summaryrefslogtreecommitdiffstats
path: root/media/libaom/src/examples
diff options
context:
space:
mode:
authorMatt A. Tobin <email@mattatobin.com>2020-04-07 23:30:51 -0400
committerwolfbeast <mcwerewolf@wolfbeast.com>2020-04-14 13:26:42 +0200
commit277f2116b6660e9bbe7f5d67524be57eceb49b8b (patch)
tree4595f7cc71418f71b9a97dfaeb03a30aa60f336a /media/libaom/src/examples
parentd270404436f6e84ffa3b92af537ac721bf10d66e (diff)
downloadUXP-277f2116b6660e9bbe7f5d67524be57eceb49b8b.tar
UXP-277f2116b6660e9bbe7f5d67524be57eceb49b8b.tar.gz
UXP-277f2116b6660e9bbe7f5d67524be57eceb49b8b.tar.lz
UXP-277f2116b6660e9bbe7f5d67524be57eceb49b8b.tar.xz
UXP-277f2116b6660e9bbe7f5d67524be57eceb49b8b.zip
Move aom source to a sub-directory under media/libaom
There is no damned reason to treat this differently than any other media lib given its license and there never was.
Diffstat (limited to 'media/libaom/src/examples')
-rw-r--r--media/libaom/src/examples/analyzer.cc723
-rw-r--r--media/libaom/src/examples/aom_cx_set_ref.c385
-rw-r--r--media/libaom/src/examples/decode_to_md5.c131
-rw-r--r--media/libaom/src/examples/decode_with_drops.c146
-rw-r--r--media/libaom/src/examples/encoder_util.c136
-rw-r--r--media/libaom/src/examples/encoder_util.h33
-rw-r--r--media/libaom/src/examples/inspect.c763
-rw-r--r--media/libaom/src/examples/lightfield_bitstream_parsing.c348
-rw-r--r--media/libaom/src/examples/lightfield_decoder.c208
-rw-r--r--media/libaom/src/examples/lightfield_encoder.c499
-rw-r--r--media/libaom/src/examples/lightfield_tile_list_decoder.c161
-rw-r--r--media/libaom/src/examples/lossless_encoder.c138
-rw-r--r--media/libaom/src/examples/noise_model.c431
-rw-r--r--media/libaom/src/examples/resize_util.c124
-rw-r--r--media/libaom/src/examples/scalable_decoder.c185
-rw-r--r--media/libaom/src/examples/scalable_encoder.c289
-rw-r--r--media/libaom/src/examples/set_maps.c208
-rw-r--r--media/libaom/src/examples/simple_decoder.c146
-rw-r--r--media/libaom/src/examples/simple_encoder.c249
-rw-r--r--media/libaom/src/examples/twopass_encoder.c250
20 files changed, 5553 insertions, 0 deletions
diff --git a/media/libaom/src/examples/analyzer.cc b/media/libaom/src/examples/analyzer.cc
new file mode 100644
index 000000000..6a42eca24
--- /dev/null
+++ b/media/libaom/src/examples/analyzer.cc
@@ -0,0 +1,723 @@
+/*
+ * Copyright (c) 2017, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+#include <wx/wx.h>
+#include <wx/aboutdlg.h>
+#include <wx/cmdline.h>
+#include <wx/dcbuffer.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aomdx.h"
+#include "av1/common/onyxc_int.h"
+#include "av1/decoder/accounting.h"
+#include "av1/decoder/inspection.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+
+#define OD_SIGNMASK(a) (-((a) < 0))
+#define OD_FLIPSIGNI(a, b) (((a) + OD_SIGNMASK(b)) ^ OD_SIGNMASK(b))
+#define OD_DIV_ROUND(x, y) (((x) + OD_FLIPSIGNI((y) >> 1, x)) / (y))
+
+enum {
+ OD_LUMA_MASK = 1 << 0,
+ OD_CB_MASK = 1 << 1,
+ OD_CR_MASK = 1 << 2,
+ OD_ALL_MASK = OD_LUMA_MASK | OD_CB_MASK | OD_CR_MASK
+};
+
+class AV1Decoder {
+ private:
+ FILE *input;
+ wxString path;
+
+ AvxVideoReader *reader;
+ const AvxVideoInfo *info;
+ const AvxInterface *decoder;
+
+ insp_frame_data frame_data;
+
+ aom_codec_ctx_t codec;
+ bool show_padding;
+
+ public:
+ aom_image_t *image;
+ int frame;
+
+ int plane_mask;
+
+ AV1Decoder();
+ ~AV1Decoder();
+
+ bool open(const wxString &path);
+ void close();
+ bool step();
+
+ int getWidthPadding() const;
+ int getHeightPadding() const;
+ void togglePadding();
+ int getWidth() const;
+ int getHeight() const;
+
+ bool getAccountingStruct(Accounting **acct);
+ bool setInspectionCallback();
+
+ static void inspect(void *decoder, void *data);
+};
+
+AV1Decoder::AV1Decoder()
+ : reader(NULL), info(NULL), decoder(NULL), show_padding(false), image(NULL),
+ frame(0) {}
+
+AV1Decoder::~AV1Decoder() {}
+
+void AV1Decoder::togglePadding() { show_padding = !show_padding; }
+
+bool AV1Decoder::open(const wxString &path) {
+ reader = aom_video_reader_open(path.mb_str());
+ if (!reader) {
+ fprintf(stderr, "Failed to open %s for reading.", path.mb_str().data());
+ return false;
+ }
+ this->path = path;
+ info = aom_video_reader_get_info(reader);
+ decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) {
+ fprintf(stderr, "Unknown input codec.");
+ return false;
+ }
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0)) {
+ fprintf(stderr, "Failed to initialize decoder.");
+ return false;
+ }
+ ifd_init(&frame_data, info->frame_width, info->frame_height);
+ setInspectionCallback();
+ return true;
+}
+
+void AV1Decoder::close() {}
+
+bool AV1Decoder::step() {
+ if (aom_video_reader_read_frame(reader)) {
+ size_t frame_size;
+ const unsigned char *frame_data;
+ frame_data = aom_video_reader_get_frame(reader, &frame_size);
+ if (aom_codec_decode(&codec, frame_data, frame_size, NULL)) {
+ fprintf(stderr, "Failed to decode frame.");
+ return false;
+ } else {
+ aom_codec_iter_t iter = NULL;
+ image = aom_codec_get_frame(&codec, &iter);
+ if (image != NULL) {
+ frame++;
+ return true;
+ }
+ return false;
+ }
+ }
+ return false;
+}
+
+int AV1Decoder::getWidth() const {
+ return info->frame_width + 2 * getWidthPadding();
+}
+
+int AV1Decoder::getWidthPadding() const {
+ return show_padding ? AOMMAX(info->frame_width + 16,
+ ALIGN_POWER_OF_TWO(info->frame_width, 6)) -
+ info->frame_width
+ : 0;
+}
+
+int AV1Decoder::getHeight() const {
+ return info->frame_height + 2 * getHeightPadding();
+}
+
+int AV1Decoder::getHeightPadding() const {
+ return show_padding ? AOMMAX(info->frame_height + 16,
+ ALIGN_POWER_OF_TWO(info->frame_height, 6)) -
+ info->frame_height
+ : 0;
+}
+
+bool AV1Decoder::getAccountingStruct(Accounting **accounting) {
+ return aom_codec_control(&codec, AV1_GET_ACCOUNTING, accounting) ==
+ AOM_CODEC_OK;
+}
+
+bool AV1Decoder::setInspectionCallback() {
+ aom_inspect_init ii;
+ ii.inspect_cb = AV1Decoder::inspect;
+ ii.inspect_ctx = (void *)this;
+ return aom_codec_control(&codec, AV1_SET_INSPECTION_CALLBACK, &ii) ==
+ AOM_CODEC_OK;
+}
+
+void AV1Decoder::inspect(void *pbi, void *data) {
+ AV1Decoder *decoder = (AV1Decoder *)data;
+ ifd_inspect(&decoder->frame_data, pbi);
+}
+
+#define MIN_ZOOM (1)
+#define MAX_ZOOM (4)
+
+class AnalyzerPanel : public wxPanel {
+ DECLARE_EVENT_TABLE()
+
+ private:
+ AV1Decoder decoder;
+ const wxString path;
+
+ int zoom;
+ unsigned char *pixels;
+
+ const bool bit_accounting;
+ double *bpp_q3;
+
+ int plane_mask;
+
+ // The display size is the decode size, scaled by the zoom.
+ int getDisplayWidth() const;
+ int getDisplayHeight() const;
+
+ bool updateDisplaySize();
+
+ void computeBitsPerPixel();
+
+ public:
+ AnalyzerPanel(wxWindow *parent, const wxString &path,
+ const bool bit_accounting);
+ ~AnalyzerPanel();
+
+ bool open(const wxString &path);
+ void close();
+ void render();
+ void togglePadding();
+ bool nextFrame();
+ void refresh();
+
+ int getZoom() const;
+ bool setZoom(int zoom);
+
+ void setShowPlane(bool show_plane, int mask);
+
+ void onPaint(wxPaintEvent &event); // NOLINT
+};
+
+BEGIN_EVENT_TABLE(AnalyzerPanel, wxPanel)
+EVT_PAINT(AnalyzerPanel::onPaint)
+END_EVENT_TABLE()
+
+AnalyzerPanel::AnalyzerPanel(wxWindow *parent, const wxString &path,
+ const bool bit_accounting)
+ : wxPanel(parent), path(path), zoom(0), pixels(NULL),
+ bit_accounting(bit_accounting), bpp_q3(NULL), plane_mask(OD_ALL_MASK) {}
+
+AnalyzerPanel::~AnalyzerPanel() { close(); }
+
+void AnalyzerPanel::setShowPlane(bool show_plane, int mask) {
+ if (show_plane) {
+ plane_mask |= mask;
+ } else {
+ plane_mask &= ~mask;
+ }
+}
+
+void AnalyzerPanel::render() {
+ aom_image_t *img = decoder.image;
+ const int hbd = !!(img->fmt & AOM_IMG_FMT_HIGHBITDEPTH);
+ int y_stride = img->stride[0] >> hbd;
+ int cb_stride = img->stride[1] >> hbd;
+ int cr_stride = img->stride[2] >> hbd;
+ int p_stride = 3 * getDisplayWidth();
+ unsigned char *y_row = img->planes[0];
+ unsigned char *cb_row = img->planes[1];
+ unsigned char *cr_row = img->planes[2];
+ uint16_t *y_row16 = reinterpret_cast<uint16_t *>(y_row);
+ uint16_t *cb_row16 = reinterpret_cast<uint16_t *>(cb_row);
+ uint16_t *cr_row16 = reinterpret_cast<uint16_t *>(cr_row);
+ unsigned char *p_row = pixels;
+ int y_width_padding = decoder.getWidthPadding();
+ int cb_width_padding = y_width_padding >> 1;
+ int cr_width_padding = y_width_padding >> 1;
+ int y_height_padding = decoder.getHeightPadding();
+ int cb_height_padding = y_height_padding >> 1;
+ int cr_height_padding = y_height_padding >> 1;
+ for (int j = 0; j < decoder.getHeight(); j++) {
+ unsigned char *y = y_row - y_stride * y_height_padding;
+ unsigned char *cb = cb_row - cb_stride * cb_height_padding;
+ unsigned char *cr = cr_row - cr_stride * cr_height_padding;
+ uint16_t *y16 = y_row16 - y_stride * y_height_padding;
+ uint16_t *cb16 = cb_row16 - cb_stride * cb_height_padding;
+ uint16_t *cr16 = cr_row16 - cr_stride * cr_height_padding;
+ unsigned char *p = p_row;
+ for (int i = 0; i < decoder.getWidth(); i++) {
+ int64_t yval;
+ int64_t cbval;
+ int64_t crval;
+ int pmask;
+ unsigned rval;
+ unsigned gval;
+ unsigned bval;
+ if (hbd) {
+ yval = *(y16 - y_width_padding);
+ cbval = *(cb16 - cb_width_padding);
+ crval = *(cr16 - cr_width_padding);
+ } else {
+ yval = *(y - y_width_padding);
+ cbval = *(cb - cb_width_padding);
+ crval = *(cr - cr_width_padding);
+ }
+ pmask = plane_mask;
+ if (pmask & OD_LUMA_MASK) {
+ yval -= 16;
+ } else {
+ yval = 128;
+ }
+ cbval = ((pmask & OD_CB_MASK) >> 1) * (cbval - 128);
+ crval = ((pmask & OD_CR_MASK) >> 2) * (crval - 128);
+ /*This is intentionally slow and very accurate.*/
+ rval = OD_CLAMPI(
+ 0,
+ (int32_t)OD_DIV_ROUND(
+ 2916394880000LL * yval + 4490222169144LL * crval, 9745792000LL),
+ 65535);
+ gval = OD_CLAMPI(0,
+ (int32_t)OD_DIV_ROUND(2916394880000LL * yval -
+ 534117096223LL * cbval -
+ 1334761232047LL * crval,
+ 9745792000LL),
+ 65535);
+ bval = OD_CLAMPI(
+ 0,
+ (int32_t)OD_DIV_ROUND(
+ 2916394880000LL * yval + 5290866304968LL * cbval, 9745792000LL),
+ 65535);
+ unsigned char *px_row = p;
+ for (int v = 0; v < zoom; v++) {
+ unsigned char *px = px_row;
+ for (int u = 0; u < zoom; u++) {
+ *(px + 0) = (unsigned char)(rval >> 8);
+ *(px + 1) = (unsigned char)(gval >> 8);
+ *(px + 2) = (unsigned char)(bval >> 8);
+ px += 3;
+ }
+ px_row += p_stride;
+ }
+ if (hbd) {
+ int dc = ((y16 - y_row16) & 1) | (1 - img->x_chroma_shift);
+ y16++;
+ cb16 += dc;
+ cr16 += dc;
+ } else {
+ int dc = ((y - y_row) & 1) | (1 - img->x_chroma_shift);
+ y++;
+ cb += dc;
+ cr += dc;
+ }
+ p += zoom * 3;
+ }
+ int dc = -((j & 1) | (1 - img->y_chroma_shift));
+ if (hbd) {
+ y_row16 += y_stride;
+ cb_row16 += dc & cb_stride;
+ cr_row16 += dc & cr_stride;
+ } else {
+ y_row += y_stride;
+ cb_row += dc & cb_stride;
+ cr_row += dc & cr_stride;
+ }
+ p_row += zoom * p_stride;
+ }
+}
+
+void AnalyzerPanel::computeBitsPerPixel() {
+ Accounting *acct;
+ double bpp_total;
+ int totals_q3[MAX_SYMBOL_TYPES] = { 0 };
+ int sym_count[MAX_SYMBOL_TYPES] = { 0 };
+ decoder.getAccountingStruct(&acct);
+ for (int j = 0; j < decoder.getHeight(); j++) {
+ for (int i = 0; i < decoder.getWidth(); i++) {
+ bpp_q3[j * decoder.getWidth() + i] = 0.0;
+ }
+ }
+ bpp_total = 0;
+ for (int i = 0; i < acct->syms.num_syms; i++) {
+ AccountingSymbol *s;
+ s = &acct->syms.syms[i];
+ totals_q3[s->id] += s->bits;
+ sym_count[s->id] += s->samples;
+ }
+ printf("=== Frame: %-3i ===\n", decoder.frame - 1);
+ for (int i = 0; i < acct->syms.dictionary.num_strs; i++) {
+ if (totals_q3[i]) {
+ printf("%30s = %10.3f (%f bit/symbol)\n", acct->syms.dictionary.strs[i],
+ (float)totals_q3[i] / 8, (float)totals_q3[i] / 8 / sym_count[i]);
+ }
+ }
+ printf("\n");
+}
+
+void AnalyzerPanel::togglePadding() {
+ decoder.togglePadding();
+ updateDisplaySize();
+}
+
+bool AnalyzerPanel::nextFrame() {
+ if (decoder.step()) {
+ refresh();
+ return true;
+ }
+ return false;
+}
+
+void AnalyzerPanel::refresh() {
+ if (bit_accounting) {
+ computeBitsPerPixel();
+ }
+ render();
+}
+
+int AnalyzerPanel::getDisplayWidth() const { return zoom * decoder.getWidth(); }
+
+int AnalyzerPanel::getDisplayHeight() const {
+ return zoom * decoder.getHeight();
+}
+
+bool AnalyzerPanel::updateDisplaySize() {
+ unsigned char *p = (unsigned char *)malloc(
+ sizeof(*p) * 3 * getDisplayWidth() * getDisplayHeight());
+ if (p == NULL) {
+ return false;
+ }
+ free(pixels);
+ pixels = p;
+ SetSize(getDisplayWidth(), getDisplayHeight());
+ return true;
+}
+
+bool AnalyzerPanel::open(const wxString &path) {
+ if (!decoder.open(path)) {
+ return false;
+ }
+ if (!setZoom(MIN_ZOOM)) {
+ return false;
+ }
+ if (bit_accounting) {
+ bpp_q3 = (double *)malloc(sizeof(*bpp_q3) * decoder.getWidth() *
+ decoder.getHeight());
+ if (bpp_q3 == NULL) {
+ fprintf(stderr, "Could not allocate memory for bit accounting\n");
+ close();
+ return false;
+ }
+ }
+ if (!nextFrame()) {
+ close();
+ return false;
+ }
+ SetFocus();
+ return true;
+}
+
+void AnalyzerPanel::close() {
+ decoder.close();
+ free(pixels);
+ pixels = NULL;
+ free(bpp_q3);
+ bpp_q3 = NULL;
+}
+
+int AnalyzerPanel::getZoom() const { return zoom; }
+
+bool AnalyzerPanel::setZoom(int z) {
+ if (z <= MAX_ZOOM && z >= MIN_ZOOM && zoom != z) {
+ int old_zoom = zoom;
+ zoom = z;
+ if (!updateDisplaySize()) {
+ zoom = old_zoom;
+ return false;
+ }
+ return true;
+ }
+ return false;
+}
+
+void AnalyzerPanel::onPaint(wxPaintEvent &) {
+ wxBitmap bmp(wxImage(getDisplayWidth(), getDisplayHeight(), pixels, true));
+ wxBufferedPaintDC dc(this, bmp);
+}
+
+class AnalyzerFrame : public wxFrame {
+ DECLARE_EVENT_TABLE()
+
+ private:
+ AnalyzerPanel *panel;
+ const bool bit_accounting;
+
+ wxMenu *fileMenu;
+ wxMenu *viewMenu;
+ wxMenu *playbackMenu;
+
+ public:
+ AnalyzerFrame(const bool bit_accounting); // NOLINT
+
+ void onOpen(wxCommandEvent &event); // NOLINT
+ void onClose(wxCommandEvent &event); // NOLINT
+ void onQuit(wxCommandEvent &event); // NOLINT
+
+ void onTogglePadding(wxCommandEvent &event); // NOLINT
+ void onZoomIn(wxCommandEvent &event); // NOLINT
+ void onZoomOut(wxCommandEvent &event); // NOLINT
+ void onActualSize(wxCommandEvent &event); // NOLINT
+
+ void onToggleViewMenuCheckBox(wxCommandEvent &event); // NOLINT
+ void onResetAndToggleViewMenuCheckBox(wxCommandEvent &event); // NOLINT
+
+ void onNextFrame(wxCommandEvent &event); // NOLINT
+ void onGotoFrame(wxCommandEvent &event); // NOLINT
+ void onRestart(wxCommandEvent &event); // NOLINT
+
+ void onAbout(wxCommandEvent &event); // NOLINT
+
+ bool open(const wxString &path);
+ bool setZoom(int zoom);
+ void updateViewMenu();
+};
+
+enum {
+ wxID_NEXT_FRAME = 6000,
+ wxID_SHOW_Y,
+ wxID_SHOW_U,
+ wxID_SHOW_V,
+ wxID_GOTO_FRAME,
+ wxID_RESTART,
+ wxID_ACTUAL_SIZE,
+ wxID_PADDING
+};
+
+BEGIN_EVENT_TABLE(AnalyzerFrame, wxFrame)
+EVT_MENU(wxID_OPEN, AnalyzerFrame::onOpen)
+EVT_MENU(wxID_CLOSE, AnalyzerFrame::onClose)
+EVT_MENU(wxID_EXIT, AnalyzerFrame::onQuit)
+EVT_MENU(wxID_PADDING, AnalyzerFrame::onTogglePadding)
+EVT_MENU(wxID_ZOOM_IN, AnalyzerFrame::onZoomIn)
+EVT_MENU(wxID_ZOOM_OUT, AnalyzerFrame::onZoomOut)
+EVT_MENU(wxID_ACTUAL_SIZE, AnalyzerFrame::onActualSize)
+EVT_MENU(wxID_SHOW_Y, AnalyzerFrame::onResetAndToggleViewMenuCheckBox)
+EVT_MENU(wxID_SHOW_U, AnalyzerFrame::onResetAndToggleViewMenuCheckBox)
+EVT_MENU(wxID_SHOW_V, AnalyzerFrame::onResetAndToggleViewMenuCheckBox)
+EVT_MENU(wxID_NEXT_FRAME, AnalyzerFrame::onNextFrame)
+EVT_MENU(wxID_GOTO_FRAME, AnalyzerFrame::onGotoFrame)
+EVT_MENU(wxID_RESTART, AnalyzerFrame::onRestart)
+EVT_MENU(wxID_ABOUT, AnalyzerFrame::onAbout)
+END_EVENT_TABLE()
+
+AnalyzerFrame::AnalyzerFrame(const bool bit_accounting)
+ : wxFrame(NULL, wxID_ANY, _("AV1 Stream Analyzer"), wxDefaultPosition,
+ wxDefaultSize, wxDEFAULT_FRAME_STYLE),
+ panel(NULL), bit_accounting(bit_accounting) {
+ wxMenuBar *mb = new wxMenuBar();
+
+ fileMenu = new wxMenu();
+ fileMenu->Append(wxID_OPEN, _("&Open...\tCtrl-O"), _("Open daala file"));
+ fileMenu->Append(wxID_CLOSE, _("&Close\tCtrl-W"), _("Close daala file"));
+ fileMenu->Enable(wxID_CLOSE, false);
+ fileMenu->Append(wxID_EXIT, _("E&xit\tCtrl-Q"), _("Quit this program"));
+ mb->Append(fileMenu, _("&File"));
+
+ wxAcceleratorEntry entries[2];
+ entries[0].Set(wxACCEL_CTRL, (int)'=', wxID_ZOOM_IN);
+ entries[1].Set(wxACCEL_CTRL | wxACCEL_SHIFT, (int)'-', wxID_ZOOM_OUT);
+ wxAcceleratorTable accel(2, entries);
+ this->SetAcceleratorTable(accel);
+
+ viewMenu = new wxMenu();
+ +viewMenu->Append(wxID_PADDING, _("Toggle padding\tCtrl-p"),
+ _("Show padding"));
+ viewMenu->Append(wxID_ZOOM_IN, _("Zoom-In\tCtrl-+"), _("Double image size"));
+ viewMenu->Append(wxID_ZOOM_OUT, _("Zoom-Out\tCtrl--"), _("Half image size"));
+ viewMenu->Append(wxID_ACTUAL_SIZE, _("Actual size\tCtrl-0"),
+ _("Actual size of the frame"));
+ viewMenu->AppendSeparator();
+ viewMenu->AppendCheckItem(wxID_SHOW_Y, _("&Y plane\tCtrl-Y"),
+ _("Show Y plane"));
+ viewMenu->AppendCheckItem(wxID_SHOW_U, _("&U plane\tCtrl-U"),
+ _("Show U plane"));
+ viewMenu->AppendCheckItem(wxID_SHOW_V, _("&V plane\tCtrl-V"),
+ _("Show V plane"));
+ mb->Append(viewMenu, _("&View"));
+
+ playbackMenu = new wxMenu();
+ playbackMenu->Append(wxID_NEXT_FRAME, _("Next frame\tCtrl-."),
+ _("Go to next frame"));
+ /*playbackMenu->Append(wxID_RESTART, _("&Restart\tCtrl-R"),
+ _("Set video to frame 0"));
+ playbackMenu->Append(wxID_GOTO_FRAME, _("Jump to Frame\tCtrl-J"),
+ _("Go to frame number"));*/
+ mb->Append(playbackMenu, _("&Playback"));
+
+ wxMenu *helpMenu = new wxMenu();
+ helpMenu->Append(wxID_ABOUT, _("&About...\tF1"), _("Show about dialog"));
+ mb->Append(helpMenu, _("&Help"));
+
+ SetMenuBar(mb);
+
+ CreateStatusBar(1);
+}
+
+void AnalyzerFrame::onOpen(wxCommandEvent &WXUNUSED(event)) {
+ wxFileDialog openFileDialog(this, _("Open file"), wxEmptyString,
+ wxEmptyString, _("AV1 files (*.ivf)|*.ivf"),
+ wxFD_OPEN | wxFD_FILE_MUST_EXIST);
+ if (openFileDialog.ShowModal() != wxID_CANCEL) {
+ open(openFileDialog.GetPath());
+ }
+}
+
+void AnalyzerFrame::onClose(wxCommandEvent &WXUNUSED(event)) {}
+
+void AnalyzerFrame::onQuit(wxCommandEvent &WXUNUSED(event)) { Close(true); }
+
+void AnalyzerFrame::onTogglePadding(wxCommandEvent &WXUNUSED(event)) {
+ panel->togglePadding();
+ SetClientSize(panel->GetSize());
+ panel->render();
+ panel->Refresh();
+}
+
+void AnalyzerFrame::onZoomIn(wxCommandEvent &WXUNUSED(event)) {
+ setZoom(panel->getZoom() + 1);
+}
+
+void AnalyzerFrame::onZoomOut(wxCommandEvent &WXUNUSED(event)) {
+ setZoom(panel->getZoom() - 1);
+}
+
+void AnalyzerFrame::onActualSize(wxCommandEvent &WXUNUSED(event)) {
+ setZoom(MIN_ZOOM);
+}
+
+void AnalyzerFrame::onToggleViewMenuCheckBox(wxCommandEvent &event) { // NOLINT
+ GetMenuBar()->Check(event.GetId(), event.IsChecked());
+ updateViewMenu();
+}
+
+void AnalyzerFrame::onResetAndToggleViewMenuCheckBox(
+ wxCommandEvent &event) { // NOLINT
+ int id = event.GetId();
+ if (id != wxID_SHOW_Y && id != wxID_SHOW_U && id != wxID_SHOW_V) {
+ GetMenuBar()->Check(wxID_SHOW_Y, true);
+ GetMenuBar()->Check(wxID_SHOW_U, true);
+ GetMenuBar()->Check(wxID_SHOW_V, true);
+ }
+ onToggleViewMenuCheckBox(event);
+}
+
+void AnalyzerFrame::onNextFrame(wxCommandEvent &WXUNUSED(event)) {
+ panel->nextFrame();
+ panel->Refresh(false);
+}
+
+void AnalyzerFrame::onGotoFrame(wxCommandEvent &WXUNUSED(event)) {}
+
+void AnalyzerFrame::onRestart(wxCommandEvent &WXUNUSED(event)) {}
+
+void AnalyzerFrame::onAbout(wxCommandEvent &WXUNUSED(event)) {
+ wxAboutDialogInfo info;
+ info.SetName(_("AV1 Bitstream Analyzer"));
+ info.SetVersion(_("0.1-beta"));
+ info.SetDescription(
+ _("This program implements a bitstream analyzer for AV1"));
+ info.SetCopyright(
+ wxT("(C) 2017 Alliance for Open Media <negge@mozilla.com>"));
+ wxAboutBox(info);
+}
+
+bool AnalyzerFrame::open(const wxString &path) {
+ panel = new AnalyzerPanel(this, path, bit_accounting);
+ if (panel->open(path)) {
+ SetClientSize(panel->GetSize());
+ return true;
+ } else {
+ delete panel;
+ return false;
+ }
+}
+
+bool AnalyzerFrame::setZoom(int zoom) {
+ if (panel->setZoom(zoom)) {
+ GetMenuBar()->Enable(wxID_ACTUAL_SIZE, zoom != MIN_ZOOM);
+ GetMenuBar()->Enable(wxID_ZOOM_IN, zoom != MAX_ZOOM);
+ GetMenuBar()->Enable(wxID_ZOOM_OUT, zoom != MIN_ZOOM);
+ SetClientSize(panel->GetSize());
+ panel->render();
+ panel->Refresh();
+ return true;
+ }
+ return false;
+}
+
+void AnalyzerFrame::updateViewMenu() {
+ panel->setShowPlane(GetMenuBar()->IsChecked(wxID_SHOW_Y), OD_LUMA_MASK);
+ panel->setShowPlane(GetMenuBar()->IsChecked(wxID_SHOW_U), OD_CB_MASK);
+ panel->setShowPlane(GetMenuBar()->IsChecked(wxID_SHOW_V), OD_CR_MASK);
+ SetClientSize(panel->GetSize());
+ panel->render();
+ panel->Refresh(false);
+}
+
+class Analyzer : public wxApp {
+ private:
+ AnalyzerFrame *frame;
+
+ public:
+ void OnInitCmdLine(wxCmdLineParser &parser); // NOLINT
+ bool OnCmdLineParsed(wxCmdLineParser &parser); // NOLINT
+};
+
+static const wxCmdLineEntryDesc CMD_LINE_DESC[] = {
+ { wxCMD_LINE_SWITCH, _("h"), _("help"), _("Display this help and exit."),
+ wxCMD_LINE_VAL_NONE, wxCMD_LINE_OPTION_HELP },
+ { wxCMD_LINE_SWITCH, _("a"), _("bit-accounting"), _("Enable bit accounting"),
+ wxCMD_LINE_VAL_NONE, wxCMD_LINE_PARAM_OPTIONAL },
+ { wxCMD_LINE_PARAM, NULL, NULL, _("input.ivf"), wxCMD_LINE_VAL_STRING,
+ wxCMD_LINE_PARAM_OPTIONAL },
+ { wxCMD_LINE_NONE }
+};
+
+void Analyzer::OnInitCmdLine(wxCmdLineParser &parser) { // NOLINT
+ parser.SetDesc(CMD_LINE_DESC);
+ parser.SetSwitchChars(_("-"));
+}
+
+bool Analyzer::OnCmdLineParsed(wxCmdLineParser &parser) { // NOLINT
+ bool bit_accounting = parser.Found(_("a"));
+ if (bit_accounting && !CONFIG_ACCOUNTING) {
+ fprintf(stderr,
+ "Bit accounting support not found. "
+ "Recompile with:\n./cmake -DCONFIG_ACCOUNTING=1\n");
+ return false;
+ }
+ frame = new AnalyzerFrame(parser.Found(_("a")));
+ frame->Show();
+ if (parser.GetParamCount() > 0) {
+ return frame->open(parser.GetParam(0));
+ }
+ return true;
+}
+
+void usage_exit(void) {
+ fprintf(stderr, "uhh\n");
+ exit(EXIT_FAILURE);
+}
+
+IMPLEMENT_APP(Analyzer)
diff --git a/media/libaom/src/examples/aom_cx_set_ref.c b/media/libaom/src/examples/aom_cx_set_ref.c
new file mode 100644
index 000000000..8e3d216fe
--- /dev/null
+++ b/media/libaom/src/examples/aom_cx_set_ref.c
@@ -0,0 +1,385 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// AV1 Set Reference Frame
+// ============================
+//
+// This is an example demonstrating how to overwrite the AV1 encoder's
+// internal reference frame. In the sample we set the last frame to the
+// current frame. This technique could be used to bounce between two cameras.
+//
+// The decoder would also have to set the reference frame to the same value
+// on the same frame, or the video will become corrupt. The 'test_decode'
+// variable is set to 1 in this example that tests if the encoder and decoder
+// results are matching.
+//
+// Usage
+// -----
+// This example encodes a raw video. And the last argument passed in specifies
+// the frame number to update the reference frame on. For example, run
+// examples/aom_cx_set_ref av1 352 288 in.yuv out.ivf 4 30
+// The parameter is parsed as follows:
+//
+//
+// Extra Variables
+// ---------------
+// This example maintains the frame number passed on the command line
+// in the `update_frame_num` variable.
+//
+//
+// Configuration
+// -------------
+//
+// The reference frame is updated on the frame specified on the command
+// line.
+//
+// Observing The Effects
+// ---------------------
+// The encoder and decoder results should be matching when the same reference
+// frame setting operation is done in both encoder and decoder. Otherwise,
+// the encoder/decoder mismatch would be seen.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aom_encoder.h"
+#include "aom/aomcx.h"
+#include "aom_scale/yv12config.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+#include "examples/encoder_util.h"
+
+static const char *exec_name;
+
+void usage_exit() {
+ fprintf(stderr,
+ "Usage: %s <codec> <width> <height> <infile> <outfile> "
+ "<frame> <limit(optional)>\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static void testing_decode(aom_codec_ctx_t *encoder, aom_codec_ctx_t *decoder,
+ unsigned int frame_out, int *mismatch_seen) {
+ aom_image_t enc_img, dec_img;
+
+ if (*mismatch_seen) return;
+
+ /* Get the internal reference frame */
+ if (aom_codec_control(encoder, AV1_GET_NEW_FRAME_IMAGE, &enc_img))
+ die_codec(encoder, "Failed to get encoder reference frame");
+ if (aom_codec_control(decoder, AV1_GET_NEW_FRAME_IMAGE, &dec_img))
+ die_codec(decoder, "Failed to get decoder reference frame");
+
+ if ((enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) !=
+ (dec_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH)) {
+ if (enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
+ aom_image_t enc_hbd_img;
+ aom_img_alloc(&enc_hbd_img, enc_img.fmt - AOM_IMG_FMT_HIGHBITDEPTH,
+ enc_img.d_w, enc_img.d_h, 16);
+ aom_img_truncate_16_to_8(&enc_hbd_img, &enc_img);
+ enc_img = enc_hbd_img;
+ }
+ if (dec_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
+ aom_image_t dec_hbd_img;
+ aom_img_alloc(&dec_hbd_img, dec_img.fmt - AOM_IMG_FMT_HIGHBITDEPTH,
+ dec_img.d_w, dec_img.d_h, 16);
+ aom_img_truncate_16_to_8(&dec_hbd_img, &dec_img);
+ dec_img = dec_hbd_img;
+ }
+ }
+
+ if (!aom_compare_img(&enc_img, &dec_img)) {
+ int y[4], u[4], v[4];
+ if (enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
+ aom_find_mismatch_high(&enc_img, &dec_img, y, u, v);
+ } else {
+ aom_find_mismatch(&enc_img, &dec_img, y, u, v);
+ }
+
+ printf(
+ "Encode/decode mismatch on frame %d at"
+ " Y[%d, %d] {%d/%d},"
+ " U[%d, %d] {%d/%d},"
+ " V[%d, %d] {%d/%d}",
+ frame_out, y[0], y[1], y[2], y[3], u[0], u[1], u[2], u[3], v[0], v[1],
+ v[2], v[3]);
+ *mismatch_seen = 1;
+ }
+
+ aom_img_free(&enc_img);
+ aom_img_free(&dec_img);
+}
+
+static int encode_frame(aom_codec_ctx_t *ecodec, aom_image_t *img,
+ unsigned int frame_in, AvxVideoWriter *writer,
+ int test_decode, aom_codec_ctx_t *dcodec,
+ unsigned int *frame_out, int *mismatch_seen,
+ aom_image_t *ext_ref) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ int got_data;
+ const aom_codec_err_t res = aom_codec_encode(ecodec, img, frame_in, 1, 0);
+ if (res != AOM_CODEC_OK) die_codec(ecodec, "Failed to encode frame");
+
+ got_data = 0;
+
+ while ((pkt = aom_codec_get_cx_data(ecodec, &iter)) != NULL) {
+ got_pkts = 1;
+
+ if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
+ const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
+
+ if (!(pkt->data.frame.flags & AOM_FRAME_IS_FRAGMENT)) {
+ *frame_out += 1;
+ }
+
+ if (!aom_video_writer_write_frame(writer, pkt->data.frame.buf,
+ pkt->data.frame.sz,
+ pkt->data.frame.pts)) {
+ die_codec(ecodec, "Failed to write compressed frame");
+ }
+ printf(keyframe ? "K" : ".");
+ fflush(stdout);
+ got_data = 1;
+
+ // Decode 1 frame.
+ if (test_decode) {
+ if (aom_codec_decode(dcodec, pkt->data.frame.buf,
+ (unsigned int)pkt->data.frame.sz, NULL))
+ die_codec(dcodec, "Failed to decode frame.");
+
+ // Copy out first decoded frame, and use it as reference later.
+ if (*frame_out == 1 && ext_ref != NULL)
+ if (aom_codec_control(dcodec, AV1_COPY_NEW_FRAME_IMAGE, ext_ref))
+ die_codec(dcodec, "Failed to get decoder new frame");
+ }
+ }
+ }
+
+ // Mismatch checking
+ if (got_data && test_decode) {
+ testing_decode(ecodec, dcodec, *frame_out, mismatch_seen);
+ }
+
+ return got_pkts;
+}
+
+int main(int argc, char **argv) {
+ FILE *infile = NULL;
+ // Encoder
+ aom_codec_ctx_t ecodec;
+ aom_codec_enc_cfg_t cfg;
+ unsigned int frame_in = 0;
+ aom_image_t raw;
+ aom_image_t raw_shift;
+ aom_image_t ext_ref;
+ aom_codec_err_t res;
+ AvxVideoInfo info;
+ AvxVideoWriter *writer = NULL;
+ const AvxInterface *encoder = NULL;
+ int flags = 0;
+ int allocated_raw_shift = 0;
+ aom_img_fmt_t raw_fmt = AOM_IMG_FMT_I420;
+ aom_img_fmt_t ref_fmt = AOM_IMG_FMT_I420;
+
+ // Test encoder/decoder mismatch.
+ int test_decode = 1;
+ // Decoder
+ aom_codec_ctx_t dcodec;
+ unsigned int frame_out = 0;
+
+ // The frame number to set reference frame on
+ unsigned int update_frame_num = 0;
+ int mismatch_seen = 0;
+
+ const int fps = 30;
+ const int bitrate = 500;
+
+ const char *codec_arg = NULL;
+ const char *width_arg = NULL;
+ const char *height_arg = NULL;
+ const char *infile_arg = NULL;
+ const char *outfile_arg = NULL;
+ const char *update_frame_num_arg = NULL;
+ unsigned int limit = 0;
+ exec_name = argv[0];
+
+ // Clear explicitly, as simply assigning "{ 0 }" generates
+ // "missing-field-initializers" warning in some compilers.
+ memset(&ecodec, 0, sizeof(ecodec));
+ memset(&cfg, 0, sizeof(cfg));
+ memset(&info, 0, sizeof(info));
+
+ if (argc < 7) die("Invalid number of arguments");
+
+ codec_arg = argv[1];
+ width_arg = argv[2];
+ height_arg = argv[3];
+ infile_arg = argv[4];
+ outfile_arg = argv[5];
+ update_frame_num_arg = argv[6];
+
+ encoder = get_aom_encoder_by_name(codec_arg);
+ if (!encoder) die("Unsupported codec.");
+
+ update_frame_num = (unsigned int)strtoul(update_frame_num_arg, NULL, 0);
+ // In AV1, the reference buffers (cm->buffer_pool->frame_bufs[i].buf) are
+ // allocated while calling aom_codec_encode(), thus, setting reference for
+ // 1st frame isn't supported.
+ if (update_frame_num <= 1) {
+ die("Couldn't parse frame number '%s'\n", update_frame_num_arg);
+ }
+
+ if (argc > 7) {
+ limit = (unsigned int)strtoul(argv[7], NULL, 0);
+ if (update_frame_num > limit)
+ die("Update frame number couldn't larger than limit\n");
+ }
+
+ info.codec_fourcc = encoder->fourcc;
+ info.frame_width = (int)strtol(width_arg, NULL, 0);
+ info.frame_height = (int)strtol(height_arg, NULL, 0);
+ info.time_base.numerator = 1;
+ info.time_base.denominator = fps;
+
+ if (info.frame_width <= 0 || info.frame_height <= 0) {
+ die("Invalid frame size: %dx%d", info.frame_width, info.frame_height);
+ }
+
+ // In this test, the bit depth of input video is 8-bit, and the input format
+ // is AOM_IMG_FMT_I420.
+ if (!aom_img_alloc(&raw, raw_fmt, info.frame_width, info.frame_height, 32)) {
+ die("Failed to allocate image.");
+ }
+
+ if (!CONFIG_LOWBITDEPTH) ref_fmt |= AOM_IMG_FMT_HIGHBITDEPTH;
+ // Allocate memory with the border so that it can be used as a reference.
+ if (!aom_img_alloc_with_border(&ext_ref, ref_fmt, info.frame_width,
+ info.frame_height, 32, 8,
+ AOM_BORDER_IN_PIXELS)) {
+ die("Failed to allocate image.");
+ }
+
+ printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
+
+ res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
+ if (res) die_codec(&ecodec, "Failed to get default codec config.");
+
+ cfg.g_w = info.frame_width;
+ cfg.g_h = info.frame_height;
+ cfg.g_timebase.num = info.time_base.numerator;
+ cfg.g_timebase.den = info.time_base.denominator;
+ cfg.rc_target_bitrate = bitrate;
+ cfg.g_lag_in_frames = 3;
+ cfg.g_bit_depth = AOM_BITS_8;
+
+ flags |= (cfg.g_bit_depth > AOM_BITS_8 || !CONFIG_LOWBITDEPTH)
+ ? AOM_CODEC_USE_HIGHBITDEPTH
+ : 0;
+
+ writer = aom_video_writer_open(outfile_arg, kContainerIVF, &info);
+ if (!writer) die("Failed to open %s for writing.", outfile_arg);
+
+ if (!(infile = fopen(infile_arg, "rb")))
+ die("Failed to open %s for reading.", infile_arg);
+
+ if (aom_codec_enc_init(&ecodec, encoder->codec_interface(), &cfg, flags))
+ die_codec(&ecodec, "Failed to initialize encoder");
+
+ // Disable alt_ref.
+ if (aom_codec_control(&ecodec, AOME_SET_ENABLEAUTOALTREF, 0))
+ die_codec(&ecodec, "Failed to set enable auto alt ref");
+
+ if (test_decode) {
+ const AvxInterface *decoder = get_aom_decoder_by_name(codec_arg);
+ if (aom_codec_dec_init(&dcodec, decoder->codec_interface(), NULL, 0))
+ die_codec(&dcodec, "Failed to initialize decoder.");
+ }
+
+ // Encode frames.
+ while (aom_img_read(&raw, infile)) {
+ if (limit && frame_in >= limit) break;
+ aom_image_t *frame_to_encode;
+
+ if (!CONFIG_LOWBITDEPTH) {
+ // Need to allocate larger buffer to use hbd internal.
+ int input_shift = 0;
+ if (!allocated_raw_shift) {
+ aom_img_alloc(&raw_shift, raw_fmt | AOM_IMG_FMT_HIGHBITDEPTH,
+ info.frame_width, info.frame_height, 32);
+ allocated_raw_shift = 1;
+ }
+ aom_img_upshift(&raw_shift, &raw, input_shift);
+ frame_to_encode = &raw_shift;
+ } else {
+ frame_to_encode = &raw;
+ }
+
+ if (update_frame_num > 1 && frame_out + 1 == update_frame_num) {
+ av1_ref_frame_t ref;
+ ref.idx = 0;
+ ref.use_external_ref = 0;
+ ref.img = ext_ref;
+ // Set reference frame in encoder.
+ if (aom_codec_control(&ecodec, AV1_SET_REFERENCE, &ref))
+ die_codec(&ecodec, "Failed to set encoder reference frame");
+ printf(" <SET_REF>");
+
+ // If set_reference in decoder is commented out, the enc/dec mismatch
+ // would be seen.
+ if (test_decode) {
+ ref.use_external_ref = 1;
+ if (aom_codec_control(&dcodec, AV1_SET_REFERENCE, &ref))
+ die_codec(&dcodec, "Failed to set decoder reference frame");
+ }
+ }
+
+ encode_frame(&ecodec, frame_to_encode, frame_in, writer, test_decode,
+ &dcodec, &frame_out, &mismatch_seen, &ext_ref);
+ frame_in++;
+ if (mismatch_seen) break;
+ }
+
+ // Flush encoder.
+ if (!mismatch_seen)
+ while (encode_frame(&ecodec, NULL, frame_in, writer, test_decode, &dcodec,
+ &frame_out, &mismatch_seen, NULL)) {
+ }
+
+ printf("\n");
+ fclose(infile);
+ printf("Processed %d frames.\n", frame_out);
+
+ if (test_decode) {
+ if (!mismatch_seen)
+ printf("Encoder/decoder results are matching.\n");
+ else
+ printf("Encoder/decoder results are NOT matching.\n");
+ }
+
+ if (test_decode)
+ if (aom_codec_destroy(&dcodec))
+ die_codec(&dcodec, "Failed to destroy decoder");
+
+ if (allocated_raw_shift) aom_img_free(&raw_shift);
+ aom_img_free(&ext_ref);
+ aom_img_free(&raw);
+ if (aom_codec_destroy(&ecodec))
+ die_codec(&ecodec, "Failed to destroy encoder.");
+
+ aom_video_writer_close(writer);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/decode_to_md5.c b/media/libaom/src/examples/decode_to_md5.c
new file mode 100644
index 000000000..bc127b78d
--- /dev/null
+++ b/media/libaom/src/examples/decode_to_md5.c
@@ -0,0 +1,131 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Frame-by-frame MD5 Checksum
+// ===========================
+//
+// This example builds upon the simple decoder loop to show how checksums
+// of the decoded output can be generated. These are used for validating
+// decoder implementations against the reference implementation, for example.
+//
+// MD5 algorithm
+// -------------
+// The Message-Digest 5 (MD5) is a well known hash function. We have provided
+// an implementation derived from the RSA Data Security, Inc. MD5 Message-Digest
+// Algorithm for your use. Our implmentation only changes the interface of this
+// reference code. You must include the `md5_utils.h` header for access to these
+// functions.
+//
+// Processing The Decoded Data
+// ---------------------------
+// Each row of the image is passed to the MD5 accumulator. First the Y plane
+// is processed, then U, then V. It is important to honor the image's `stride`
+// values.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aomdx.h"
+#include "common/md5_utils.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+
+static void get_image_md5(const aom_image_t *img, unsigned char digest[16]) {
+ int plane, y;
+ MD5Context md5;
+
+ MD5Init(&md5);
+
+ for (plane = 0; plane < 3; ++plane) {
+ const unsigned char *buf = img->planes[plane];
+ const int stride = img->stride[plane];
+ const int w = plane ? (img->d_w + 1) >> 1 : img->d_w;
+ const int h = plane ? (img->d_h + 1) >> 1 : img->d_h;
+
+ for (y = 0; y < h; ++y) {
+ MD5Update(&md5, buf, w);
+ buf += stride;
+ }
+ }
+
+ MD5Final(digest, &md5);
+}
+
+static void print_md5(FILE *stream, unsigned char digest[16]) {
+ int i;
+
+ for (i = 0; i < 16; ++i) fprintf(stream, "%02x", digest[i]);
+}
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s <infile> <outfile>\n", exec_name);
+ exit(EXIT_FAILURE);
+}
+
+int main(int argc, char **argv) {
+ int frame_cnt = 0;
+ FILE *outfile = NULL;
+ aom_codec_ctx_t codec;
+ AvxVideoReader *reader = NULL;
+ const AvxVideoInfo *info = NULL;
+ const AvxInterface *decoder = NULL;
+
+ exec_name = argv[0];
+
+ if (argc != 3) die("Invalid number of arguments.");
+
+ reader = aom_video_reader_open(argv[1]);
+ if (!reader) die("Failed to open %s for reading.", argv[1]);
+
+ if (!(outfile = fopen(argv[2], "wb")))
+ die("Failed to open %s for writing.", argv[2]);
+
+ info = aom_video_reader_get_info(reader);
+
+ decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) die("Unknown input codec.");
+
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder");
+
+ while (aom_video_reader_read_frame(reader)) {
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img = NULL;
+ size_t frame_size = 0;
+ const unsigned char *frame =
+ aom_video_reader_get_frame(reader, &frame_size);
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode frame");
+
+ while ((img = aom_codec_get_frame(&codec, &iter)) != NULL) {
+ unsigned char digest[16];
+
+ get_image_md5(img, digest);
+ print_md5(outfile, digest);
+ fprintf(outfile, " img-%dx%d-%04d.i420\n", img->d_w, img->d_h,
+ ++frame_cnt);
+ }
+ }
+
+ printf("Processed %d frames.\n", frame_cnt);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ aom_video_reader_close(reader);
+
+ fclose(outfile);
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/decode_with_drops.c b/media/libaom/src/examples/decode_with_drops.c
new file mode 100644
index 000000000..214401958
--- /dev/null
+++ b/media/libaom/src/examples/decode_with_drops.c
@@ -0,0 +1,146 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Decode With Drops Example
+// =========================
+//
+// This is an example utility which drops a series of frames, as specified
+// on the command line. This is useful for observing the error recovery
+// features of the codec.
+//
+// Usage
+// -----
+// This example adds a single argument to the `simple_decoder` example,
+// which specifies the range or pattern of frames to drop. The parameter is
+// parsed as follows:
+//
+// Dropping A Range Of Frames
+// --------------------------
+// To drop a range of frames, specify the starting frame and the ending
+// frame to drop, separated by a dash. The following command will drop
+// frames 5 through 10 (base 1).
+//
+// $ ./decode_with_drops in.ivf out.i420 5-10
+//
+//
+// Dropping A Pattern Of Frames
+// ----------------------------
+// To drop a pattern of frames, specify the number of frames to drop and
+// the number of frames after which to repeat the pattern, separated by
+// a forward-slash. The following command will drop 3 of 7 frames.
+// Specifically, it will decode 4 frames, then drop 3 frames, and then
+// repeat.
+//
+// $ ./decode_with_drops in.ivf out.i420 3/7
+//
+//
+// Extra Variables
+// ---------------
+// This example maintains the pattern passed on the command line in the
+// `n`, `m`, and `is_range` variables:
+//
+//
+// Making The Drop Decision
+// ------------------------
+// The example decides whether to drop the frame based on the current
+// frame number, immediately before decoding the frame.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aomdx.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s <infile> <outfile> <N-M|N/M>\n", exec_name);
+ exit(EXIT_FAILURE);
+}
+
+int main(int argc, char **argv) {
+ int frame_cnt = 0;
+ FILE *outfile = NULL;
+ aom_codec_ctx_t codec;
+ const AvxInterface *decoder = NULL;
+ AvxVideoReader *reader = NULL;
+ const AvxVideoInfo *info = NULL;
+ int n = 0;
+ int m = 0;
+ int is_range = 0;
+ char *nptr = NULL;
+
+ exec_name = argv[0];
+
+ if (argc != 4) die("Invalid number of arguments.");
+
+ reader = aom_video_reader_open(argv[1]);
+ if (!reader) die("Failed to open %s for reading.", argv[1]);
+
+ if (!(outfile = fopen(argv[2], "wb")))
+ die("Failed to open %s for writing.", argv[2]);
+
+ n = (int)strtol(argv[3], &nptr, 0);
+ m = (int)strtol(nptr + 1, NULL, 0);
+ is_range = (*nptr == '-');
+ if (!n || !m || (*nptr != '-' && *nptr != '/'))
+ die("Couldn't parse pattern %s.\n", argv[3]);
+
+ info = aom_video_reader_get_info(reader);
+
+ decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) die("Unknown input codec.");
+
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder.");
+
+ while (aom_video_reader_read_frame(reader)) {
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img = NULL;
+ size_t frame_size = 0;
+ int skip;
+ const unsigned char *frame =
+ aom_video_reader_get_frame(reader, &frame_size);
+ ++frame_cnt;
+
+ skip = (is_range && frame_cnt >= n && frame_cnt <= m) ||
+ (!is_range && m - (frame_cnt - 1) % m <= n);
+
+ if (!skip) {
+ putc('.', stdout);
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode frame.");
+
+ while ((img = aom_codec_get_frame(&codec, &iter)) != NULL)
+ aom_img_write(img, outfile);
+ } else {
+ putc('X', stdout);
+ }
+
+ fflush(stdout);
+ }
+
+ printf("Processed %d frames.\n", frame_cnt);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ printf("Play: ffplay -f rawvideo -pix_fmt yuv420p -s %dx%d %s\n",
+ info->frame_width, info->frame_height, argv[2]);
+
+ aom_video_reader_close(reader);
+ fclose(outfile);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/encoder_util.c b/media/libaom/src/examples/encoder_util.c
new file mode 100644
index 000000000..e43b37250
--- /dev/null
+++ b/media/libaom/src/examples/encoder_util.c
@@ -0,0 +1,136 @@
+/*
+ * Copyright (c) 2017, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Utility functions used by encoder binaries.
+
+#include "examples/encoder_util.h"
+
+#include <assert.h>
+#include <string.h>
+
+#include "aom/aom_integer.h"
+
+#define mmin(a, b) ((a) < (b) ? (a) : (b))
+
+static void find_mismatch_plane(const aom_image_t *const img1,
+ const aom_image_t *const img2, int plane,
+ int use_highbitdepth, int loc[4]) {
+ const unsigned char *const p1 = img1->planes[plane];
+ const int p1_stride = img1->stride[plane] >> use_highbitdepth;
+ const unsigned char *const p2 = img2->planes[plane];
+ const int p2_stride = img2->stride[plane] >> use_highbitdepth;
+ const uint32_t bsize = 64;
+ const int is_y_plane = (plane == AOM_PLANE_Y);
+ const uint32_t bsizex = is_y_plane ? bsize : bsize >> img1->x_chroma_shift;
+ const uint32_t bsizey = is_y_plane ? bsize : bsize >> img1->y_chroma_shift;
+ const uint32_t c_w =
+ is_y_plane ? img1->d_w
+ : (img1->d_w + img1->x_chroma_shift) >> img1->x_chroma_shift;
+ const uint32_t c_h =
+ is_y_plane ? img1->d_h
+ : (img1->d_h + img1->y_chroma_shift) >> img1->y_chroma_shift;
+ assert(img1->d_w == img2->d_w && img1->d_h == img2->d_h);
+ assert(img1->x_chroma_shift == img2->x_chroma_shift &&
+ img1->y_chroma_shift == img2->y_chroma_shift);
+ loc[0] = loc[1] = loc[2] = loc[3] = -1;
+ if (img1->monochrome && img2->monochrome && plane) return;
+ int match = 1;
+ uint32_t i, j;
+ for (i = 0; match && i < c_h; i += bsizey) {
+ for (j = 0; match && j < c_w; j += bsizex) {
+ const int si =
+ is_y_plane ? mmin(i + bsizey, c_h) - i : mmin(i + bsizey, c_h - i);
+ const int sj =
+ is_y_plane ? mmin(j + bsizex, c_w) - j : mmin(j + bsizex, c_w - j);
+ int k, l;
+ for (k = 0; match && k < si; ++k) {
+ for (l = 0; match && l < sj; ++l) {
+ const int row = i + k;
+ const int col = j + l;
+ const int offset1 = row * p1_stride + col;
+ const int offset2 = row * p2_stride + col;
+ const int val1 = use_highbitdepth
+ ? p1[2 * offset1] | (p1[2 * offset1 + 1] << 8)
+ : p1[offset1];
+ const int val2 = use_highbitdepth
+ ? p2[2 * offset2] | (p2[2 * offset2 + 1] << 8)
+ : p2[offset2];
+ if (val1 != val2) {
+ loc[0] = row;
+ loc[1] = col;
+ loc[2] = val1;
+ loc[3] = val2;
+ match = 0;
+ break;
+ }
+ }
+ }
+ }
+ }
+}
+
+static void find_mismatch_helper(const aom_image_t *const img1,
+ const aom_image_t *const img2,
+ int use_highbitdepth, int yloc[4], int uloc[4],
+ int vloc[4]) {
+ find_mismatch_plane(img1, img2, AOM_PLANE_Y, use_highbitdepth, yloc);
+ find_mismatch_plane(img1, img2, AOM_PLANE_U, use_highbitdepth, uloc);
+ find_mismatch_plane(img1, img2, AOM_PLANE_V, use_highbitdepth, vloc);
+}
+
+void aom_find_mismatch_high(const aom_image_t *const img1,
+ const aom_image_t *const img2, int yloc[4],
+ int uloc[4], int vloc[4]) {
+ find_mismatch_helper(img1, img2, 1, yloc, uloc, vloc);
+}
+
+void aom_find_mismatch(const aom_image_t *const img1,
+ const aom_image_t *const img2, int yloc[4], int uloc[4],
+ int vloc[4]) {
+ find_mismatch_helper(img1, img2, 0, yloc, uloc, vloc);
+}
+
+int aom_compare_img(const aom_image_t *const img1,
+ const aom_image_t *const img2) {
+ assert(img1->cp == img2->cp);
+ assert(img1->tc == img2->tc);
+ assert(img1->mc == img2->mc);
+ assert(img1->monochrome == img2->monochrome);
+
+ int num_planes = img1->monochrome ? 1 : 3;
+
+ uint32_t l_w = img1->d_w;
+ uint32_t c_w = (img1->d_w + img1->x_chroma_shift) >> img1->x_chroma_shift;
+ const uint32_t c_h =
+ (img1->d_h + img1->y_chroma_shift) >> img1->y_chroma_shift;
+ int match = 1;
+
+ match &= (img1->fmt == img2->fmt);
+ match &= (img1->d_w == img2->d_w);
+ match &= (img1->d_h == img2->d_h);
+ if (img1->fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
+ l_w *= 2;
+ c_w *= 2;
+ }
+
+ for (int plane = 0; plane < num_planes; ++plane) {
+ uint32_t height = plane ? c_h : img1->d_h;
+ uint32_t width = plane ? c_w : l_w;
+
+ for (uint32_t i = 0; i < height; ++i) {
+ match &=
+ (memcmp(img1->planes[plane] + i * img1->stride[plane],
+ img2->planes[plane] + i * img2->stride[plane], width) == 0);
+ }
+ }
+
+ return match;
+}
diff --git a/media/libaom/src/examples/encoder_util.h b/media/libaom/src/examples/encoder_util.h
new file mode 100644
index 000000000..a6bb3fb48
--- /dev/null
+++ b/media/libaom/src/examples/encoder_util.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright (c) 2017, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Utility functions used by encoder binaries.
+
+#ifndef AOM_EXAMPLES_ENCODER_UTIL_H_
+#define AOM_EXAMPLES_ENCODER_UTIL_H_
+
+#include "aom/aom_image.h"
+
+// Returns mismatch location (?loc[0],?loc[1]) and the values at that location
+// in img1 (?loc[2]) and img2 (?loc[3]).
+void aom_find_mismatch_high(const aom_image_t *const img1,
+ const aom_image_t *const img2, int yloc[4],
+ int uloc[4], int vloc[4]);
+
+void aom_find_mismatch(const aom_image_t *const img1,
+ const aom_image_t *const img2, int yloc[4], int uloc[4],
+ int vloc[4]);
+
+// Returns 1 if the two images match.
+int aom_compare_img(const aom_image_t *const img1,
+ const aom_image_t *const img2);
+
+#endif // AOM_EXAMPLES_ENCODER_UTIL_H_
diff --git a/media/libaom/src/examples/inspect.c b/media/libaom/src/examples/inspect.c
new file mode 100644
index 000000000..9d5f0dcfc
--- /dev/null
+++ b/media/libaom/src/examples/inspect.c
@@ -0,0 +1,763 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Inspect Decoder
+// ================
+//
+// This is a simple decoder loop that writes JSON stats to stdout. This tool
+// can also be compiled with Emscripten and used as a library.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#ifdef __EMSCRIPTEN__
+#include <emscripten.h>
+#else
+#define EMSCRIPTEN_KEEPALIVE
+#endif
+
+#include "config/aom_config.h"
+
+#include "aom/aom_decoder.h"
+#include "aom/aomdx.h"
+#include "av1/common/onyxc_int.h"
+
+#if CONFIG_ACCOUNTING
+#include "av1/decoder/accounting.h"
+#endif
+
+#include "av1/decoder/inspection.h"
+#include "common/args.h"
+#include "common/tools_common.h"
+#include "common/video_common.h"
+#include "common/video_reader.h"
+
+// Max JSON buffer size.
+const int MAX_BUFFER = 1024 * 1024 * 32;
+
+typedef enum {
+ ACCOUNTING_LAYER = 1,
+ BLOCK_SIZE_LAYER = 1 << 1,
+ TRANSFORM_SIZE_LAYER = 1 << 2,
+ TRANSFORM_TYPE_LAYER = 1 << 3,
+ MODE_LAYER = 1 << 4,
+ SKIP_LAYER = 1 << 5,
+ FILTER_LAYER = 1 << 6,
+ CDEF_LAYER = 1 << 7,
+ REFERENCE_FRAME_LAYER = 1 << 8,
+ MOTION_VECTORS_LAYER = 1 << 9,
+ UV_MODE_LAYER = 1 << 10,
+ CFL_LAYER = 1 << 11,
+ DUAL_FILTER_LAYER = 1 << 12,
+ Q_INDEX_LAYER = 1 << 13,
+ SEGMENT_ID_LAYER = 1 << 14,
+ ALL_LAYERS = (1 << 15) - 1
+} LayerType;
+
+static LayerType layers = 0;
+
+static int stop_after = 0;
+static int compress = 0;
+
+static const arg_def_t limit_arg =
+ ARG_DEF(NULL, "limit", 1, "Stop decoding after n frames");
+static const arg_def_t dump_all_arg = ARG_DEF("A", "all", 0, "Dump All");
+static const arg_def_t compress_arg =
+ ARG_DEF("x", "compress", 0, "Compress JSON using RLE");
+static const arg_def_t dump_accounting_arg =
+ ARG_DEF("a", "accounting", 0, "Dump Accounting");
+static const arg_def_t dump_block_size_arg =
+ ARG_DEF("bs", "blockSize", 0, "Dump Block Size");
+static const arg_def_t dump_motion_vectors_arg =
+ ARG_DEF("mv", "motionVectors", 0, "Dump Motion Vectors");
+static const arg_def_t dump_transform_size_arg =
+ ARG_DEF("ts", "transformSize", 0, "Dump Transform Size");
+static const arg_def_t dump_transform_type_arg =
+ ARG_DEF("tt", "transformType", 0, "Dump Transform Type");
+static const arg_def_t dump_mode_arg = ARG_DEF("m", "mode", 0, "Dump Mode");
+static const arg_def_t dump_uv_mode_arg =
+ ARG_DEF("uvm", "uv_mode", 0, "Dump UV Intra Prediction Modes");
+static const arg_def_t dump_skip_arg = ARG_DEF("s", "skip", 0, "Dump Skip");
+static const arg_def_t dump_filter_arg =
+ ARG_DEF("f", "filter", 0, "Dump Filter");
+static const arg_def_t dump_cdef_arg = ARG_DEF("c", "cdef", 0, "Dump CDEF");
+static const arg_def_t dump_cfl_arg =
+ ARG_DEF("cfl", "chroma_from_luma", 0, "Dump Chroma from Luma Alphas");
+static const arg_def_t dump_dual_filter_type_arg =
+ ARG_DEF("df", "dualFilterType", 0, "Dump Dual Filter Type");
+static const arg_def_t dump_reference_frame_arg =
+ ARG_DEF("r", "referenceFrame", 0, "Dump Reference Frame");
+static const arg_def_t dump_delta_q_arg =
+ ARG_DEF("dq", "delta_q", 0, "Dump QIndex");
+static const arg_def_t dump_seg_id_arg =
+ ARG_DEF("si", "seg_id", 0, "Dump Segment ID");
+static const arg_def_t usage_arg = ARG_DEF("h", "help", 0, "Help");
+
+static const arg_def_t *main_args[] = { &limit_arg,
+ &dump_all_arg,
+ &compress_arg,
+#if CONFIG_ACCOUNTING
+ &dump_accounting_arg,
+#endif
+ &dump_block_size_arg,
+ &dump_transform_size_arg,
+ &dump_transform_type_arg,
+ &dump_mode_arg,
+ &dump_uv_mode_arg,
+ &dump_skip_arg,
+ &dump_filter_arg,
+ &dump_cdef_arg,
+ &dump_dual_filter_type_arg,
+ &dump_cfl_arg,
+ &dump_reference_frame_arg,
+ &dump_motion_vectors_arg,
+ &dump_delta_q_arg,
+ &dump_seg_id_arg,
+ &usage_arg,
+ NULL };
+#define ENUM(name) \
+ { #name, name }
+#define LAST_ENUM \
+ { NULL, 0 }
+typedef struct map_entry {
+ const char *name;
+ int value;
+} map_entry;
+
+const map_entry refs_map[] = {
+ ENUM(INTRA_FRAME), ENUM(LAST_FRAME), ENUM(LAST2_FRAME),
+ ENUM(LAST3_FRAME), ENUM(GOLDEN_FRAME), ENUM(BWDREF_FRAME),
+ ENUM(ALTREF2_FRAME), ENUM(ALTREF_FRAME), LAST_ENUM
+};
+
+const map_entry block_size_map[] = {
+ ENUM(BLOCK_4X4), ENUM(BLOCK_4X8), ENUM(BLOCK_8X4),
+ ENUM(BLOCK_8X8), ENUM(BLOCK_8X16), ENUM(BLOCK_16X8),
+ ENUM(BLOCK_16X16), ENUM(BLOCK_16X32), ENUM(BLOCK_32X16),
+ ENUM(BLOCK_32X32), ENUM(BLOCK_32X64), ENUM(BLOCK_64X32),
+ ENUM(BLOCK_64X64), ENUM(BLOCK_64X128), ENUM(BLOCK_128X64),
+ ENUM(BLOCK_128X128), ENUM(BLOCK_4X16), ENUM(BLOCK_16X4),
+ ENUM(BLOCK_8X32), ENUM(BLOCK_32X8), ENUM(BLOCK_16X64),
+ ENUM(BLOCK_64X16), LAST_ENUM
+};
+
+const map_entry tx_size_map[] = {
+ ENUM(TX_4X4), ENUM(TX_8X8), ENUM(TX_16X16), ENUM(TX_32X32),
+ ENUM(TX_64X64), ENUM(TX_4X8), ENUM(TX_8X4), ENUM(TX_8X16),
+ ENUM(TX_16X8), ENUM(TX_16X32), ENUM(TX_32X16), ENUM(TX_32X64),
+ ENUM(TX_64X32), ENUM(TX_4X16), ENUM(TX_16X4), ENUM(TX_8X32),
+ ENUM(TX_32X8), LAST_ENUM
+};
+
+const map_entry tx_type_map[] = { ENUM(DCT_DCT),
+ ENUM(ADST_DCT),
+ ENUM(DCT_ADST),
+ ENUM(ADST_ADST),
+ ENUM(FLIPADST_DCT),
+ ENUM(DCT_FLIPADST),
+ ENUM(FLIPADST_FLIPADST),
+ ENUM(ADST_FLIPADST),
+ ENUM(FLIPADST_ADST),
+ ENUM(IDTX),
+ ENUM(V_DCT),
+ ENUM(H_DCT),
+ ENUM(V_ADST),
+ ENUM(H_ADST),
+ ENUM(V_FLIPADST),
+ ENUM(H_FLIPADST),
+ LAST_ENUM };
+const map_entry dual_filter_map[] = { ENUM(REG_REG), ENUM(REG_SMOOTH),
+ ENUM(REG_SHARP), ENUM(SMOOTH_REG),
+ ENUM(SMOOTH_SMOOTH), ENUM(SMOOTH_SHARP),
+ ENUM(SHARP_REG), ENUM(SHARP_SMOOTH),
+ ENUM(SHARP_SHARP), LAST_ENUM };
+
+const map_entry prediction_mode_map[] = {
+ ENUM(DC_PRED), ENUM(V_PRED), ENUM(H_PRED),
+ ENUM(D45_PRED), ENUM(D135_PRED), ENUM(D113_PRED),
+ ENUM(D157_PRED), ENUM(D203_PRED), ENUM(D67_PRED),
+ ENUM(SMOOTH_PRED), ENUM(SMOOTH_V_PRED), ENUM(SMOOTH_H_PRED),
+ ENUM(PAETH_PRED), ENUM(NEARESTMV), ENUM(NEARMV),
+ ENUM(GLOBALMV), ENUM(NEWMV), ENUM(NEAREST_NEARESTMV),
+ ENUM(NEAR_NEARMV), ENUM(NEAREST_NEWMV), ENUM(NEW_NEARESTMV),
+ ENUM(NEAR_NEWMV), ENUM(NEW_NEARMV), ENUM(GLOBAL_GLOBALMV),
+ ENUM(NEW_NEWMV), ENUM(INTRA_INVALID), LAST_ENUM
+};
+
+const map_entry uv_prediction_mode_map[] = {
+ ENUM(UV_DC_PRED), ENUM(UV_V_PRED),
+ ENUM(UV_H_PRED), ENUM(UV_D45_PRED),
+ ENUM(UV_D135_PRED), ENUM(UV_D113_PRED),
+ ENUM(UV_D157_PRED), ENUM(UV_D203_PRED),
+ ENUM(UV_D67_PRED), ENUM(UV_SMOOTH_PRED),
+ ENUM(UV_SMOOTH_V_PRED), ENUM(UV_SMOOTH_H_PRED),
+ ENUM(UV_PAETH_PRED), ENUM(UV_CFL_PRED),
+ ENUM(UV_MODE_INVALID), LAST_ENUM
+};
+#define NO_SKIP 0
+#define SKIP 1
+
+const map_entry skip_map[] = { ENUM(SKIP), ENUM(NO_SKIP), LAST_ENUM };
+
+const map_entry config_map[] = { ENUM(MI_SIZE), LAST_ENUM };
+
+static const char *exec_name;
+
+insp_frame_data frame_data;
+int frame_count = 0;
+int decoded_frame_count = 0;
+aom_codec_ctx_t codec;
+AvxVideoReader *reader = NULL;
+const AvxVideoInfo *info = NULL;
+aom_image_t *img = NULL;
+
+void on_frame_decoded_dump(char *json) {
+#ifdef __EMSCRIPTEN__
+ EM_ASM_({ Module.on_frame_decoded_json($0); }, json);
+#else
+ printf("%s", json);
+#endif
+}
+
+// Writing out the JSON buffer using snprintf is very slow, especially when
+// compiled with emscripten, these functions speed things up quite a bit.
+int put_str(char *buffer, const char *str) {
+ int i;
+ for (i = 0; str[i] != '\0'; i++) {
+ buffer[i] = str[i];
+ }
+ return i;
+}
+
+int put_str_with_escape(char *buffer, const char *str) {
+ int i;
+ int j = 0;
+ for (i = 0; str[i] != '\0'; i++) {
+ if (str[i] < ' ') {
+ continue;
+ } else if (str[i] == '"' || str[i] == '\\') {
+ buffer[j++] = '\\';
+ }
+ buffer[j++] = str[i];
+ }
+ return j;
+}
+
+int put_num(char *buffer, char prefix, int num, char suffix) {
+ int i = 0;
+ char *buf = buffer;
+ int is_neg = 0;
+ if (prefix) {
+ buf[i++] = prefix;
+ }
+ if (num == 0) {
+ buf[i++] = '0';
+ } else {
+ if (num < 0) {
+ num = -num;
+ is_neg = 1;
+ }
+ int s = i;
+ while (num != 0) {
+ buf[i++] = '0' + (num % 10);
+ num = num / 10;
+ }
+ if (is_neg) {
+ buf[i++] = '-';
+ }
+ int e = i - 1;
+ while (s < e) {
+ int t = buf[s];
+ buf[s] = buf[e];
+ buf[e] = t;
+ s++;
+ e--;
+ }
+ }
+ if (suffix) {
+ buf[i++] = suffix;
+ }
+ return i;
+}
+
+int put_map(char *buffer, const map_entry *map) {
+ char *buf = buffer;
+ const map_entry *entry = map;
+ while (entry->name != NULL) {
+ *(buf++) = '"';
+ buf += put_str(buf, entry->name);
+ *(buf++) = '"';
+ buf += put_num(buf, ':', entry->value, 0);
+ entry++;
+ if (entry->name != NULL) {
+ *(buf++) = ',';
+ }
+ }
+ return (int)(buf - buffer);
+}
+
+int put_reference_frame(char *buffer) {
+ const int mi_rows = frame_data.mi_rows;
+ const int mi_cols = frame_data.mi_cols;
+ char *buf = buffer;
+ int r, c, t;
+ buf += put_str(buf, " \"referenceFrameMap\": {");
+ buf += put_map(buf, refs_map);
+ buf += put_str(buf, "},\n");
+ buf += put_str(buf, " \"referenceFrame\": [");
+ for (r = 0; r < mi_rows; ++r) {
+ *(buf++) = '[';
+ for (c = 0; c < mi_cols; ++c) {
+ insp_mi_data *mi = &frame_data.mi_grid[r * mi_cols + c];
+ buf += put_num(buf, '[', mi->ref_frame[0], 0);
+ buf += put_num(buf, ',', mi->ref_frame[1], ']');
+ if (compress) { // RLE
+ for (t = c + 1; t < mi_cols; ++t) {
+ insp_mi_data *next_mi = &frame_data.mi_grid[r * mi_cols + t];
+ if (mi->ref_frame[0] != next_mi->ref_frame[0] ||
+ mi->ref_frame[1] != next_mi->ref_frame[1]) {
+ break;
+ }
+ }
+ if (t - c > 1) {
+ *(buf++) = ',';
+ buf += put_num(buf, '[', t - c - 1, ']');
+ c = t - 1;
+ }
+ }
+ if (c < mi_cols - 1) *(buf++) = ',';
+ }
+ *(buf++) = ']';
+ if (r < mi_rows - 1) *(buf++) = ',';
+ }
+ buf += put_str(buf, "],\n");
+ return (int)(buf - buffer);
+}
+
+int put_motion_vectors(char *buffer) {
+ const int mi_rows = frame_data.mi_rows;
+ const int mi_cols = frame_data.mi_cols;
+ char *buf = buffer;
+ int r, c, t;
+ buf += put_str(buf, " \"motionVectors\": [");
+ for (r = 0; r < mi_rows; ++r) {
+ *(buf++) = '[';
+ for (c = 0; c < mi_cols; ++c) {
+ insp_mi_data *mi = &frame_data.mi_grid[r * mi_cols + c];
+ buf += put_num(buf, '[', mi->mv[0].col, 0);
+ buf += put_num(buf, ',', mi->mv[0].row, 0);
+ buf += put_num(buf, ',', mi->mv[1].col, 0);
+ buf += put_num(buf, ',', mi->mv[1].row, ']');
+ if (compress) { // RLE
+ for (t = c + 1; t < mi_cols; ++t) {
+ insp_mi_data *next_mi = &frame_data.mi_grid[r * mi_cols + t];
+ if (mi->mv[0].col != next_mi->mv[0].col ||
+ mi->mv[0].row != next_mi->mv[0].row ||
+ mi->mv[1].col != next_mi->mv[1].col ||
+ mi->mv[1].row != next_mi->mv[1].row) {
+ break;
+ }
+ }
+ if (t - c > 1) {
+ *(buf++) = ',';
+ buf += put_num(buf, '[', t - c - 1, ']');
+ c = t - 1;
+ }
+ }
+ if (c < mi_cols - 1) *(buf++) = ',';
+ }
+ *(buf++) = ']';
+ if (r < mi_rows - 1) *(buf++) = ',';
+ }
+ buf += put_str(buf, "],\n");
+ return (int)(buf - buffer);
+}
+
+int put_block_info(char *buffer, const map_entry *map, const char *name,
+ size_t offset, int len) {
+ const int mi_rows = frame_data.mi_rows;
+ const int mi_cols = frame_data.mi_cols;
+ char *buf = buffer;
+ int r, c, t, i;
+ if (compress && len == 1) {
+ die("Can't encode scalars as arrays when RLE compression is enabled.");
+ return -1;
+ }
+ if (map) {
+ buf += snprintf(buf, MAX_BUFFER, " \"%sMap\": {", name);
+ buf += put_map(buf, map);
+ buf += put_str(buf, "},\n");
+ }
+ buf += snprintf(buf, MAX_BUFFER, " \"%s\": [", name);
+ for (r = 0; r < mi_rows; ++r) {
+ *(buf++) = '[';
+ for (c = 0; c < mi_cols; ++c) {
+ insp_mi_data *mi = &frame_data.mi_grid[r * mi_cols + c];
+ int16_t *v = (int16_t *)(((int8_t *)mi) + offset);
+ if (len == 0) {
+ buf += put_num(buf, 0, v[0], 0);
+ } else {
+ buf += put_str(buf, "[");
+ for (i = 0; i < len; i++) {
+ buf += put_num(buf, 0, v[i], 0);
+ if (i < len - 1) {
+ buf += put_str(buf, ",");
+ }
+ }
+ buf += put_str(buf, "]");
+ }
+ if (compress) { // RLE
+ for (t = c + 1; t < mi_cols; ++t) {
+ insp_mi_data *next_mi = &frame_data.mi_grid[r * mi_cols + t];
+ int16_t *nv = (int16_t *)(((int8_t *)next_mi) + offset);
+ int same = 0;
+ if (len == 0) {
+ same = v[0] == nv[0];
+ } else {
+ for (i = 0; i < len; i++) {
+ same = v[i] == nv[i];
+ if (!same) {
+ break;
+ }
+ }
+ }
+ if (!same) {
+ break;
+ }
+ }
+ if (t - c > 1) {
+ *(buf++) = ',';
+ buf += put_num(buf, '[', t - c - 1, ']');
+ c = t - 1;
+ }
+ }
+ if (c < mi_cols - 1) *(buf++) = ',';
+ }
+ *(buf++) = ']';
+ if (r < mi_rows - 1) *(buf++) = ',';
+ }
+ buf += put_str(buf, "],\n");
+ return (int)(buf - buffer);
+}
+
+#if CONFIG_ACCOUNTING
+int put_accounting(char *buffer) {
+ char *buf = buffer;
+ int i;
+ const Accounting *accounting = frame_data.accounting;
+ if (accounting == NULL) {
+ printf("XXX\n");
+ return 0;
+ }
+ const int num_syms = accounting->syms.num_syms;
+ const int num_strs = accounting->syms.dictionary.num_strs;
+ buf += put_str(buf, " \"symbolsMap\": [");
+ for (i = 0; i < num_strs; i++) {
+ buf += snprintf(buf, MAX_BUFFER, "\"%s\"",
+ accounting->syms.dictionary.strs[i]);
+ if (i < num_strs - 1) *(buf++) = ',';
+ }
+ buf += put_str(buf, "],\n");
+ buf += put_str(buf, " \"symbols\": [\n ");
+ AccountingSymbolContext context;
+ context.x = -2;
+ context.y = -2;
+ AccountingSymbol *sym;
+ for (i = 0; i < num_syms; i++) {
+ sym = &accounting->syms.syms[i];
+ if (memcmp(&context, &sym->context, sizeof(AccountingSymbolContext)) != 0) {
+ buf += put_num(buf, '[', sym->context.x, 0);
+ buf += put_num(buf, ',', sym->context.y, ']');
+ } else {
+ buf += put_num(buf, '[', sym->id, 0);
+ buf += put_num(buf, ',', sym->bits, 0);
+ buf += put_num(buf, ',', sym->samples, ']');
+ }
+ context = sym->context;
+ if (i < num_syms - 1) *(buf++) = ',';
+ }
+ buf += put_str(buf, "],\n");
+ return (int)(buf - buffer);
+}
+#endif
+
+void inspect(void *pbi, void *data) {
+ /* Fetch frame data. */
+ ifd_inspect(&frame_data, pbi);
+ (void)data;
+ // We allocate enough space and hope we don't write out of bounds. Totally
+ // unsafe but this speeds things up, especially when compiled to Javascript.
+ char *buffer = aom_malloc(MAX_BUFFER);
+ char *buf = buffer;
+ buf += put_str(buf, "{\n");
+ if (layers & BLOCK_SIZE_LAYER) {
+ buf += put_block_info(buf, block_size_map, "blockSize",
+ offsetof(insp_mi_data, sb_type), 0);
+ }
+ if (layers & TRANSFORM_SIZE_LAYER) {
+ buf += put_block_info(buf, tx_size_map, "transformSize",
+ offsetof(insp_mi_data, tx_size), 0);
+ }
+ if (layers & TRANSFORM_TYPE_LAYER) {
+ buf += put_block_info(buf, tx_type_map, "transformType",
+ offsetof(insp_mi_data, tx_type), 0);
+ }
+ if (layers & DUAL_FILTER_LAYER) {
+ buf += put_block_info(buf, dual_filter_map, "dualFilterType",
+ offsetof(insp_mi_data, dual_filter_type), 0);
+ }
+ if (layers & MODE_LAYER) {
+ buf += put_block_info(buf, prediction_mode_map, "mode",
+ offsetof(insp_mi_data, mode), 0);
+ }
+ if (layers & UV_MODE_LAYER) {
+ buf += put_block_info(buf, uv_prediction_mode_map, "uv_mode",
+ offsetof(insp_mi_data, uv_mode), 0);
+ }
+ if (layers & SKIP_LAYER) {
+ buf +=
+ put_block_info(buf, skip_map, "skip", offsetof(insp_mi_data, skip), 0);
+ }
+ if (layers & FILTER_LAYER) {
+ buf +=
+ put_block_info(buf, NULL, "filter", offsetof(insp_mi_data, filter), 2);
+ }
+ if (layers & CDEF_LAYER) {
+ buf += put_block_info(buf, NULL, "cdef_level",
+ offsetof(insp_mi_data, cdef_level), 0);
+ buf += put_block_info(buf, NULL, "cdef_strength",
+ offsetof(insp_mi_data, cdef_strength), 0);
+ }
+ if (layers & CFL_LAYER) {
+ buf += put_block_info(buf, NULL, "cfl_alpha_idx",
+ offsetof(insp_mi_data, cfl_alpha_idx), 0);
+ buf += put_block_info(buf, NULL, "cfl_alpha_sign",
+ offsetof(insp_mi_data, cfl_alpha_sign), 0);
+ }
+ if (layers & Q_INDEX_LAYER) {
+ buf += put_block_info(buf, NULL, "delta_q",
+ offsetof(insp_mi_data, current_qindex), 0);
+ }
+ if (layers & SEGMENT_ID_LAYER) {
+ buf += put_block_info(buf, NULL, "seg_id",
+ offsetof(insp_mi_data, segment_id), 0);
+ }
+ if (layers & MOTION_VECTORS_LAYER) {
+ buf += put_motion_vectors(buf);
+ }
+ if (layers & REFERENCE_FRAME_LAYER) {
+ buf += put_block_info(buf, refs_map, "referenceFrame",
+ offsetof(insp_mi_data, ref_frame), 2);
+ }
+#if CONFIG_ACCOUNTING
+ if (layers & ACCOUNTING_LAYER) {
+ buf += put_accounting(buf);
+ }
+#endif
+ buf += snprintf(buf, MAX_BUFFER, " \"frame\": %d,\n", decoded_frame_count);
+ buf += snprintf(buf, MAX_BUFFER, " \"showFrame\": %d,\n",
+ frame_data.show_frame);
+ buf += snprintf(buf, MAX_BUFFER, " \"frameType\": %d,\n",
+ frame_data.frame_type);
+ buf += snprintf(buf, MAX_BUFFER, " \"baseQIndex\": %d,\n",
+ frame_data.base_qindex);
+ buf += snprintf(buf, MAX_BUFFER, " \"tileCols\": %d,\n",
+ frame_data.tile_mi_cols);
+ buf += snprintf(buf, MAX_BUFFER, " \"tileRows\": %d,\n",
+ frame_data.tile_mi_rows);
+ buf += snprintf(buf, MAX_BUFFER, " \"deltaQPresentFlag\": %d,\n",
+ frame_data.delta_q_present_flag);
+ buf += snprintf(buf, MAX_BUFFER, " \"deltaQRes\": %d,\n",
+ frame_data.delta_q_res);
+ buf += put_str(buf, " \"config\": {");
+ buf += put_map(buf, config_map);
+ buf += put_str(buf, "},\n");
+ buf += put_str(buf, " \"configString\": \"");
+ buf += put_str_with_escape(buf, aom_codec_build_config());
+ buf += put_str(buf, "\"\n");
+ decoded_frame_count++;
+ buf += put_str(buf, "},\n");
+ *(buf++) = 0;
+ on_frame_decoded_dump(buffer);
+ aom_free(buffer);
+}
+
+void ifd_init_cb() {
+ aom_inspect_init ii;
+ ii.inspect_cb = inspect;
+ ii.inspect_ctx = NULL;
+ aom_codec_control(&codec, AV1_SET_INSPECTION_CALLBACK, &ii);
+}
+
+EMSCRIPTEN_KEEPALIVE
+int open_file(char *file) {
+ if (file == NULL) {
+ // The JS analyzer puts the .ivf file at this location.
+ file = "/tmp/input.ivf";
+ }
+ reader = aom_video_reader_open(file);
+ if (!reader) die("Failed to open %s for reading.", file);
+ info = aom_video_reader_get_info(reader);
+ const AvxInterface *decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) die("Unknown input codec.");
+ fprintf(stderr, "Using %s\n",
+ aom_codec_iface_name(decoder->codec_interface()));
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder.");
+ ifd_init(&frame_data, info->frame_width, info->frame_height);
+ ifd_init_cb();
+ return EXIT_SUCCESS;
+}
+
+EMSCRIPTEN_KEEPALIVE
+int read_frame() {
+ if (!aom_video_reader_read_frame(reader)) return EXIT_FAILURE;
+ img = NULL;
+ aom_codec_iter_t iter = NULL;
+ size_t frame_size = 0;
+ const unsigned char *frame = aom_video_reader_get_frame(reader, &frame_size);
+ if (aom_codec_decode(&codec, frame, (unsigned int)frame_size, NULL) !=
+ AOM_CODEC_OK) {
+ die_codec(&codec, "Failed to decode frame.");
+ }
+ int got_any_frames = 0;
+ aom_image_t *frame_img;
+ while ((frame_img = aom_codec_get_frame(&codec, &iter))) {
+ img = frame_img;
+ ++frame_count;
+ got_any_frames = 1;
+ }
+ if (!got_any_frames) {
+ return EXIT_FAILURE;
+ }
+ return EXIT_SUCCESS;
+}
+
+EMSCRIPTEN_KEEPALIVE
+const char *get_aom_codec_build_config() { return aom_codec_build_config(); }
+
+EMSCRIPTEN_KEEPALIVE
+int get_bit_depth() { return img->bit_depth; }
+
+EMSCRIPTEN_KEEPALIVE
+int get_bits_per_sample() { return img->bps; }
+
+EMSCRIPTEN_KEEPALIVE
+int get_image_format() { return img->fmt; }
+
+EMSCRIPTEN_KEEPALIVE
+unsigned char *get_plane(int plane) { return img->planes[plane]; }
+
+EMSCRIPTEN_KEEPALIVE
+int get_plane_stride(int plane) { return img->stride[plane]; }
+
+EMSCRIPTEN_KEEPALIVE
+int get_plane_width(int plane) { return aom_img_plane_width(img, plane); }
+
+EMSCRIPTEN_KEEPALIVE
+int get_plane_height(int plane) { return aom_img_plane_height(img, plane); }
+
+EMSCRIPTEN_KEEPALIVE
+int get_frame_width() { return info->frame_width; }
+
+EMSCRIPTEN_KEEPALIVE
+int get_frame_height() { return info->frame_height; }
+
+static void parse_args(char **argv) {
+ char **argi, **argj;
+ struct arg arg;
+ (void)dump_accounting_arg;
+ (void)dump_cdef_arg;
+ for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step) {
+ arg.argv_step = 1;
+ if (arg_match(&arg, &dump_block_size_arg, argi)) layers |= BLOCK_SIZE_LAYER;
+#if CONFIG_ACCOUNTING
+ else if (arg_match(&arg, &dump_accounting_arg, argi))
+ layers |= ACCOUNTING_LAYER;
+#endif
+ else if (arg_match(&arg, &dump_transform_size_arg, argi))
+ layers |= TRANSFORM_SIZE_LAYER;
+ else if (arg_match(&arg, &dump_transform_type_arg, argi))
+ layers |= TRANSFORM_TYPE_LAYER;
+ else if (arg_match(&arg, &dump_mode_arg, argi))
+ layers |= MODE_LAYER;
+ else if (arg_match(&arg, &dump_uv_mode_arg, argi))
+ layers |= UV_MODE_LAYER;
+ else if (arg_match(&arg, &dump_skip_arg, argi))
+ layers |= SKIP_LAYER;
+ else if (arg_match(&arg, &dump_filter_arg, argi))
+ layers |= FILTER_LAYER;
+ else if (arg_match(&arg, &dump_cdef_arg, argi))
+ layers |= CDEF_LAYER;
+ else if (arg_match(&arg, &dump_cfl_arg, argi))
+ layers |= CFL_LAYER;
+ else if (arg_match(&arg, &dump_reference_frame_arg, argi))
+ layers |= REFERENCE_FRAME_LAYER;
+ else if (arg_match(&arg, &dump_motion_vectors_arg, argi))
+ layers |= MOTION_VECTORS_LAYER;
+ else if (arg_match(&arg, &dump_dual_filter_type_arg, argi))
+ layers |= DUAL_FILTER_LAYER;
+ else if (arg_match(&arg, &dump_delta_q_arg, argi))
+ layers |= Q_INDEX_LAYER;
+ else if (arg_match(&arg, &dump_seg_id_arg, argi))
+ layers |= SEGMENT_ID_LAYER;
+ else if (arg_match(&arg, &dump_all_arg, argi))
+ layers |= ALL_LAYERS;
+ else if (arg_match(&arg, &compress_arg, argi))
+ compress = 1;
+ else if (arg_match(&arg, &usage_arg, argi))
+ usage_exit();
+ else if (arg_match(&arg, &limit_arg, argi))
+ stop_after = arg_parse_uint(&arg);
+ else
+ argj++;
+ }
+}
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s src_filename <options>\n", exec_name);
+ fprintf(stderr, "\nOptions:\n");
+ arg_show_usage(stderr, main_args);
+ exit(EXIT_FAILURE);
+}
+
+EMSCRIPTEN_KEEPALIVE
+int main(int argc, char **argv) {
+ exec_name = argv[0];
+ parse_args(argv);
+ if (argc >= 2) {
+ open_file(argv[1]);
+ printf("[\n");
+ while (1) {
+ if (stop_after && (decoded_frame_count >= stop_after)) break;
+ if (read_frame()) break;
+ }
+ printf("null\n");
+ printf("]");
+ } else {
+ usage_exit();
+ }
+}
+
+EMSCRIPTEN_KEEPALIVE
+void quit() {
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
+ aom_video_reader_close(reader);
+}
+
+EMSCRIPTEN_KEEPALIVE
+void set_layers(LayerType v) { layers = v; }
+
+EMSCRIPTEN_KEEPALIVE
+void set_compress(int v) { compress = v; }
diff --git a/media/libaom/src/examples/lightfield_bitstream_parsing.c b/media/libaom/src/examples/lightfield_bitstream_parsing.c
new file mode 100644
index 000000000..159f1617a
--- /dev/null
+++ b/media/libaom/src/examples/lightfield_bitstream_parsing.c
@@ -0,0 +1,348 @@
+/*
+ * Copyright (c) 2018, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Lightfield Bitstream Parsing
+// ============================
+//
+// This is a lightfield bitstream parsing example. It takes an input file
+// containing the whole compressed lightfield bitstream(ivf file), and parses it
+// and constructs and outputs a new bitstream that can be decoded by an AV1
+// decoder. The output bitstream contains reference frames(i.e. anchor frames),
+// camera frame header, and tile list OBUs. num_references is the number of
+// anchor frames coded at the beginning of the light field file.
+// After running the lightfield encoder, run lightfield bitstream parsing:
+// examples/lightfield_bitstream_parsing vase10x10.ivf vase_tile_list.ivf 4
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aom_encoder.h"
+#include "aom/aom_integer.h"
+#include "aom/aomdx.h"
+#include "aom_dsp/bitwriter_buffer.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+#include "common/video_writer.h"
+
+#define MAX_TILES 512
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s <infile> <outfile> <num_references> \n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+#define ALIGN_POWER_OF_TWO(value, n) \
+ (((value) + ((1 << (n)) - 1)) & ~((1 << (n)) - 1))
+
+// SB size: 64x64
+const uint8_t output_frame_width_in_tiles_minus_1 = 512 / 64 - 1;
+const uint8_t output_frame_height_in_tiles_minus_1 = 512 / 64 - 1;
+
+// Spec:
+// typedef struct {
+// uint8_t anchor_frame_idx;
+// uint8_t tile_row;
+// uint8_t tile_col;
+// uint16_t coded_tile_data_size_minus_1;
+// uint8_t *coded_tile_data;
+// } TILE_LIST_ENTRY;
+
+// Tile list entry provided by the application
+typedef struct {
+ int image_idx;
+ int reference_idx;
+ int tile_col;
+ int tile_row;
+} TILE_LIST_INFO;
+
+// M references: 0 - M-1; N images(including references): 0 - N-1;
+// Note: order the image index incrementally, so that we only go through the
+// bitstream once to construct the tile list.
+const int num_tile_lists = 2;
+const uint16_t tile_count_minus_1 = 9 - 1;
+const TILE_LIST_INFO tile_list[2][9] = {
+ { { 16, 0, 4, 5 },
+ { 83, 3, 13, 2 },
+ { 57, 2, 2, 6 },
+ { 31, 1, 11, 5 },
+ { 2, 0, 7, 4 },
+ { 77, 3, 9, 9 },
+ { 49, 1, 0, 1 },
+ { 6, 0, 3, 10 },
+ { 63, 2, 5, 8 } },
+ { { 65, 2, 11, 1 },
+ { 42, 1, 3, 7 },
+ { 88, 3, 8, 4 },
+ { 76, 3, 1, 15 },
+ { 1, 0, 2, 2 },
+ { 19, 0, 5, 6 },
+ { 60, 2, 4, 0 },
+ { 25, 1, 11, 15 },
+ { 50, 2, 5, 4 } },
+};
+
+static int get_image_bps(aom_img_fmt_t fmt) {
+ switch (fmt) {
+ case AOM_IMG_FMT_I420: return 12;
+ case AOM_IMG_FMT_I422: return 16;
+ case AOM_IMG_FMT_I444: return 24;
+ case AOM_IMG_FMT_I42016: return 24;
+ case AOM_IMG_FMT_I42216: return 32;
+ case AOM_IMG_FMT_I44416: return 48;
+ default: die("Invalid image format");
+ }
+ return 0;
+}
+
+int main(int argc, char **argv) {
+ aom_codec_ctx_t codec;
+ AvxVideoReader *reader = NULL;
+ AvxVideoWriter *writer = NULL;
+ const AvxInterface *decoder = NULL;
+ const AvxVideoInfo *info = NULL;
+ int num_references;
+ int n, i;
+ aom_codec_pts_t pts;
+
+ exec_name = argv[0];
+ if (argc != 4) die("Invalid number of arguments.");
+
+ reader = aom_video_reader_open(argv[1]);
+ if (!reader) die("Failed to open %s for reading.", argv[1]);
+
+ num_references = (int)strtol(argv[3], NULL, 0);
+ info = aom_video_reader_get_info(reader);
+
+ // The writer to write out ivf file in tile list OBU, which can be decoded by
+ // AV1 decoder.
+ writer = aom_video_writer_open(argv[2], kContainerIVF, info);
+ if (!writer) die("Failed to open %s for writing", argv[2]);
+
+ decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) die("Unknown input codec.");
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder.");
+
+ // Decode anchor frames.
+ aom_codec_control_(&codec, AV1_SET_TILE_MODE, 0);
+
+ for (i = 0; i < num_references; ++i) {
+ aom_video_reader_read_frame(reader);
+
+ size_t frame_size = 0;
+ const unsigned char *frame =
+ aom_video_reader_get_frame(reader, &frame_size);
+ pts = (aom_codec_pts_t)aom_video_reader_get_frame_pts(reader);
+
+ // Copy references bitstream directly.
+ if (!aom_video_writer_write_frame(writer, frame, frame_size, pts))
+ die_codec(&codec, "Failed to copy compressed anchor frame.");
+
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode frame.");
+ }
+
+ // Decode camera frames.
+ aom_codec_control_(&codec, AV1_SET_TILE_MODE, 1);
+ aom_codec_control_(&codec, AV1D_EXT_TILE_DEBUG, 1);
+
+ FILE *infile = aom_video_reader_get_file(reader);
+ // Record the offset of the first camera image.
+ const FileOffset camera_frame_pos = ftello(infile);
+
+ // Read out the first camera frame.
+ aom_video_reader_read_frame(reader);
+
+ // Copy first camera frame for getting camera frame header. This is done
+ // only once.
+ {
+ size_t frame_size = 0;
+ const unsigned char *frame =
+ aom_video_reader_get_frame(reader, &frame_size);
+ pts = (aom_codec_pts_t)aom_video_reader_get_frame_pts(reader);
+ aom_tile_data frame_header_info = { 0, NULL, 0 };
+
+ // Need to decode frame header to get camera frame header info. So, here
+ // decoding 1 tile is enough.
+ aom_codec_control_(&codec, AV1_SET_DECODE_TILE_ROW, 0);
+ aom_codec_control_(&codec, AV1_SET_DECODE_TILE_COL, 0);
+
+ aom_codec_err_t aom_status =
+ aom_codec_decode(&codec, frame, frame_size, NULL);
+ if (aom_status) die_codec(&codec, "Failed to decode tile.");
+
+ aom_codec_control_(&codec, AV1D_GET_FRAME_HEADER_INFO, &frame_header_info);
+
+ size_t obu_size_offset =
+ (uint8_t *)frame_header_info.coded_tile_data - frame;
+ size_t length_field_size = frame_header_info.coded_tile_data_size;
+ // Remove ext-tile tile info.
+ uint32_t frame_header_size = (uint32_t)frame_header_info.extra_size - 1;
+ size_t bytes_to_copy =
+ obu_size_offset + length_field_size + frame_header_size;
+
+ unsigned char *frame_hdr_buf = (unsigned char *)malloc(bytes_to_copy);
+ if (frame_hdr_buf == NULL)
+ die_codec(&codec, "Failed to allocate frame header buffer.");
+
+ memcpy(frame_hdr_buf, frame, bytes_to_copy);
+
+ // Update frame header OBU size.
+ size_t bytes_written = 0;
+ if (aom_uleb_encode_fixed_size(
+ frame_header_size, length_field_size, length_field_size,
+ frame_hdr_buf + obu_size_offset, &bytes_written))
+ die_codec(&codec, "Failed to encode the tile list obu size.");
+
+ // Copy camera frame header bitstream.
+ if (!aom_video_writer_write_frame(writer, frame_hdr_buf, bytes_to_copy,
+ pts))
+ die_codec(&codec, "Failed to copy compressed camera frame header.");
+ free(frame_hdr_buf);
+ }
+
+ // Read out the image format.
+ aom_img_fmt_t ref_fmt = 0;
+ if (aom_codec_control(&codec, AV1D_GET_IMG_FORMAT, &ref_fmt))
+ die_codec(&codec, "Failed to get the image format");
+ const int bps = get_image_bps(ref_fmt);
+ if (!bps) die_codec(&codec, "Invalid image format.");
+ // read out the tile size.
+ unsigned int tile_size = 0;
+ if (aom_codec_control(&codec, AV1D_GET_TILE_SIZE, &tile_size))
+ die_codec(&codec, "Failed to get the tile size");
+ const unsigned int tile_width = tile_size >> 16;
+ const unsigned int tile_height = tile_size & 65535;
+ // Allocate a buffer to store tile list bitstream.
+ const size_t data_sz = MAX_TILES * ALIGN_POWER_OF_TWO(tile_width, 5) *
+ ALIGN_POWER_OF_TWO(tile_height, 5) * bps / 8;
+ unsigned char *tl_buf = (unsigned char *)malloc(data_sz);
+ if (tl_buf == NULL) die_codec(&codec, "Failed to allocate tile list buffer.");
+
+ aom_codec_pts_t tl_pts = pts;
+
+ // Process 1 tile list.
+ for (n = 0; n < num_tile_lists; n++) {
+ unsigned char *tl = tl_buf;
+ struct aom_write_bit_buffer wb = { tl, 0 };
+ unsigned char *saved_obu_size_loc = NULL;
+ uint32_t tile_list_obu_header_size = 0;
+ uint32_t tile_list_obu_size = 0;
+
+ // Write the tile list OBU header that is 1 byte long.
+ aom_wb_write_literal(&wb, 0, 1); // forbidden bit.
+ aom_wb_write_literal(&wb, 8, 4); // tile list OBU: "1000"
+ aom_wb_write_literal(&wb, 0, 1); // obu_extension = 0
+ aom_wb_write_literal(&wb, 1, 1); // obu_has_size_field
+ aom_wb_write_literal(&wb, 0, 1); // reserved
+ tl++;
+ tile_list_obu_header_size++;
+
+ // Write the OBU size using a fixed length_field_size of 4 bytes.
+ saved_obu_size_loc = tl;
+ // aom_wb_write_unsigned_literal(&wb, data, bits) requires that bits <= 32.
+ aom_wb_write_unsigned_literal(&wb, 0, 32);
+ tl += 4;
+ tile_list_obu_header_size += 4;
+
+ // write_tile_list_obu()
+ aom_wb_write_literal(&wb, output_frame_width_in_tiles_minus_1, 8);
+ aom_wb_write_literal(&wb, output_frame_height_in_tiles_minus_1, 8);
+ aom_wb_write_literal(&wb, tile_count_minus_1, 16);
+ tl += 4;
+ tile_list_obu_size += 4;
+
+ // Write each tile's data
+ for (i = 0; i <= tile_count_minus_1; i++) {
+ aom_tile_data tile_data = { 0, NULL, 0 };
+
+ int image_idx = tile_list[n][i].image_idx;
+ int ref_idx = tile_list[n][i].reference_idx;
+ int tc = tile_list[n][i].tile_col;
+ int tr = tile_list[n][i].tile_row;
+ int frame_cnt = -1;
+
+ // Reset bit writer to the right location.
+ wb.bit_buffer = tl;
+ wb.bit_offset = 0;
+
+ // Seek to the first camera image.
+ fseeko(infile, camera_frame_pos, SEEK_SET);
+
+ // Read out the camera image
+ while (frame_cnt != image_idx) {
+ aom_video_reader_read_frame(reader);
+ frame_cnt++;
+ }
+
+ size_t frame_size = 0;
+ const unsigned char *frame =
+ aom_video_reader_get_frame(reader, &frame_size);
+
+ aom_codec_control_(&codec, AV1_SET_DECODE_TILE_ROW, tr);
+ aom_codec_control_(&codec, AV1_SET_DECODE_TILE_COL, tc);
+
+ aom_codec_err_t aom_status =
+ aom_codec_decode(&codec, frame, frame_size, NULL);
+ if (aom_status) die_codec(&codec, "Failed to decode tile.");
+
+ aom_codec_control_(&codec, AV1D_GET_TILE_DATA, &tile_data);
+
+ // Copy over tile info.
+ // uint8_t anchor_frame_idx;
+ // uint8_t tile_row;
+ // uint8_t tile_col;
+ // uint16_t coded_tile_data_size_minus_1;
+ // uint8_t *coded_tile_data;
+ uint32_t tile_info_bytes = 5;
+ aom_wb_write_literal(&wb, ref_idx, 8);
+ aom_wb_write_literal(&wb, tr, 8);
+ aom_wb_write_literal(&wb, tc, 8);
+ aom_wb_write_literal(&wb, (int)tile_data.coded_tile_data_size - 1, 16);
+ tl += tile_info_bytes;
+
+ memcpy(tl, (uint8_t *)tile_data.coded_tile_data,
+ tile_data.coded_tile_data_size);
+ tl += tile_data.coded_tile_data_size;
+
+ tile_list_obu_size +=
+ tile_info_bytes + (uint32_t)tile_data.coded_tile_data_size;
+ }
+
+ // Write tile list OBU size.
+ size_t bytes_written = 0;
+ if (aom_uleb_encode_fixed_size(tile_list_obu_size, 4, 4, saved_obu_size_loc,
+ &bytes_written))
+ die_codec(&codec, "Failed to encode the tile list obu size.");
+
+ // Copy the tile list.
+ if (!aom_video_writer_write_frame(
+ writer, tl_buf, tile_list_obu_header_size + tile_list_obu_size,
+ tl_pts))
+ die_codec(&codec, "Failed to copy compressed tile list.");
+
+ tl_pts++;
+ }
+
+ free(tl_buf);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
+ aom_video_writer_close(writer);
+ aom_video_reader_close(reader);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/lightfield_decoder.c b/media/libaom/src/examples/lightfield_decoder.c
new file mode 100644
index 000000000..f5e54db7f
--- /dev/null
+++ b/media/libaom/src/examples/lightfield_decoder.c
@@ -0,0 +1,208 @@
+/*
+ * Copyright (c) 2017, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Lightfield Decoder
+// ==================
+//
+// This is an example of a simple lightfield decoder. It builds upon the
+// simple_decoder.c example. It takes an input file containing the compressed
+// data (in ivf format), treating it as a lightfield instead of a video.
+// After running the lightfield encoder, run lightfield decoder to decode a
+// batch of tiles:
+// examples/lightfield_decoder vase10x10.ivf vase_reference.yuv 4
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aomdx.h"
+#include "aom_scale/yv12config.h"
+#include "av1/common/enums.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s <infile> <outfile> <num_references>\n", exec_name);
+ exit(EXIT_FAILURE);
+}
+
+// Tile list entry provided by the application
+typedef struct {
+ int image_idx;
+ int reference_idx;
+ int tile_col;
+ int tile_row;
+} TILE_LIST_INFO;
+
+// M references: 0 - M-1; N images(including references): 0 - N-1;
+// Note: order the image index incrementally, so that we only go through the
+// bitstream once to construct the tile list.
+const int num_tile_lists = 2;
+const uint16_t tile_count_minus_1 = 9 - 1;
+const TILE_LIST_INFO tile_list[2][9] = {
+ { { 16, 0, 4, 5 },
+ { 83, 3, 13, 2 },
+ { 57, 2, 2, 6 },
+ { 31, 1, 11, 5 },
+ { 2, 0, 7, 4 },
+ { 77, 3, 9, 9 },
+ { 49, 1, 0, 1 },
+ { 6, 0, 3, 10 },
+ { 63, 2, 5, 8 } },
+ { { 65, 2, 11, 1 },
+ { 42, 1, 3, 7 },
+ { 88, 3, 8, 4 },
+ { 76, 3, 1, 15 },
+ { 1, 0, 2, 2 },
+ { 19, 0, 5, 6 },
+ { 60, 2, 4, 0 },
+ { 25, 1, 11, 15 },
+ { 50, 2, 5, 4 } },
+};
+
+int main(int argc, char **argv) {
+ FILE *outfile = NULL;
+ aom_codec_ctx_t codec;
+ AvxVideoReader *reader = NULL;
+ const AvxInterface *decoder = NULL;
+ const AvxVideoInfo *info = NULL;
+ int num_references;
+ aom_image_t reference_images[MAX_EXTERNAL_REFERENCES];
+ size_t frame_size = 0;
+ const unsigned char *frame = NULL;
+ int n, i, j;
+ exec_name = argv[0];
+
+ if (argc != 4) die("Invalid number of arguments.");
+
+ reader = aom_video_reader_open(argv[1]);
+ if (!reader) die("Failed to open %s for reading.", argv[1]);
+
+ if (!(outfile = fopen(argv[2], "wb")))
+ die("Failed to open %s for writing.", argv[2]);
+
+ num_references = (int)strtol(argv[3], NULL, 0);
+
+ info = aom_video_reader_get_info(reader);
+
+ decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) die("Unknown input codec.");
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder.");
+
+ if (aom_codec_control(&codec, AV1D_SET_IS_ANNEXB, info->is_annexb)) {
+ die("Failed to set annex b status");
+ }
+
+ // Decode anchor frames.
+ aom_codec_control_(&codec, AV1_SET_TILE_MODE, 0);
+ for (i = 0; i < num_references; ++i) {
+ aom_video_reader_read_frame(reader);
+ frame = aom_video_reader_get_frame(reader, &frame_size);
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode frame.");
+
+ if (i == 0) {
+ aom_img_fmt_t ref_fmt = 0;
+ if (aom_codec_control(&codec, AV1D_GET_IMG_FORMAT, &ref_fmt))
+ die_codec(&codec, "Failed to get the image format");
+
+ int frame_res[2];
+ if (aom_codec_control(&codec, AV1D_GET_FRAME_SIZE, frame_res))
+ die_codec(&codec, "Failed to get the image frame size");
+
+ // Allocate memory to store decoded references. Allocate memory with the
+ // border so that it can be used as a reference.
+ for (j = 0; j < num_references; j++) {
+ unsigned int border = AOM_BORDER_IN_PIXELS;
+ if (!aom_img_alloc_with_border(&reference_images[j], ref_fmt,
+ frame_res[0], frame_res[1], 32, 8,
+ border)) {
+ die("Failed to allocate references.");
+ }
+ }
+ }
+
+ if (aom_codec_control(&codec, AV1_COPY_NEW_FRAME_IMAGE,
+ &reference_images[i]))
+ die_codec(&codec, "Failed to copy decoded reference frame");
+
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img = NULL;
+ while ((img = aom_codec_get_frame(&codec, &iter)) != NULL) {
+ char name[1024];
+ snprintf(name, sizeof(name), "ref_%d.yuv", i);
+ printf("writing ref image to %s, %d, %d\n", name, img->d_w, img->d_h);
+ FILE *ref_file = fopen(name, "wb");
+ aom_img_write(img, ref_file);
+ fclose(ref_file);
+ }
+ }
+
+ FILE *infile = aom_video_reader_get_file(reader);
+ // Record the offset of the first camera image.
+ const FileOffset camera_frame_pos = ftello(infile);
+
+ // Process 1 tile.
+ for (n = 0; n < num_tile_lists; n++) {
+ for (i = 0; i <= tile_count_minus_1; i++) {
+ int image_idx = tile_list[n][i].image_idx;
+ int ref_idx = tile_list[n][i].reference_idx;
+ int tc = tile_list[n][i].tile_col;
+ int tr = tile_list[n][i].tile_row;
+ int frame_cnt = -1;
+
+ // Seek to the first camera image.
+ fseeko(infile, camera_frame_pos, SEEK_SET);
+
+ // Read out the camera image
+ while (frame_cnt != image_idx) {
+ aom_video_reader_read_frame(reader);
+ frame_cnt++;
+ }
+
+ frame = aom_video_reader_get_frame(reader, &frame_size);
+
+ aom_codec_control_(&codec, AV1_SET_TILE_MODE, 1);
+ aom_codec_control_(&codec, AV1D_EXT_TILE_DEBUG, 1);
+ aom_codec_control_(&codec, AV1_SET_DECODE_TILE_ROW, tr);
+ aom_codec_control_(&codec, AV1_SET_DECODE_TILE_COL, tc);
+
+ av1_ref_frame_t ref;
+ ref.idx = 0;
+ ref.use_external_ref = 1;
+ ref.img = reference_images[ref_idx];
+ if (aom_codec_control(&codec, AV1_SET_REFERENCE, &ref)) {
+ die_codec(&codec, "Failed to set reference frame.");
+ }
+
+ aom_codec_err_t aom_status =
+ aom_codec_decode(&codec, frame, frame_size, NULL);
+ if (aom_status) die_codec(&codec, "Failed to decode tile.");
+
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img = aom_codec_get_frame(&codec, &iter);
+ aom_img_write(img, outfile);
+ }
+ }
+
+ for (i = 0; i < num_references; i++) aom_img_free(&reference_images[i]);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
+ aom_video_reader_close(reader);
+ fclose(outfile);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/lightfield_encoder.c b/media/libaom/src/examples/lightfield_encoder.c
new file mode 100644
index 000000000..e55cd5ce3
--- /dev/null
+++ b/media/libaom/src/examples/lightfield_encoder.c
@@ -0,0 +1,499 @@
+/*
+ * Copyright (c) 2017, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Lightfield Encoder
+// ==================
+//
+// This is an example of a simple lightfield encoder. It builds upon the
+// twopass_encoder.c example. It takes an input file in YV12 format,
+// treating it as a planar lightfield instead of a video. The img_width
+// and img_height arguments are the dimensions of the lightfield images,
+// while the lf_width and lf_height arguments are the number of
+// lightfield images in each dimension. The lf_blocksize determines the
+// number of reference images used for MCP. For example, 5 means that there
+// is a reference image for every 5x5 lightfield image block. All images
+// within a block will use the center image in that block as the reference
+// image for MCP.
+// Run "make test" to download lightfield test data: vase10x10.yuv.
+// Run lightfield encoder to encode whole lightfield:
+// examples/lightfield_encoder 1024 1024 vase10x10.yuv vase10x10.ivf 10 10 5
+
+// Note: In bitstream.c and encoder.c, define EXT_TILE_DEBUG as 1 will print
+// out the uncompressed header and the frame contexts, which can be used to
+// test the bit exactness of the headers and the frame contexts for large scale
+// tile coded frames.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_encoder.h"
+#include "aom/aomcx.h"
+#include "aom_scale/yv12config.h"
+#include "av1/common/enums.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr,
+ "Usage: %s <img_width> <img_height> <infile> <outfile> "
+ "<lf_width> <lf_height> <lf_blocksize>\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static int aom_img_size_bytes(aom_image_t *img) {
+ int image_size_bytes = 0;
+ int plane;
+ for (plane = 0; plane < 3; ++plane) {
+ const int w = aom_img_plane_width(img, plane) *
+ ((img->fmt & AOM_IMG_FMT_HIGHBITDEPTH) ? 2 : 1);
+ const int h = aom_img_plane_height(img, plane);
+ image_size_bytes += w * h;
+ }
+ return image_size_bytes;
+}
+
+static int get_frame_stats(aom_codec_ctx_t *ctx, const aom_image_t *img,
+ aom_codec_pts_t pts, unsigned int duration,
+ aom_enc_frame_flags_t flags,
+ aom_fixed_buf_t *stats) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res = aom_codec_encode(ctx, img, pts, duration, flags);
+ if (res != AOM_CODEC_OK) die_codec(ctx, "Failed to get frame stats.");
+
+ while ((pkt = aom_codec_get_cx_data(ctx, &iter)) != NULL) {
+ got_pkts = 1;
+
+ if (pkt->kind == AOM_CODEC_STATS_PKT) {
+ const uint8_t *const pkt_buf = pkt->data.twopass_stats.buf;
+ const size_t pkt_size = pkt->data.twopass_stats.sz;
+ stats->buf = realloc(stats->buf, stats->sz + pkt_size);
+ memcpy((uint8_t *)stats->buf + stats->sz, pkt_buf, pkt_size);
+ stats->sz += pkt_size;
+ }
+ }
+
+ return got_pkts;
+}
+
+static int encode_frame(aom_codec_ctx_t *ctx, const aom_image_t *img,
+ aom_codec_pts_t pts, unsigned int duration,
+ aom_enc_frame_flags_t flags, AvxVideoWriter *writer) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res = aom_codec_encode(ctx, img, pts, duration, flags);
+ if (res != AOM_CODEC_OK) die_codec(ctx, "Failed to encode frame.");
+
+ while ((pkt = aom_codec_get_cx_data(ctx, &iter)) != NULL) {
+ got_pkts = 1;
+ if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
+ const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
+
+ if (!aom_video_writer_write_frame(writer, pkt->data.frame.buf,
+ pkt->data.frame.sz,
+ pkt->data.frame.pts))
+ die_codec(ctx, "Failed to write compressed frame.");
+ printf(keyframe ? "K" : ".");
+ fflush(stdout);
+ }
+ }
+
+ return got_pkts;
+}
+
+static void get_raw_image(aom_image_t **frame_to_encode, aom_image_t *raw,
+ aom_image_t *raw_shift) {
+ if (!CONFIG_LOWBITDEPTH) {
+ // Need to allocate larger buffer to use hbd internal.
+ int input_shift = 0;
+ aom_img_upshift(raw_shift, raw, input_shift);
+ *frame_to_encode = raw_shift;
+ } else {
+ *frame_to_encode = raw;
+ }
+}
+
+static aom_fixed_buf_t pass0(aom_image_t *raw, FILE *infile,
+ const AvxInterface *encoder,
+ const aom_codec_enc_cfg_t *cfg, int lf_width,
+ int lf_height, int lf_blocksize, int flags,
+ aom_image_t *raw_shift) {
+ aom_codec_ctx_t codec;
+ int frame_count = 0;
+ int image_size_bytes = aom_img_size_bytes(raw);
+ int u_blocks, v_blocks;
+ int bu, bv;
+ aom_fixed_buf_t stats = { NULL, 0 };
+ aom_image_t *frame_to_encode;
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), cfg, flags))
+ die_codec(&codec, "Failed to initialize encoder");
+ if (aom_codec_control(&codec, AOME_SET_ENABLEAUTOALTREF, 0))
+ die_codec(&codec, "Failed to turn off auto altref");
+ if (aom_codec_control(&codec, AV1E_SET_FRAME_PARALLEL_DECODING, 0))
+ die_codec(&codec, "Failed to set frame parallel decoding");
+
+ // How many reference images we need to encode.
+ u_blocks = (lf_width + lf_blocksize - 1) / lf_blocksize;
+ v_blocks = (lf_height + lf_blocksize - 1) / lf_blocksize;
+
+ printf("\n First pass: ");
+
+ for (bv = 0; bv < v_blocks; ++bv) {
+ for (bu = 0; bu < u_blocks; ++bu) {
+ const int block_u_min = bu * lf_blocksize;
+ const int block_v_min = bv * lf_blocksize;
+ int block_u_end = (bu + 1) * lf_blocksize;
+ int block_v_end = (bv + 1) * lf_blocksize;
+ int u_block_size, v_block_size;
+ int block_ref_u, block_ref_v;
+
+ block_u_end = block_u_end < lf_width ? block_u_end : lf_width;
+ block_v_end = block_v_end < lf_height ? block_v_end : lf_height;
+ u_block_size = block_u_end - block_u_min;
+ v_block_size = block_v_end - block_v_min;
+ block_ref_u = block_u_min + u_block_size / 2;
+ block_ref_v = block_v_min + v_block_size / 2;
+
+ printf("A%d, ", (block_ref_u + block_ref_v * lf_width));
+ fseek(infile, (block_ref_u + block_ref_v * lf_width) * image_size_bytes,
+ SEEK_SET);
+ aom_img_read(raw, infile);
+ get_raw_image(&frame_to_encode, raw, raw_shift);
+
+ // Reference frames can be encoded encoded without tiles.
+ ++frame_count;
+ get_frame_stats(&codec, frame_to_encode, frame_count, 1,
+ AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 |
+ AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF |
+ AOM_EFLAG_NO_REF_BWD | AOM_EFLAG_NO_REF_ARF2 |
+ AOM_EFLAG_NO_UPD_LAST | AOM_EFLAG_NO_UPD_GF |
+ AOM_EFLAG_NO_UPD_ARF,
+ &stats);
+ }
+ }
+
+ if (aom_codec_control(&codec, AV1E_SET_FRAME_PARALLEL_DECODING, 1))
+ die_codec(&codec, "Failed to set frame parallel decoding");
+
+ for (bv = 0; bv < v_blocks; ++bv) {
+ for (bu = 0; bu < u_blocks; ++bu) {
+ const int block_u_min = bu * lf_blocksize;
+ const int block_v_min = bv * lf_blocksize;
+ int block_u_end = (bu + 1) * lf_blocksize;
+ int block_v_end = (bv + 1) * lf_blocksize;
+ int u, v;
+ block_u_end = block_u_end < lf_width ? block_u_end : lf_width;
+ block_v_end = block_v_end < lf_height ? block_v_end : lf_height;
+ for (v = block_v_min; v < block_v_end; ++v) {
+ for (u = block_u_min; u < block_u_end; ++u) {
+ printf("C%d, ", (u + v * lf_width));
+ fseek(infile, (u + v * lf_width) * image_size_bytes, SEEK_SET);
+ aom_img_read(raw, infile);
+ get_raw_image(&frame_to_encode, raw, raw_shift);
+
+ ++frame_count;
+ get_frame_stats(&codec, frame_to_encode, frame_count, 1,
+ AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 |
+ AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF |
+ AOM_EFLAG_NO_REF_BWD | AOM_EFLAG_NO_REF_ARF2 |
+ AOM_EFLAG_NO_UPD_LAST | AOM_EFLAG_NO_UPD_GF |
+ AOM_EFLAG_NO_UPD_ARF | AOM_EFLAG_NO_UPD_ENTROPY,
+ &stats);
+ }
+ }
+ }
+ }
+ // Flush encoder.
+ // No ARF, this should not be needed.
+ while (get_frame_stats(&codec, NULL, frame_count, 1, 0, &stats)) {
+ }
+
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ printf("\nFirst pass complete. Processed %d frames.\n", frame_count);
+
+ return stats;
+}
+
+static void pass1(aom_image_t *raw, FILE *infile, const char *outfile_name,
+ const AvxInterface *encoder, aom_codec_enc_cfg_t *cfg,
+ int lf_width, int lf_height, int lf_blocksize, int flags,
+ aom_image_t *raw_shift) {
+ AvxVideoInfo info = { encoder->fourcc,
+ cfg->g_w,
+ cfg->g_h,
+ { cfg->g_timebase.num, cfg->g_timebase.den },
+ 0 };
+ AvxVideoWriter *writer = NULL;
+ aom_codec_ctx_t codec;
+ int frame_count = 0;
+ int image_size_bytes = aom_img_size_bytes(raw);
+ int bu, bv;
+ int u_blocks, v_blocks;
+ aom_image_t *frame_to_encode;
+ aom_image_t reference_images[MAX_EXTERNAL_REFERENCES];
+ int reference_image_num = 0;
+ int i;
+
+ writer = aom_video_writer_open(outfile_name, kContainerIVF, &info);
+ if (!writer) die("Failed to open %s for writing", outfile_name);
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), cfg, flags))
+ die_codec(&codec, "Failed to initialize encoder");
+ if (aom_codec_control(&codec, AOME_SET_ENABLEAUTOALTREF, 0))
+ die_codec(&codec, "Failed to turn off auto altref");
+ if (aom_codec_control(&codec, AV1E_SET_FRAME_PARALLEL_DECODING, 0))
+ die_codec(&codec, "Failed to set frame parallel decoding");
+ // Note: The superblock is a sequence parameter and has to be the same for 1
+ // sequence. In lightfield application, must choose the superblock size(either
+ // 64x64 or 128x128) before the encoding starts. Otherwise, the default is
+ // AOM_SUPERBLOCK_SIZE_DYNAMIC, and the superblock size will be set to 64x64
+ // internally.
+ if (aom_codec_control(&codec, AV1E_SET_SUPERBLOCK_SIZE,
+ AOM_SUPERBLOCK_SIZE_64X64))
+ die_codec(&codec, "Failed to set SB size");
+
+ u_blocks = (lf_width + lf_blocksize - 1) / lf_blocksize;
+ v_blocks = (lf_height + lf_blocksize - 1) / lf_blocksize;
+
+ reference_image_num = u_blocks * v_blocks;
+ aom_img_fmt_t ref_fmt = AOM_IMG_FMT_I420;
+ if (!CONFIG_LOWBITDEPTH) ref_fmt |= AOM_IMG_FMT_HIGHBITDEPTH;
+ // Allocate memory with the border so that it can be used as a reference.
+ for (i = 0; i < reference_image_num; i++) {
+ if (!aom_img_alloc_with_border(&reference_images[i], ref_fmt, cfg->g_w,
+ cfg->g_h, 32, 8, AOM_BORDER_IN_PIXELS)) {
+ die("Failed to allocate image.");
+ }
+ }
+
+ printf("\n Second pass: ");
+
+ // Encode reference images first.
+ printf("Encoding Reference Images\n");
+ for (bv = 0; bv < v_blocks; ++bv) {
+ for (bu = 0; bu < u_blocks; ++bu) {
+ const int block_u_min = bu * lf_blocksize;
+ const int block_v_min = bv * lf_blocksize;
+ int block_u_end = (bu + 1) * lf_blocksize;
+ int block_v_end = (bv + 1) * lf_blocksize;
+ int u_block_size, v_block_size;
+ int block_ref_u, block_ref_v;
+
+ block_u_end = block_u_end < lf_width ? block_u_end : lf_width;
+ block_v_end = block_v_end < lf_height ? block_v_end : lf_height;
+ u_block_size = block_u_end - block_u_min;
+ v_block_size = block_v_end - block_v_min;
+ block_ref_u = block_u_min + u_block_size / 2;
+ block_ref_v = block_v_min + v_block_size / 2;
+
+ printf("A%d, ", (block_ref_u + block_ref_v * lf_width));
+ fseek(infile, (block_ref_u + block_ref_v * lf_width) * image_size_bytes,
+ SEEK_SET);
+ aom_img_read(raw, infile);
+
+ get_raw_image(&frame_to_encode, raw, raw_shift);
+
+ // Reference frames may be encoded without tiles.
+ ++frame_count;
+ printf("Encoding reference image %d of %d\n", bv * u_blocks + bu,
+ u_blocks * v_blocks);
+ encode_frame(&codec, frame_to_encode, frame_count, 1,
+ AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 |
+ AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF |
+ AOM_EFLAG_NO_REF_BWD | AOM_EFLAG_NO_REF_ARF2 |
+ AOM_EFLAG_NO_UPD_LAST | AOM_EFLAG_NO_UPD_GF |
+ AOM_EFLAG_NO_UPD_ARF | AOM_EFLAG_NO_UPD_ENTROPY,
+ writer);
+
+ if (aom_codec_control(&codec, AV1_COPY_NEW_FRAME_IMAGE,
+ &reference_images[frame_count - 1]))
+ die_codec(&codec, "Failed to copy decoder reference frame");
+ }
+ }
+
+ cfg->large_scale_tile = 1;
+ // Fixed q encoding for camera frames.
+ cfg->rc_end_usage = AOM_Q;
+ if (aom_codec_enc_config_set(&codec, cfg))
+ die_codec(&codec, "Failed to configure encoder");
+
+ // The fixed q value used in encoding.
+ if (aom_codec_control(&codec, AOME_SET_CQ_LEVEL, 36))
+ die_codec(&codec, "Failed to set cq level");
+ if (aom_codec_control(&codec, AV1E_SET_FRAME_PARALLEL_DECODING, 1))
+ die_codec(&codec, "Failed to set frame parallel decoding");
+ if (aom_codec_control(&codec, AV1E_SET_SINGLE_TILE_DECODING, 1))
+ die_codec(&codec, "Failed to turn on single tile decoding");
+ // Set tile_columns and tile_rows to MAX values, which guarantees the tile
+ // size of 64 x 64 pixels(i.e. 1 SB) for <= 4k resolution.
+ if (aom_codec_control(&codec, AV1E_SET_TILE_COLUMNS, 6))
+ die_codec(&codec, "Failed to set tile width");
+ if (aom_codec_control(&codec, AV1E_SET_TILE_ROWS, 6))
+ die_codec(&codec, "Failed to set tile height");
+
+ for (bv = 0; bv < v_blocks; ++bv) {
+ for (bu = 0; bu < u_blocks; ++bu) {
+ const int block_u_min = bu * lf_blocksize;
+ const int block_v_min = bv * lf_blocksize;
+ int block_u_end = (bu + 1) * lf_blocksize;
+ int block_v_end = (bv + 1) * lf_blocksize;
+ int u, v;
+ block_u_end = block_u_end < lf_width ? block_u_end : lf_width;
+ block_v_end = block_v_end < lf_height ? block_v_end : lf_height;
+ for (v = block_v_min; v < block_v_end; ++v) {
+ for (u = block_u_min; u < block_u_end; ++u) {
+ av1_ref_frame_t ref;
+ ref.idx = 0;
+ ref.use_external_ref = 1;
+ ref.img = reference_images[bv * u_blocks + bu];
+ if (aom_codec_control(&codec, AV1_SET_REFERENCE, &ref))
+ die_codec(&codec, "Failed to set reference frame");
+
+ printf("C%d, ", (u + v * lf_width));
+ fseek(infile, (u + v * lf_width) * image_size_bytes, SEEK_SET);
+ aom_img_read(raw, infile);
+ get_raw_image(&frame_to_encode, raw, raw_shift);
+
+ ++frame_count;
+ printf("Encoding image %d of %d\n",
+ frame_count - (u_blocks * v_blocks), lf_width * lf_height);
+ encode_frame(&codec, frame_to_encode, frame_count, 1,
+ AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 |
+ AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF |
+ AOM_EFLAG_NO_REF_BWD | AOM_EFLAG_NO_REF_ARF2 |
+ AOM_EFLAG_NO_UPD_LAST | AOM_EFLAG_NO_UPD_GF |
+ AOM_EFLAG_NO_UPD_ARF | AOM_EFLAG_NO_UPD_ENTROPY,
+ writer);
+ }
+ }
+ }
+ }
+
+ // Flush encoder.
+ // No ARF, this should not be needed.
+ while (encode_frame(&codec, NULL, -1, 1, 0, writer)) {
+ }
+
+ for (i = 0; i < reference_image_num; i++) aom_img_free(&reference_images[i]);
+
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+ aom_video_writer_close(writer);
+
+ printf("\nSecond pass complete. Processed %d frames.\n", frame_count);
+}
+
+int main(int argc, char **argv) {
+ FILE *infile = NULL;
+ int w, h;
+ // The number of lightfield images in the u and v dimensions.
+ int lf_width, lf_height;
+ // Defines how many images refer to the same reference image for MCP.
+ // lf_blocksize X lf_blocksize images will all use the reference image
+ // in the middle of the block of images.
+ int lf_blocksize;
+ aom_codec_ctx_t codec;
+ aom_codec_enc_cfg_t cfg;
+ aom_image_t raw;
+ aom_image_t raw_shift;
+ aom_codec_err_t res;
+ aom_fixed_buf_t stats;
+ int flags = 0;
+
+ const AvxInterface *encoder = NULL;
+ const int fps = 30;
+ const int bitrate = 200; // kbit/s
+ const char *const width_arg = argv[1];
+ const char *const height_arg = argv[2];
+ const char *const infile_arg = argv[3];
+ const char *const outfile_arg = argv[4];
+ const char *const lf_width_arg = argv[5];
+ const char *const lf_height_arg = argv[6];
+ const char *lf_blocksize_arg = argv[7];
+ exec_name = argv[0];
+
+ if (argc < 8) die("Invalid number of arguments");
+
+ encoder = get_aom_encoder_by_name("av1");
+ if (!encoder) die("Unsupported codec.");
+
+ w = (int)strtol(width_arg, NULL, 0);
+ h = (int)strtol(height_arg, NULL, 0);
+ lf_width = (int)strtol(lf_width_arg, NULL, 0);
+ lf_height = (int)strtol(lf_height_arg, NULL, 0);
+ lf_blocksize = (int)strtol(lf_blocksize_arg, NULL, 0);
+ lf_blocksize = lf_blocksize < lf_width ? lf_blocksize : lf_width;
+ lf_blocksize = lf_blocksize < lf_height ? lf_blocksize : lf_height;
+
+ if (w <= 0 || h <= 0 || (w % 2) != 0 || (h % 2) != 0)
+ die("Invalid frame size: %dx%d", w, h);
+ if (lf_width <= 0 || lf_height <= 0)
+ die("Invalid lf_width and/or lf_height: %dx%d", lf_width, lf_height);
+ if (lf_blocksize <= 0) die("Invalid lf_blocksize: %d", lf_blocksize);
+
+ if (!aom_img_alloc(&raw, AOM_IMG_FMT_I420, w, h, 32)) {
+ die("Failed to allocate image.");
+ }
+ if (!CONFIG_LOWBITDEPTH) {
+ // Need to allocate larger buffer to use hbd internal.
+ aom_img_alloc(&raw_shift, AOM_IMG_FMT_I420 | AOM_IMG_FMT_HIGHBITDEPTH, w, h,
+ 32);
+ }
+
+ printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
+
+ // Configuration
+ res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
+ if (res) die_codec(&codec, "Failed to get default codec config.");
+
+ cfg.g_w = w;
+ cfg.g_h = h;
+ cfg.g_timebase.num = 1;
+ cfg.g_timebase.den = fps;
+ cfg.rc_target_bitrate = bitrate;
+ cfg.g_error_resilient = 0; // This is required.
+ cfg.g_lag_in_frames = 0; // need to set this since default is 19.
+ cfg.kf_mode = AOM_KF_DISABLED;
+ cfg.large_scale_tile = 0; // Only set it to 1 for camera frame encoding.
+ cfg.g_bit_depth = AOM_BITS_8;
+ flags |= (cfg.g_bit_depth > AOM_BITS_8 || !CONFIG_LOWBITDEPTH)
+ ? AOM_CODEC_USE_HIGHBITDEPTH
+ : 0;
+
+ if (!(infile = fopen(infile_arg, "rb")))
+ die("Failed to open %s for reading", infile_arg);
+
+ // Pass 0
+ cfg.g_pass = AOM_RC_FIRST_PASS;
+ stats = pass0(&raw, infile, encoder, &cfg, lf_width, lf_height, lf_blocksize,
+ flags, &raw_shift);
+
+ // Pass 1
+ rewind(infile);
+ cfg.g_pass = AOM_RC_LAST_PASS;
+ cfg.rc_twopass_stats_in = stats;
+ pass1(&raw, infile, outfile_arg, encoder, &cfg, lf_width, lf_height,
+ lf_blocksize, flags, &raw_shift);
+ free(stats.buf);
+
+ if (!CONFIG_LOWBITDEPTH) aom_img_free(&raw_shift);
+ aom_img_free(&raw);
+ fclose(infile);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/lightfield_tile_list_decoder.c b/media/libaom/src/examples/lightfield_tile_list_decoder.c
new file mode 100644
index 000000000..5556bf0e7
--- /dev/null
+++ b/media/libaom/src/examples/lightfield_tile_list_decoder.c
@@ -0,0 +1,161 @@
+/*
+ * Copyright (c) 2018, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Lightfield Tile List Decoder
+// ============================
+//
+// This is a lightfield tile list decoder example. It takes an input file that
+// contains the anchor frames that are references of the coded tiles, the camera
+// frame header, and tile list OBUs that include the tile information and the
+// compressed tile data. This input file is reconstructed from the encoded
+// lightfield ivf file, and is decodable by AV1 decoder. num_references is
+// the number of anchor frames coded at the beginning of the light field file.
+// num_tile_lists is the number of tile lists need to be decoded.
+// Run lightfield tile list decoder to decode an AV1 tile list file:
+// examples/lightfield_tile_list_decoder vase_tile_list.ivf vase_tile_list.yuv
+// 4 2
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aomdx.h"
+#include "aom_scale/yv12config.h"
+#include "av1/common/enums.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr,
+ "Usage: %s <infile> <outfile> <num_references> <num_tile_lists>\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+int main(int argc, char **argv) {
+ FILE *outfile = NULL;
+ aom_codec_ctx_t codec;
+ AvxVideoReader *reader = NULL;
+ const AvxInterface *decoder = NULL;
+ const AvxVideoInfo *info = NULL;
+ int num_references;
+ int num_tile_lists;
+ aom_image_t reference_images[MAX_EXTERNAL_REFERENCES];
+ size_t frame_size = 0;
+ const unsigned char *frame = NULL;
+ int i, j, n;
+
+ exec_name = argv[0];
+
+ if (argc != 5) die("Invalid number of arguments.");
+
+ reader = aom_video_reader_open(argv[1]);
+ if (!reader) die("Failed to open %s for reading.", argv[1]);
+
+ if (!(outfile = fopen(argv[2], "wb")))
+ die("Failed to open %s for writing.", argv[2]);
+
+ num_references = (int)strtol(argv[3], NULL, 0);
+ num_tile_lists = (int)strtol(argv[4], NULL, 0);
+
+ info = aom_video_reader_get_info(reader);
+
+ decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) die("Unknown input codec.");
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder.");
+
+ if (aom_codec_control(&codec, AV1D_SET_IS_ANNEXB, info->is_annexb)) {
+ die("Failed to set annex b status");
+ }
+
+ // Decode anchor frames.
+ aom_codec_control_(&codec, AV1_SET_TILE_MODE, 0);
+ for (i = 0; i < num_references; ++i) {
+ aom_video_reader_read_frame(reader);
+ frame = aom_video_reader_get_frame(reader, &frame_size);
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode frame.");
+
+ if (i == 0) {
+ aom_img_fmt_t ref_fmt = 0;
+ if (aom_codec_control(&codec, AV1D_GET_IMG_FORMAT, &ref_fmt))
+ die_codec(&codec, "Failed to get the image format");
+
+ int frame_res[2];
+ if (aom_codec_control(&codec, AV1D_GET_FRAME_SIZE, frame_res))
+ die_codec(&codec, "Failed to get the image frame size");
+
+ // Allocate memory to store decoded references. Allocate memory with the
+ // border so that it can be used as a reference.
+ for (j = 0; j < num_references; j++) {
+ unsigned int border = AOM_BORDER_IN_PIXELS;
+ if (!aom_img_alloc_with_border(&reference_images[j], ref_fmt,
+ frame_res[0], frame_res[1], 32, 8,
+ border)) {
+ die("Failed to allocate references.");
+ }
+ }
+ }
+
+ if (aom_codec_control(&codec, AV1_COPY_NEW_FRAME_IMAGE,
+ &reference_images[i]))
+ die_codec(&codec, "Failed to copy decoded reference frame");
+
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img = NULL;
+ while ((img = aom_codec_get_frame(&codec, &iter)) != NULL) {
+ char name[1024];
+ snprintf(name, sizeof(name), "ref_%d.yuv", i);
+ printf("writing ref image to %s, %d, %d\n", name, img->d_w, img->d_h);
+ FILE *ref_file = fopen(name, "wb");
+ aom_img_write(img, ref_file);
+ fclose(ref_file);
+ }
+ }
+
+ // Decode the lightfield.
+ aom_codec_control_(&codec, AV1_SET_TILE_MODE, 1);
+
+ // Set external references.
+ av1_ext_ref_frame_t set_ext_ref = { &reference_images[0], num_references };
+ aom_codec_control_(&codec, AV1D_SET_EXT_REF_PTR, &set_ext_ref);
+ // Must decode the camera frame header first.
+ aom_video_reader_read_frame(reader);
+ frame = aom_video_reader_get_frame(reader, &frame_size);
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode the frame.");
+ // Decode tile lists one by one.
+ for (n = 0; n < num_tile_lists; n++) {
+ aom_video_reader_read_frame(reader);
+ frame = aom_video_reader_get_frame(reader, &frame_size);
+
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode the tile list.");
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img;
+ while ((img = aom_codec_get_frame(&codec, &iter)))
+ fwrite(img->img_data, 1, img->sz, outfile);
+ }
+
+ for (i = 0; i < num_references; i++) aom_img_free(&reference_images[i]);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
+ aom_video_reader_close(reader);
+ fclose(outfile);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/lossless_encoder.c b/media/libaom/src/examples/lossless_encoder.c
new file mode 100644
index 000000000..438ff21c6
--- /dev/null
+++ b/media/libaom/src/examples/lossless_encoder.c
@@ -0,0 +1,138 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_encoder.h"
+#include "aom/aomcx.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr,
+ "lossless_encoder: Example demonstrating lossless "
+ "encoding feature. Supports raw input only.\n");
+ fprintf(stderr, "Usage: %s <width> <height> <infile> <outfile>\n", exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static int encode_frame(aom_codec_ctx_t *codec, aom_image_t *img,
+ int frame_index, int flags, AvxVideoWriter *writer) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res =
+ aom_codec_encode(codec, img, frame_index, 1, flags);
+ if (res != AOM_CODEC_OK) die_codec(codec, "Failed to encode frame");
+
+ while ((pkt = aom_codec_get_cx_data(codec, &iter)) != NULL) {
+ got_pkts = 1;
+
+ if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
+ const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
+ if (!aom_video_writer_write_frame(writer, pkt->data.frame.buf,
+ pkt->data.frame.sz,
+ pkt->data.frame.pts)) {
+ die_codec(codec, "Failed to write compressed frame");
+ }
+ printf(keyframe ? "K" : ".");
+ fflush(stdout);
+ }
+ }
+
+ return got_pkts;
+}
+
+int main(int argc, char **argv) {
+ FILE *infile = NULL;
+ aom_codec_ctx_t codec;
+ aom_codec_enc_cfg_t cfg;
+ int frame_count = 0;
+ aom_image_t raw;
+ aom_codec_err_t res;
+ AvxVideoInfo info;
+ AvxVideoWriter *writer = NULL;
+ const AvxInterface *encoder = NULL;
+ const int fps = 30;
+
+ exec_name = argv[0];
+
+ // Clear explicitly, as simply assigning "{ 0 }" generates
+ // "missing-field-initializers" warning in some compilers.
+ memset(&info, 0, sizeof(info));
+
+ if (argc < 5) die("Invalid number of arguments");
+
+ encoder = get_aom_encoder_by_name("av1");
+ if (!encoder) die("Unsupported codec.");
+
+ info.codec_fourcc = encoder->fourcc;
+ info.frame_width = (int)strtol(argv[1], NULL, 0);
+ info.frame_height = (int)strtol(argv[2], NULL, 0);
+ info.time_base.numerator = 1;
+ info.time_base.denominator = fps;
+
+ if (info.frame_width <= 0 || info.frame_height <= 0 ||
+ (info.frame_width % 2) != 0 || (info.frame_height % 2) != 0) {
+ die("Invalid frame size: %dx%d", info.frame_width, info.frame_height);
+ }
+
+ if (!aom_img_alloc(&raw, AOM_IMG_FMT_I420, info.frame_width,
+ info.frame_height, 1)) {
+ die("Failed to allocate image.");
+ }
+
+ printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
+
+ res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
+ if (res) die_codec(&codec, "Failed to get default codec config.");
+
+ cfg.g_w = info.frame_width;
+ cfg.g_h = info.frame_height;
+ cfg.g_timebase.num = info.time_base.numerator;
+ cfg.g_timebase.den = info.time_base.denominator;
+
+ writer = aom_video_writer_open(argv[4], kContainerIVF, &info);
+ if (!writer) die("Failed to open %s for writing.", argv[4]);
+
+ if (!(infile = fopen(argv[3], "rb")))
+ die("Failed to open %s for reading.", argv[3]);
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), &cfg, 0))
+ die_codec(&codec, "Failed to initialize encoder");
+
+ if (aom_codec_control_(&codec, AV1E_SET_LOSSLESS, 1))
+ die_codec(&codec, "Failed to use lossless mode");
+
+ // Encode frames.
+ while (aom_img_read(&raw, infile)) {
+ encode_frame(&codec, &raw, frame_count++, 0, writer);
+ }
+
+ // Flush encoder.
+ while (encode_frame(&codec, NULL, -1, 0, writer)) {
+ }
+
+ printf("\n");
+ fclose(infile);
+ printf("Processed %d frames.\n", frame_count);
+
+ aom_img_free(&raw);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ aom_video_writer_close(writer);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/noise_model.c b/media/libaom/src/examples/noise_model.c
new file mode 100644
index 000000000..5cc6003b6
--- /dev/null
+++ b/media/libaom/src/examples/noise_model.c
@@ -0,0 +1,431 @@
+/*
+ * Copyright (c) 2018, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+/*!\file
+ * \brief This is an sample binary to create noise params from input video.
+ *
+ * To allow for external denoising applications, this sample binary illustrates
+ * how to create a film grain table (film grain params as a function of time)
+ * from an input video and its corresponding denoised source.
+ *
+ * The --output-grain-table file can be passed as input to the encoder (in
+ * aomenc this is done through the "--film-grain-table" parameter).
+ *
+ * As an example, where the input source is an 854x480 yuv420p 8-bit video
+ * named "input.854_480.yuv" you would use steps similar to the following:
+ *
+ * # Run your denoiser (e.g, using hqdn3d filter):
+ * ffmpeg -vcodec rawvideo -video_size 854x480 -i input.854_480.yuv \
+ * -vf hqdn3d=5:5:5:5 -vcodec rawvideo -an -f rawvideo \
+ * denoised.854_480.yuv
+ *
+ * # Model the noise between the denoised version and original source:
+ * ./examples/noise_model --fps=25/1 --width=854 --height=480 --i420 \
+ * --input-denoised=denoised.854_480.yuv --input=original.854_480.yuv \
+ * --output-grain-table=film_grain.tbl
+ *
+ * # Encode with your favorite settings (including the grain table):
+ * aomenc --limit=100 --cpu-used=4 --input-bit-depth=8 \
+ * --i420 -w 854 -h 480 --end-usage=q --cq-level=25 --lag-in-frames=25 \
+ * --auto-alt-ref=2 --bit-depth=8 --film-grain-table=film_grain.tbl \
+ * -o denoised_with_grain_params.ivf denoised.854_480.yuv
+ */
+#include <math.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_encoder.h"
+#include "aom_dsp/aom_dsp_common.h"
+
+#if CONFIG_AV1_DECODER
+#include "aom_dsp/grain_synthesis.h"
+#endif
+
+#include "aom_dsp/grain_table.h"
+#include "aom_dsp/noise_model.h"
+#include "aom_dsp/noise_util.h"
+#include "aom_mem/aom_mem.h"
+#include "common/args.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr,
+ "Usage: %s --input=<input> --input-denoised=<denoised> "
+ "--output-grain-table=<outfile> "
+ "See comments in noise_model.c for more information.\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static const arg_def_t help =
+ ARG_DEF(NULL, "help", 0, "Show usage options and exit");
+static const arg_def_t width_arg =
+ ARG_DEF("w", "width", 1, "Input width (if rawvideo)");
+static const arg_def_t height_arg =
+ ARG_DEF("h", "height", 1, "Input height (if rawvideo)");
+static const arg_def_t skip_frames_arg =
+ ARG_DEF("s", "skip-frames", 1, "Number of frames to skip (default = 1)");
+static const arg_def_t fps_arg = ARG_DEF(NULL, "fps", 1, "Frame rate");
+static const arg_def_t input_arg = ARG_DEF("-i", "input", 1, "Input filename");
+static const arg_def_t output_grain_table_arg =
+ ARG_DEF("n", "output-grain-table", 1, "Output noise file");
+static const arg_def_t input_denoised_arg =
+ ARG_DEF("d", "input-denoised", 1, "Input denoised filename (YUV) only");
+static const arg_def_t flat_block_finder_arg =
+ ARG_DEF("b", "flat-block-finder", 1, "Run the flat block finder");
+static const arg_def_t block_size_arg =
+ ARG_DEF("b", "block-size", 1, "Block size");
+static const arg_def_t bit_depth_arg =
+ ARG_DEF(NULL, "bit-depth", 1, "Bit depth of input");
+static const arg_def_t use_i420 =
+ ARG_DEF(NULL, "i420", 0, "Input file (and denoised) is I420 (default)");
+static const arg_def_t use_i422 =
+ ARG_DEF(NULL, "i422", 0, "Input file (and denoised) is I422");
+static const arg_def_t use_i444 =
+ ARG_DEF(NULL, "i444", 0, "Input file (and denoised) is I444");
+static const arg_def_t debug_file_arg =
+ ARG_DEF(NULL, "debug-file", 1, "File to output debug info");
+
+typedef struct {
+ int width;
+ int height;
+ struct aom_rational fps;
+ const char *input;
+ const char *input_denoised;
+ const char *output_grain_table;
+ int img_fmt;
+ int block_size;
+ int bit_depth;
+ int run_flat_block_finder;
+ int force_flat_psd;
+ int skip_frames;
+ const char *debug_file;
+} noise_model_args_t;
+
+void parse_args(noise_model_args_t *noise_args, int *argc, char **argv) {
+ struct arg arg;
+ static const arg_def_t *main_args[] = { &help,
+ &input_arg,
+ &fps_arg,
+ &width_arg,
+ &height_arg,
+ &block_size_arg,
+ &output_grain_table_arg,
+ &input_denoised_arg,
+ &use_i420,
+ &use_i422,
+ &use_i444,
+ &debug_file_arg,
+ NULL };
+ for (int argi = *argc + 1; *argv; argi++, argv++) {
+ if (arg_match(&arg, &help, argv)) {
+ fprintf(stdout, "\nOptions:\n");
+ arg_show_usage(stdout, main_args);
+ exit(0);
+ } else if (arg_match(&arg, &width_arg, argv)) {
+ noise_args->width = atoi(arg.val);
+ } else if (arg_match(&arg, &height_arg, argv)) {
+ noise_args->height = atoi(arg.val);
+ } else if (arg_match(&arg, &input_arg, argv)) {
+ noise_args->input = arg.val;
+ } else if (arg_match(&arg, &input_denoised_arg, argv)) {
+ noise_args->input_denoised = arg.val;
+ } else if (arg_match(&arg, &output_grain_table_arg, argv)) {
+ noise_args->output_grain_table = arg.val;
+ } else if (arg_match(&arg, &block_size_arg, argv)) {
+ noise_args->block_size = atoi(arg.val);
+ } else if (arg_match(&arg, &bit_depth_arg, argv)) {
+ noise_args->bit_depth = atoi(arg.val);
+ } else if (arg_match(&arg, &flat_block_finder_arg, argv)) {
+ noise_args->run_flat_block_finder = atoi(arg.val);
+ } else if (arg_match(&arg, &fps_arg, argv)) {
+ noise_args->fps = arg_parse_rational(&arg);
+ } else if (arg_match(&arg, &use_i420, argv)) {
+ noise_args->img_fmt = AOM_IMG_FMT_I420;
+ } else if (arg_match(&arg, &use_i422, argv)) {
+ noise_args->img_fmt = AOM_IMG_FMT_I422;
+ } else if (arg_match(&arg, &use_i444, argv)) {
+ noise_args->img_fmt = AOM_IMG_FMT_I444;
+ } else if (arg_match(&arg, &skip_frames_arg, argv)) {
+ noise_args->skip_frames = atoi(arg.val);
+ } else if (arg_match(&arg, &debug_file_arg, argv)) {
+ noise_args->debug_file = arg.val;
+ } else {
+ fprintf(stdout, "Unknown arg: %s\n\nUsage:\n", *argv);
+ arg_show_usage(stdout, main_args);
+ exit(0);
+ }
+ }
+ if (noise_args->bit_depth > 8) {
+ noise_args->img_fmt |= AOM_IMG_FMT_HIGHBITDEPTH;
+ }
+}
+
+#if CONFIG_AV1_DECODER
+static void print_variance_y(FILE *debug_file, aom_image_t *raw,
+ aom_image_t *denoised, const uint8_t *flat_blocks,
+ int block_size, aom_film_grain_t *grain) {
+ aom_image_t renoised;
+ grain->apply_grain = 1;
+ grain->random_seed = 7391;
+ aom_img_alloc(&renoised, raw->fmt, raw->w, raw->h, 1);
+
+ if (av1_add_film_grain(grain, denoised, &renoised)) {
+ fprintf(stderr, "Internal failure in av1_add_film_grain().\n");
+ aom_img_free(&renoised);
+ return;
+ }
+
+ const int num_blocks_w = (raw->w + block_size - 1) / block_size;
+ const int num_blocks_h = (raw->h + block_size - 1) / block_size;
+ fprintf(debug_file, "x = [");
+ for (int by = 0; by < num_blocks_h; by++) {
+ for (int bx = 0; bx < num_blocks_w; bx++) {
+ double block_mean = 0;
+ double noise_std = 0, noise_mean = 0;
+ double renoise_std = 0, renoise_mean = 0;
+ for (int yi = 0; yi < block_size; ++yi) {
+ const int y = by * block_size + yi;
+ for (int xi = 0; xi < block_size; ++xi) {
+ const int x = bx * block_size + xi;
+ const double noise_v = (raw->planes[0][y * raw->stride[0] + x] -
+ denoised->planes[0][y * raw->stride[0] + x]);
+ noise_mean += noise_v;
+ noise_std += noise_v * noise_v;
+
+ block_mean += raw->planes[0][y * raw->stride[0] + x];
+
+ const double renoise_v =
+ (renoised.planes[0][y * raw->stride[0] + x] -
+ denoised->planes[0][y * raw->stride[0] + x]);
+ renoise_mean += renoise_v;
+ renoise_std += renoise_v * renoise_v;
+ }
+ }
+ int n = (block_size * block_size);
+ block_mean /= n;
+ noise_mean /= n;
+ renoise_mean /= n;
+ noise_std = sqrt(noise_std / n - noise_mean * noise_mean);
+ renoise_std = sqrt(renoise_std / n - renoise_mean * renoise_mean);
+ fprintf(debug_file, "%d %3.2lf %3.2lf %3.2lf ",
+ flat_blocks[by * num_blocks_w + bx], block_mean, noise_std,
+ renoise_std);
+ }
+ fprintf(debug_file, "\n");
+ }
+ fprintf(debug_file, "];\n");
+
+ if (raw->fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
+ fprintf(stderr,
+ "Detailed debug info not supported for high bit"
+ "depth formats\n");
+ } else {
+ fprintf(debug_file, "figure(2); clf;\n");
+ fprintf(debug_file,
+ "scatter(x(:, 2:4:end), x(:, 3:4:end), 'r'); hold on;\n");
+ fprintf(debug_file, "scatter(x(:, 2:4:end), x(:, 4:4:end), 'b');\n");
+ fprintf(debug_file,
+ "plot(linspace(0, 255, length(noise_strength_0)), "
+ "noise_strength_0, 'b');\n");
+ fprintf(debug_file,
+ "title('Scatter plot of intensity vs noise strength');\n");
+ fprintf(debug_file,
+ "legend('Actual', 'Estimated', 'Estimated strength');\n");
+ fprintf(debug_file, "figure(3); clf;\n");
+ fprintf(debug_file, "scatter(x(:, 3:4:end), x(:, 4:4:end), 'k');\n");
+ fprintf(debug_file, "title('Actual vs Estimated');\n");
+ fprintf(debug_file, "pause(3);\n");
+ }
+ aom_img_free(&renoised);
+}
+#endif
+
+static void print_debug_info(FILE *debug_file, aom_image_t *raw,
+ aom_image_t *denoised, uint8_t *flat_blocks,
+ int block_size, aom_noise_model_t *noise_model) {
+ (void)raw;
+ (void)denoised;
+ (void)flat_blocks;
+ (void)block_size;
+ fprintf(debug_file, "figure(3); clf;\n");
+ fprintf(debug_file, "figure(2); clf;\n");
+ fprintf(debug_file, "figure(1); clf;\n");
+ for (int c = 0; c < 3; ++c) {
+ fprintf(debug_file, "noise_strength_%d = [\n", c);
+ const aom_equation_system_t *eqns =
+ &noise_model->combined_state[c].strength_solver.eqns;
+ for (int k = 0; k < eqns->n; ++k) {
+ fprintf(debug_file, "%lf ", eqns->x[k]);
+ }
+ fprintf(debug_file, "];\n");
+ fprintf(debug_file, "plot(noise_strength_%d); hold on;\n", c);
+ }
+ fprintf(debug_file, "legend('Y', 'cb', 'cr');\n");
+ fprintf(debug_file, "title('Noise strength function');\n");
+
+#if CONFIG_AV1_DECODER
+ aom_film_grain_t grain;
+ aom_noise_model_get_grain_parameters(noise_model, &grain);
+ print_variance_y(debug_file, raw, denoised, flat_blocks, block_size, &grain);
+#endif
+ fflush(debug_file);
+}
+
+int main(int argc, char *argv[]) {
+ noise_model_args_t args = { 0, 0, { 25, 1 }, 0, 0, 0, AOM_IMG_FMT_I420,
+ 32, 8, 1, 0, 1, NULL };
+ aom_image_t raw, denoised;
+ FILE *infile = NULL;
+ AvxVideoInfo info;
+
+ memset(&info, 0, sizeof(info));
+
+ exec_name = argv[0];
+ parse_args(&args, &argc, argv + 1);
+
+ info.frame_width = args.width;
+ info.frame_height = args.height;
+ info.time_base.numerator = args.fps.den;
+ info.time_base.denominator = args.fps.num;
+
+ if (info.frame_width <= 0 || info.frame_height <= 0 ||
+ (info.frame_width % 2) != 0 || (info.frame_height % 2) != 0) {
+ die("Invalid frame size: %dx%d", info.frame_width, info.frame_height);
+ }
+ if (!aom_img_alloc(&raw, args.img_fmt, info.frame_width, info.frame_height,
+ 1)) {
+ die("Failed to allocate image.");
+ }
+ if (!aom_img_alloc(&denoised, args.img_fmt, info.frame_width,
+ info.frame_height, 1)) {
+ die("Failed to allocate image.");
+ }
+ infile = fopen(args.input, "r");
+ if (!infile) {
+ die("Failed to open input file:", args.input);
+ }
+ fprintf(stderr, "Bit depth: %d stride:%d\n", args.bit_depth, raw.stride[0]);
+
+ const int high_bd = args.bit_depth > 8;
+ const int block_size = args.block_size;
+ aom_flat_block_finder_t block_finder;
+ aom_flat_block_finder_init(&block_finder, block_size, args.bit_depth,
+ high_bd);
+
+ const int num_blocks_w = (info.frame_width + block_size - 1) / block_size;
+ const int num_blocks_h = (info.frame_height + block_size - 1) / block_size;
+ uint8_t *flat_blocks = (uint8_t *)aom_malloc(num_blocks_w * num_blocks_h);
+ // Sets the random seed on the first entry in the output table
+ int16_t random_seed = 7391;
+ aom_noise_model_t noise_model;
+ aom_noise_model_params_t params = { AOM_NOISE_SHAPE_SQUARE, 3, args.bit_depth,
+ high_bd };
+ aom_noise_model_init(&noise_model, params);
+
+ FILE *denoised_file = 0;
+ if (args.input_denoised) {
+ denoised_file = fopen(args.input_denoised, "rb");
+ if (!denoised_file)
+ die("Unable to open input_denoised: %s", args.input_denoised);
+ } else {
+ die("--input-denoised file must be specified");
+ }
+ FILE *debug_file = 0;
+ if (args.debug_file) {
+ debug_file = fopen(args.debug_file, "w");
+ }
+ aom_film_grain_table_t grain_table = { 0, 0 };
+
+ int64_t prev_timestamp = 0;
+ int frame_count = 0;
+ while (aom_img_read(&raw, infile)) {
+ if (args.input_denoised) {
+ if (!aom_img_read(&denoised, denoised_file)) {
+ die("Unable to read input denoised file");
+ }
+ }
+ if (frame_count % args.skip_frames == 0) {
+ int num_flat_blocks = num_blocks_w * num_blocks_h;
+ memset(flat_blocks, 1, num_flat_blocks);
+ if (args.run_flat_block_finder) {
+ memset(flat_blocks, 0, num_flat_blocks);
+ num_flat_blocks = aom_flat_block_finder_run(
+ &block_finder, raw.planes[0], info.frame_width, info.frame_height,
+ info.frame_width, flat_blocks);
+ fprintf(stdout, "Num flat blocks %d\n", num_flat_blocks);
+ }
+
+ const uint8_t *planes[3] = { raw.planes[0], raw.planes[1],
+ raw.planes[2] };
+ uint8_t *denoised_planes[3] = { denoised.planes[0], denoised.planes[1],
+ denoised.planes[2] };
+ int strides[3] = { raw.stride[0] >> high_bd, raw.stride[1] >> high_bd,
+ raw.stride[2] >> high_bd };
+ int chroma_sub[3] = { raw.x_chroma_shift, raw.y_chroma_shift, 0 };
+
+ fprintf(stdout, "Updating noise model...\n");
+ aom_noise_status_t status = aom_noise_model_update(
+ &noise_model, (const uint8_t *const *)planes,
+ (const uint8_t *const *)denoised_planes, info.frame_width,
+ info.frame_height, strides, chroma_sub, flat_blocks, block_size);
+
+ int64_t cur_timestamp =
+ frame_count * 10000000ULL * args.fps.den / args.fps.num;
+ if (status == AOM_NOISE_STATUS_DIFFERENT_NOISE_TYPE) {
+ fprintf(stdout,
+ "Noise type is different, updating parameters for time "
+ "[ %" PRId64 ", %" PRId64 ")\n",
+ prev_timestamp, cur_timestamp);
+ aom_film_grain_t grain;
+ aom_noise_model_get_grain_parameters(&noise_model, &grain);
+ grain.random_seed = random_seed;
+ random_seed = 0;
+ aom_film_grain_table_append(&grain_table, prev_timestamp, cur_timestamp,
+ &grain);
+ aom_noise_model_save_latest(&noise_model);
+ prev_timestamp = cur_timestamp;
+ }
+ if (debug_file) {
+ print_debug_info(debug_file, &raw, &denoised, flat_blocks, block_size,
+ &noise_model);
+ }
+ fprintf(stdout, "Done noise model update, status = %d\n", status);
+ }
+ frame_count++;
+ }
+
+ aom_film_grain_t grain;
+ aom_noise_model_get_grain_parameters(&noise_model, &grain);
+ grain.random_seed = random_seed;
+ aom_film_grain_table_append(&grain_table, prev_timestamp, INT64_MAX, &grain);
+ if (args.output_grain_table) {
+ struct aom_internal_error_info error_info;
+ if (AOM_CODEC_OK != aom_film_grain_table_write(&grain_table,
+ args.output_grain_table,
+ &error_info)) {
+ die("Unable to write output film grain table");
+ }
+ }
+ aom_film_grain_table_free(&grain_table);
+
+ if (infile) fclose(infile);
+ if (denoised_file) fclose(denoised_file);
+ if (debug_file) fclose(debug_file);
+ aom_img_free(&raw);
+ aom_img_free(&denoised);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/resize_util.c b/media/libaom/src/examples/resize_util.c
new file mode 100644
index 000000000..6a84d5740
--- /dev/null
+++ b/media/libaom/src/examples/resize_util.c
@@ -0,0 +1,124 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#include <assert.h>
+#include <limits.h>
+#include <math.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "av1/common/resize.h"
+#include "common/tools_common.h"
+
+static const char *exec_name = NULL;
+
+static void usage() {
+ printf("Usage:\n");
+ printf("%s <input_yuv> <width>x<height> <target_width>x<target_height> ",
+ exec_name);
+ printf("<output_yuv> [<frames>]\n");
+}
+
+void usage_exit(void) {
+ usage();
+ exit(EXIT_FAILURE);
+}
+
+static int parse_dim(char *v, int *width, int *height) {
+ char *x = strchr(v, 'x');
+ if (x == NULL) x = strchr(v, 'X');
+ if (x == NULL) return 0;
+ *width = atoi(v);
+ *height = atoi(&x[1]);
+ if (*width <= 0 || *height <= 0)
+ return 0;
+ else
+ return 1;
+}
+
+int main(int argc, char *argv[]) {
+ char *fin, *fout;
+ FILE *fpin, *fpout;
+ uint8_t *inbuf, *outbuf;
+ uint8_t *inbuf_u, *outbuf_u;
+ uint8_t *inbuf_v, *outbuf_v;
+ int f, frames;
+ int width, height, target_width, target_height;
+
+ exec_name = argv[0];
+
+ if (argc < 5) {
+ printf("Incorrect parameters:\n");
+ usage();
+ return 1;
+ }
+
+ fin = argv[1];
+ fout = argv[4];
+ if (!parse_dim(argv[2], &width, &height)) {
+ printf("Incorrect parameters: %s\n", argv[2]);
+ usage();
+ return 1;
+ }
+ if (!parse_dim(argv[3], &target_width, &target_height)) {
+ printf("Incorrect parameters: %s\n", argv[3]);
+ usage();
+ return 1;
+ }
+
+ fpin = fopen(fin, "rb");
+ if (fpin == NULL) {
+ printf("Can't open file %s to read\n", fin);
+ usage();
+ return 1;
+ }
+ fpout = fopen(fout, "wb");
+ if (fpout == NULL) {
+ printf("Can't open file %s to write\n", fout);
+ usage();
+ return 1;
+ }
+ if (argc >= 6)
+ frames = atoi(argv[5]);
+ else
+ frames = INT_MAX;
+
+ printf("Input size: %dx%d\n", width, height);
+ printf("Target size: %dx%d, Frames: ", target_width, target_height);
+ if (frames == INT_MAX)
+ printf("All\n");
+ else
+ printf("%d\n", frames);
+
+ inbuf = (uint8_t *)malloc(width * height * 3 / 2);
+ outbuf = (uint8_t *)malloc(target_width * target_height * 3 / 2);
+ inbuf_u = inbuf + width * height;
+ inbuf_v = inbuf_u + width * height / 4;
+ outbuf_u = outbuf + target_width * target_height;
+ outbuf_v = outbuf_u + target_width * target_height / 4;
+ f = 0;
+ while (f < frames) {
+ if (fread(inbuf, width * height * 3 / 2, 1, fpin) != 1) break;
+ av1_resize_frame420(inbuf, width, inbuf_u, inbuf_v, width / 2, height,
+ width, outbuf, target_width, outbuf_u, outbuf_v,
+ target_width / 2, target_height, target_width);
+ fwrite(outbuf, target_width * target_height * 3 / 2, 1, fpout);
+ f++;
+ }
+ printf("%d frames processed\n", f);
+ fclose(fpin);
+ fclose(fpout);
+
+ free(inbuf);
+ free(outbuf);
+ return 0;
+}
diff --git a/media/libaom/src/examples/scalable_decoder.c b/media/libaom/src/examples/scalable_decoder.c
new file mode 100644
index 000000000..c22924223
--- /dev/null
+++ b/media/libaom/src/examples/scalable_decoder.c
@@ -0,0 +1,185 @@
+/*
+ * Copyright (c) 2018, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Scalable Decoder
+// ==============
+//
+// This is an example of a scalable decoder loop. It takes a 2-spatial-layer
+// input file
+// containing the compressed data (in OBU format), passes it through the
+// decoder, and writes the decompressed frames to disk. The base layer and
+// enhancement layers are stored as separate files, out_lyr0.yuv and
+// out_lyr1.yuv, respectively.
+//
+// Standard Includes
+// -----------------
+// For decoders, you only have to include `aom_decoder.h` and then any
+// header files for the specific codecs you use. In this case, we're using
+// av1.
+//
+// Initializing The Codec
+// ----------------------
+// The libaom decoder is initialized by the call to aom_codec_dec_init().
+// Determining the codec interface to use is handled by AvxVideoReader and the
+// functions prefixed with aom_video_reader_. Discussion of those functions is
+// beyond the scope of this example, but the main gist is to open the input file
+// and parse just enough of it to determine if it's a AVx file and which AVx
+// codec is contained within the file.
+// Note the NULL pointer passed to aom_codec_dec_init(). We do that in this
+// example because we want the algorithm to determine the stream configuration
+// (width/height) and allocate memory automatically.
+//
+// Decoding A Frame
+// ----------------
+// Once the frame has been read into memory, it is decoded using the
+// `aom_codec_decode` function. The call takes a pointer to the data
+// (`frame`) and the length of the data (`frame_size`). No application data
+// is associated with the frame in this example, so the `user_priv`
+// parameter is NULL. The `deadline` parameter is left at zero for this
+// example. This parameter is generally only used when doing adaptive post
+// processing.
+//
+// Codecs may produce a variable number of output frames for every call to
+// `aom_codec_decode`. These frames are retrieved by the
+// `aom_codec_get_frame` iterator function. The iterator variable `iter` is
+// initialized to NULL each time `aom_codec_decode` is called.
+// `aom_codec_get_frame` is called in a loop, returning a pointer to a
+// decoded image or NULL to indicate the end of list.
+//
+// Processing The Decoded Data
+// ---------------------------
+// In this example, we simply write the encoded data to disk. It is
+// important to honor the image's `stride` values.
+//
+// Cleanup
+// -------
+// The `aom_codec_destroy` call frees any memory allocated by the codec.
+//
+// Error Handling
+// --------------
+// This example does not special case any error return codes. If there was
+// an error, a descriptive message is printed and the program exits. With
+// few exceptions, aom_codec functions return an enumerated error status,
+// with the value `0` indicating success.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_decoder.h"
+#include "aom/aomdx.h"
+#include "common/obudec.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+
+static const char *exec_name;
+
+#define MAX_LAYERS 5
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s <infile>\n", exec_name);
+ exit(EXIT_FAILURE);
+}
+
+int main(int argc, char **argv) {
+ int frame_cnt = 0;
+ FILE *outfile[MAX_LAYERS];
+ char filename[80];
+ aom_codec_ctx_t codec;
+ const AvxInterface *decoder = NULL;
+ FILE *inputfile = NULL;
+ uint8_t *buf = NULL;
+ size_t bytes_in_buffer = 0;
+ size_t buffer_size = 0;
+ struct AvxInputContext aom_input_ctx;
+ struct ObuDecInputContext obu_ctx = { &aom_input_ctx, NULL, 0, 0, 0 };
+ aom_codec_stream_info_t si;
+ uint8_t tmpbuf[32];
+ unsigned int i;
+
+ exec_name = argv[0];
+
+ if (argc != 2) die("Invalid number of arguments.");
+
+ if (!(inputfile = fopen(argv[1], "rb")))
+ die("Failed to open %s for read.", argv[1]);
+ obu_ctx.avx_ctx->file = inputfile;
+ obu_ctx.avx_ctx->filename = argv[1];
+
+ decoder = get_aom_decoder_by_index(0);
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder.");
+
+ if (aom_codec_control(&codec, AV1D_SET_OUTPUT_ALL_LAYERS, 1)) {
+ die_codec(&codec, "Failed to set output_all_layers control.");
+ }
+
+ // peak sequence header OBU to get number of spatial layers
+ const size_t ret = fread(tmpbuf, 1, 32, inputfile);
+ if (ret != 32) die_codec(&codec, "Input is not a valid obu file");
+ si.is_annexb = 0;
+ if (aom_codec_peek_stream_info(decoder->codec_interface(), tmpbuf, 32, &si)) {
+ die_codec(&codec, "Input is not a valid obu file");
+ }
+ fseek(inputfile, -32, SEEK_CUR);
+
+ if (!file_is_obu(&obu_ctx))
+ die_codec(&codec, "Input is not a valid obu file");
+
+ // open base layer output yuv file
+ snprintf(filename, sizeof(filename), "out_lyr%d.yuv", 0);
+ if (!(outfile[0] = fopen(filename, "wb")))
+ die("Failed top open output for writing.");
+
+ // open any enhancement layer output yuv files
+ for (i = 1; i < si.number_spatial_layers; i++) {
+ snprintf(filename, sizeof(filename), "out_lyr%d.yuv", i);
+ if (!(outfile[i] = fopen(filename, "wb")))
+ die("Failed to open output for writing.");
+ }
+
+ while (!obudec_read_temporal_unit(&obu_ctx, &buf, &bytes_in_buffer,
+ &buffer_size)) {
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img = NULL;
+ if (aom_codec_decode(&codec, buf, bytes_in_buffer, NULL))
+ die_codec(&codec, "Failed to decode frame.");
+
+ while ((img = aom_codec_get_frame(&codec, &iter)) != NULL) {
+ aom_image_t *img_shifted =
+ aom_img_alloc(NULL, AOM_IMG_FMT_I420, img->d_w, img->d_h, 16);
+ img_shifted->bit_depth = 8;
+ aom_img_downshift(img_shifted, img,
+ img->bit_depth - img_shifted->bit_depth);
+ if (img->spatial_id == 0) {
+ printf("Writing base layer 0 %d\n", frame_cnt);
+ aom_img_write(img_shifted, outfile[0]);
+ } else if (img->spatial_id <= (int)(si.number_spatial_layers - 1)) {
+ printf("Writing enhancement layer %d %d\n", img->spatial_id, frame_cnt);
+ aom_img_write(img_shifted, outfile[img->spatial_id]);
+ } else {
+ die_codec(&codec, "Invalid bitstream. Layer id exceeds layer count");
+ }
+ if (img->spatial_id == (int)(si.number_spatial_layers - 1)) ++frame_cnt;
+ }
+ }
+
+ printf("Processed %d frames.\n", frame_cnt);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
+
+ for (i = 0; i < si.number_spatial_layers; i++) fclose(outfile[i]);
+
+ fclose(inputfile);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/scalable_encoder.c b/media/libaom/src/examples/scalable_encoder.c
new file mode 100644
index 000000000..7af03e29f
--- /dev/null
+++ b/media/libaom/src/examples/scalable_encoder.c
@@ -0,0 +1,289 @@
+/*
+ * Copyright (c) 2018, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Scalable Encoder
+// ==============
+//
+// This is an example of a scalable encoder loop. It takes two input files in
+// YV12 format, passes it through the encoder, and writes the compressed
+// frames to disk in OBU format.
+//
+// Getting The Default Configuration
+// ---------------------------------
+// Encoders have the notion of "usage profiles." For example, an encoder
+// may want to publish default configurations for both a video
+// conferencing application and a best quality offline encoder. These
+// obviously have very different default settings. Consult the
+// documentation for your codec to see if it provides any default
+// configurations. All codecs provide a default configuration, number 0,
+// which is valid for material in the vacinity of QCIF/QVGA.
+//
+// Updating The Configuration
+// ---------------------------------
+// Almost all applications will want to update the default configuration
+// with settings specific to their usage. Here we set the width and height
+// of the video file to that specified on the command line. We also scale
+// the default bitrate based on the ratio between the default resolution
+// and the resolution specified on the command line.
+//
+// Encoding A Frame
+// ----------------
+// The frame is read as a continuous block (size = width * height * 3 / 2)
+// from the input file. If a frame was read (the input file has not hit
+// EOF) then the frame is passed to the encoder. Otherwise, a NULL
+// is passed, indicating the End-Of-Stream condition to the encoder. The
+// `frame_cnt` is reused as the presentation time stamp (PTS) and each
+// frame is shown for one frame-time in duration. The flags parameter is
+// unused in this example.
+
+// Forced Keyframes
+// ----------------
+// Keyframes can be forced by setting the AOM_EFLAG_FORCE_KF bit of the
+// flags passed to `aom_codec_control()`. In this example, we force a
+// keyframe every <keyframe-interval> frames. Note, the output stream can
+// contain additional keyframes beyond those that have been forced using the
+// AOM_EFLAG_FORCE_KF flag because of automatic keyframe placement by the
+// encoder.
+//
+// Processing The Encoded Data
+// ---------------------------
+// Each packet of type `AOM_CODEC_CX_FRAME_PKT` contains the encoded data
+// for this frame. We write a IVF frame header, followed by the raw data.
+//
+// Cleanup
+// -------
+// The `aom_codec_destroy` call frees any memory allocated by the codec.
+//
+// Error Handling
+// --------------
+// This example does not special case any error return codes. If there was
+// an error, a descriptive message is printed and the program exits. With
+// few exeptions, aom_codec functions return an enumerated error status,
+// with the value `0` indicating success.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_encoder.h"
+#include "aom/aomcx.h"
+#include "av1/common/enums.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr,
+ "Usage: %s <codec> <width> <height> <infile0> <infile1> "
+ "<outfile> <frames to encode>\n"
+ "See comments in scalable_encoder.c for more information.\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static int encode_frame(aom_codec_ctx_t *codec, aom_image_t *img,
+ int frame_index, int flags, FILE *outfile) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res =
+ aom_codec_encode(codec, img, frame_index, 1, flags);
+ if (res != AOM_CODEC_OK) die_codec(codec, "Failed to encode frame");
+
+ while ((pkt = aom_codec_get_cx_data(codec, &iter)) != NULL) {
+ got_pkts = 1;
+
+ if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
+ const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
+ if (fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz, outfile) !=
+ pkt->data.frame.sz) {
+ die_codec(codec, "Failed to write compressed frame");
+ }
+ printf(keyframe ? "K" : ".");
+ printf(" %6d\n", (int)pkt->data.frame.sz);
+ fflush(stdout);
+ }
+ }
+
+ return got_pkts;
+}
+
+int main(int argc, char **argv) {
+ FILE *infile0 = NULL;
+ FILE *infile1 = NULL;
+ aom_codec_ctx_t codec;
+ aom_codec_enc_cfg_t cfg;
+ int frame_count = 0;
+ aom_image_t raw0, raw1;
+ aom_codec_err_t res;
+ AvxVideoInfo info;
+ const AvxInterface *encoder = NULL;
+ const int fps = 30;
+ const int bitrate = 200;
+ int keyframe_interval = 0;
+ int max_frames = 0;
+ int frames_encoded = 0;
+ const char *codec_arg = NULL;
+ const char *width_arg = NULL;
+ const char *height_arg = NULL;
+ const char *infile0_arg = NULL;
+ const char *infile1_arg = NULL;
+ const char *outfile_arg = NULL;
+ // const char *keyframe_interval_arg = NULL;
+ FILE *outfile = NULL;
+
+ exec_name = argv[0];
+
+ // Clear explicitly, as simply assigning "{ 0 }" generates
+ // "missing-field-initializers" warning in some compilers.
+ memset(&info, 0, sizeof(info));
+
+ if (argc != 8) die("Invalid number of arguments");
+
+ codec_arg = argv[1];
+ width_arg = argv[2];
+ height_arg = argv[3];
+ infile0_arg = argv[4];
+ infile1_arg = argv[5];
+ outfile_arg = argv[6];
+ max_frames = (int)strtol(argv[7], NULL, 0);
+
+ encoder = get_aom_encoder_by_name(codec_arg);
+ if (!encoder) die("Unsupported codec.");
+
+ info.codec_fourcc = encoder->fourcc;
+ info.frame_width = (int)strtol(width_arg, NULL, 0);
+ info.frame_height = (int)strtol(height_arg, NULL, 0);
+ info.time_base.numerator = 1;
+ info.time_base.denominator = fps;
+
+ if (info.frame_width <= 0 || info.frame_height <= 0 ||
+ (info.frame_width % 2) != 0 || (info.frame_height % 2) != 0) {
+ die("Invalid frame size: %dx%d", info.frame_width, info.frame_height);
+ }
+
+ if (!aom_img_alloc(&raw0, AOM_IMG_FMT_I420, info.frame_width,
+ info.frame_height, 1)) {
+ die("Failed to allocate image for layer 0.");
+ }
+ if (!aom_img_alloc(&raw1, AOM_IMG_FMT_I420, info.frame_width,
+ info.frame_height, 1)) {
+ die("Failed to allocate image for layer 1.");
+ }
+
+ // keyframe_interval = (int)strtol(keyframe_interval_arg, NULL, 0);
+ keyframe_interval = 100;
+ if (keyframe_interval < 0) die("Invalid keyframe interval value.");
+
+ printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
+
+ res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
+ if (res) die_codec(&codec, "Failed to get default codec config.");
+
+ cfg.g_w = info.frame_width;
+ cfg.g_h = info.frame_height;
+ cfg.g_timebase.num = info.time_base.numerator;
+ cfg.g_timebase.den = info.time_base.denominator;
+ cfg.rc_target_bitrate = bitrate;
+ cfg.g_error_resilient = 0;
+ cfg.g_lag_in_frames = 0;
+ cfg.rc_end_usage = AOM_Q;
+ cfg.save_as_annexb = 0;
+
+ outfile = fopen(outfile_arg, "wb");
+ if (!outfile) die("Failed to open %s for writing.", outfile_arg);
+
+ if (!(infile0 = fopen(infile0_arg, "rb")))
+ die("Failed to open %s for reading.", infile0_arg);
+ if (!(infile1 = fopen(infile1_arg, "rb")))
+ die("Failed to open %s for reading.", infile0_arg);
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), &cfg, 0))
+ die_codec(&codec, "Failed to initialize encoder");
+ if (aom_codec_control(&codec, AOME_SET_CPUUSED, 8))
+ die_codec(&codec, "Failed to set cpu to 8");
+
+ if (aom_codec_control(&codec, AV1E_SET_TILE_COLUMNS, 2))
+ die_codec(&codec, "Failed to set tile columns to 2");
+ if (aom_codec_control(&codec, AV1E_SET_NUM_TG, 3))
+ die_codec(&codec, "Failed to set num of tile groups to 3");
+
+ if (aom_codec_control(&codec, AOME_SET_NUMBER_SPATIAL_LAYERS, 2))
+ die_codec(&codec, "Failed to set number of spatial layers to 2");
+
+ // Encode frames.
+ while (aom_img_read(&raw0, infile0)) {
+ int flags = 0;
+
+ // configure and encode base layer
+
+ if (keyframe_interval > 0 && frames_encoded % keyframe_interval == 0)
+ flags |= AOM_EFLAG_FORCE_KF;
+ else
+ // use previous base layer (LAST) as sole reference
+ // save this frame as LAST to be used as reference by enhanmcent layer
+ // and next base layer
+ flags |= AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 |
+ AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF |
+ AOM_EFLAG_NO_REF_BWD | AOM_EFLAG_NO_REF_ARF2 |
+ AOM_EFLAG_NO_UPD_GF | AOM_EFLAG_NO_UPD_ARF |
+ AOM_EFLAG_NO_UPD_ENTROPY;
+ cfg.g_w = info.frame_width;
+ cfg.g_h = info.frame_height;
+ if (aom_codec_enc_config_set(&codec, &cfg))
+ die_codec(&codec, "Failed to set enc cfg for layer 0");
+ if (aom_codec_control(&codec, AOME_SET_SPATIAL_LAYER_ID, 0))
+ die_codec(&codec, "Failed to set layer id to 0");
+ if (aom_codec_control(&codec, AOME_SET_CQ_LEVEL, 62))
+ die_codec(&codec, "Failed to set cq level");
+ encode_frame(&codec, &raw0, frame_count++, flags, outfile);
+
+ // configure and encode enhancement layer
+
+ // use LAST (base layer) as sole reference
+ flags = AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 |
+ AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF | AOM_EFLAG_NO_REF_BWD |
+ AOM_EFLAG_NO_REF_ARF2 | AOM_EFLAG_NO_UPD_LAST |
+ AOM_EFLAG_NO_UPD_GF | AOM_EFLAG_NO_UPD_ARF |
+ AOM_EFLAG_NO_UPD_ENTROPY;
+ cfg.g_w = info.frame_width;
+ cfg.g_h = info.frame_height;
+ aom_img_read(&raw1, infile1);
+ if (aom_codec_enc_config_set(&codec, &cfg))
+ die_codec(&codec, "Failed to set enc cfg for layer 1");
+ if (aom_codec_control(&codec, AOME_SET_SPATIAL_LAYER_ID, 1))
+ die_codec(&codec, "Failed to set layer id to 1");
+ if (aom_codec_control(&codec, AOME_SET_CQ_LEVEL, 10))
+ die_codec(&codec, "Failed to set cq level");
+ encode_frame(&codec, &raw1, frame_count++, flags, outfile);
+
+ frames_encoded++;
+
+ if (max_frames > 0 && frames_encoded >= max_frames) break;
+ }
+
+ // Flush encoder.
+ while (encode_frame(&codec, NULL, -1, 0, outfile)) continue;
+
+ printf("\n");
+ fclose(infile0);
+ fclose(infile1);
+ printf("Processed %d frames.\n", frame_count / 2);
+
+ aom_img_free(&raw0);
+ aom_img_free(&raw1);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ fclose(outfile);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/set_maps.c b/media/libaom/src/examples/set_maps.c
new file mode 100644
index 000000000..9aeb96e43
--- /dev/null
+++ b/media/libaom/src/examples/set_maps.c
@@ -0,0 +1,208 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// AOM Set Active and ROI Maps
+// ===========================
+//
+// This is an example demonstrating how to control the AOM encoder's
+// ROI and Active maps.
+//
+// ROI (Reigon of Interest) maps are a way for the application to assign
+// each macroblock in the image to a region, and then set quantizer and
+// filtering parameters on that image.
+//
+// Active maps are a way for the application to specify on a
+// macroblock-by-macroblock basis whether there is any activity in that
+// macroblock.
+//
+//
+// Configuration
+// -------------
+// An ROI map is set on frame 22. If the width of the image in macroblocks
+// is evenly divisble by 4, then the output will appear to have distinct
+// columns, where the quantizer, loopfilter, and static threshold differ
+// from column to column.
+//
+// An active map is set on frame 33. If the width of the image in macroblocks
+// is evenly divisble by 4, then the output will appear to have distinct
+// columns, where one column will have motion and the next will not.
+//
+// The active map is cleared on frame 44.
+//
+// Observing The Effects
+// ---------------------
+// Use the `simple_decoder` example to decode this sample, and observe
+// the change in the image at frames 22, 33, and 44.
+
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_encoder.h"
+#include "aom/aomcx.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s <codec> <width> <height> <infile> <outfile>\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static void set_active_map(const aom_codec_enc_cfg_t *cfg,
+ aom_codec_ctx_t *codec) {
+ unsigned int i;
+ aom_active_map_t map = { 0, 0, 0 };
+
+ map.rows = (cfg->g_h + 15) / 16;
+ map.cols = (cfg->g_w + 15) / 16;
+
+ map.active_map = (uint8_t *)malloc(map.rows * map.cols);
+ for (i = 0; i < map.rows * map.cols; ++i) map.active_map[i] = i % 2;
+
+ if (aom_codec_control(codec, AOME_SET_ACTIVEMAP, &map))
+ die_codec(codec, "Failed to set active map");
+
+ free(map.active_map);
+}
+
+static void unset_active_map(const aom_codec_enc_cfg_t *cfg,
+ aom_codec_ctx_t *codec) {
+ aom_active_map_t map = { 0, 0, 0 };
+
+ map.rows = (cfg->g_h + 15) / 16;
+ map.cols = (cfg->g_w + 15) / 16;
+ map.active_map = NULL;
+
+ if (aom_codec_control(codec, AOME_SET_ACTIVEMAP, &map))
+ die_codec(codec, "Failed to set active map");
+}
+
+static int encode_frame(aom_codec_ctx_t *codec, aom_image_t *img,
+ int frame_index, AvxVideoWriter *writer) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res = aom_codec_encode(codec, img, frame_index, 1, 0);
+ if (res != AOM_CODEC_OK) die_codec(codec, "Failed to encode frame");
+
+ while ((pkt = aom_codec_get_cx_data(codec, &iter)) != NULL) {
+ got_pkts = 1;
+
+ if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
+ const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
+ if (!aom_video_writer_write_frame(writer, pkt->data.frame.buf,
+ pkt->data.frame.sz,
+ pkt->data.frame.pts)) {
+ die_codec(codec, "Failed to write compressed frame");
+ }
+
+ printf(keyframe ? "K" : ".");
+ fflush(stdout);
+ }
+ }
+
+ return got_pkts;
+}
+
+int main(int argc, char **argv) {
+ FILE *infile = NULL;
+ aom_codec_ctx_t codec;
+ aom_codec_enc_cfg_t cfg;
+ int frame_count = 0;
+ const int limit = 15;
+ aom_image_t raw;
+ aom_codec_err_t res;
+ AvxVideoInfo info;
+ AvxVideoWriter *writer = NULL;
+ const AvxInterface *encoder = NULL;
+ const int fps = 2; // TODO(dkovalev) add command line argument
+ const double bits_per_pixel_per_frame = 0.067;
+
+ exec_name = argv[0];
+ if (argc != 6) die("Invalid number of arguments");
+
+ memset(&info, 0, sizeof(info));
+
+ encoder = get_aom_encoder_by_name(argv[1]);
+ if (encoder == NULL) {
+ die("Unsupported codec.");
+ }
+ assert(encoder != NULL);
+ info.codec_fourcc = encoder->fourcc;
+ info.frame_width = (int)strtol(argv[2], NULL, 0);
+ info.frame_height = (int)strtol(argv[3], NULL, 0);
+ info.time_base.numerator = 1;
+ info.time_base.denominator = fps;
+
+ if (info.frame_width <= 0 || info.frame_height <= 0 ||
+ (info.frame_width % 2) != 0 || (info.frame_height % 2) != 0) {
+ die("Invalid frame size: %dx%d", info.frame_width, info.frame_height);
+ }
+
+ if (!aom_img_alloc(&raw, AOM_IMG_FMT_I420, info.frame_width,
+ info.frame_height, 1)) {
+ die("Failed to allocate image.");
+ }
+
+ printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
+
+ res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
+ if (res) die_codec(&codec, "Failed to get default codec config.");
+
+ cfg.g_w = info.frame_width;
+ cfg.g_h = info.frame_height;
+ cfg.g_timebase.num = info.time_base.numerator;
+ cfg.g_timebase.den = info.time_base.denominator;
+ cfg.rc_target_bitrate =
+ (unsigned int)(bits_per_pixel_per_frame * cfg.g_w * cfg.g_h * fps / 1000);
+ cfg.g_lag_in_frames = 0;
+
+ writer = aom_video_writer_open(argv[5], kContainerIVF, &info);
+ if (!writer) die("Failed to open %s for writing.", argv[5]);
+
+ if (!(infile = fopen(argv[4], "rb")))
+ die("Failed to open %s for reading.", argv[4]);
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), &cfg, 0))
+ die_codec(&codec, "Failed to initialize encoder");
+
+ // Encode frames.
+ while (aom_img_read(&raw, infile) && frame_count < limit) {
+ ++frame_count;
+
+ if (frame_count == 5) {
+ set_active_map(&cfg, &codec);
+ } else if (frame_count == 11) {
+ unset_active_map(&cfg, &codec);
+ }
+
+ encode_frame(&codec, &raw, frame_count, writer);
+ }
+
+ // Flush encoder.
+ while (encode_frame(&codec, NULL, -1, writer)) {
+ }
+
+ printf("\n");
+ fclose(infile);
+ printf("Processed %d frames.\n", frame_count);
+
+ aom_img_free(&raw);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ aom_video_writer_close(writer);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/simple_decoder.c b/media/libaom/src/examples/simple_decoder.c
new file mode 100644
index 000000000..d098d1e0b
--- /dev/null
+++ b/media/libaom/src/examples/simple_decoder.c
@@ -0,0 +1,146 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Simple Decoder
+// ==============
+//
+// This is an example of a simple decoder loop. It takes an input file
+// containing the compressed data (in IVF format), passes it through the
+// decoder, and writes the decompressed frames to disk. Other decoder
+// examples build upon this one.
+//
+// The details of the IVF format have been elided from this example for
+// simplicity of presentation, as IVF files will not generally be used by
+// your application. In general, an IVF file consists of a file header,
+// followed by a variable number of frames. Each frame consists of a frame
+// header followed by a variable length payload. The length of the payload
+// is specified in the first four bytes of the frame header. The payload is
+// the raw compressed data.
+//
+// Standard Includes
+// -----------------
+// For decoders, you only have to include `aom_decoder.h` and then any
+// header files for the specific codecs you use. In this case, we're using
+// aom.
+//
+// Initializing The Codec
+// ----------------------
+// The libaom decoder is initialized by the call to aom_codec_dec_init().
+// Determining the codec interface to use is handled by AvxVideoReader and the
+// functions prefixed with aom_video_reader_. Discussion of those functions is
+// beyond the scope of this example, but the main gist is to open the input file
+// and parse just enough of it to determine if it's a AVx file and which AVx
+// codec is contained within the file.
+// Note the NULL pointer passed to aom_codec_dec_init(). We do that in this
+// example because we want the algorithm to determine the stream configuration
+// (width/height) and allocate memory automatically.
+//
+// Decoding A Frame
+// ----------------
+// Once the frame has been read into memory, it is decoded using the
+// `aom_codec_decode` function. The call takes a pointer to the data
+// (`frame`) and the length of the data (`frame_size`). No application data
+// is associated with the frame in this example, so the `user_priv`
+// parameter is NULL.
+//
+// Codecs may produce a variable number of output frames for every call to
+// `aom_codec_decode`. These frames are retrieved by the
+// `aom_codec_get_frame` iterator function. The iterator variable `iter` is
+// initialized to NULL each time `aom_codec_decode` is called.
+// `aom_codec_get_frame` is called in a loop, returning a pointer to a
+// decoded image or NULL to indicate the end of list.
+//
+// Processing The Decoded Data
+// ---------------------------
+// In this example, we simply write the encoded data to disk. It is
+// important to honor the image's `stride` values.
+//
+// Cleanup
+// -------
+// The `aom_codec_destroy` call frees any memory allocated by the codec.
+//
+// Error Handling
+// --------------
+// This example does not special case any error return codes. If there was
+// an error, a descriptive message is printed and the program exits. With
+// few exceptions, aom_codec functions return an enumerated error status,
+// with the value `0` indicating success.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_decoder.h"
+#include "common/tools_common.h"
+#include "common/video_reader.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr, "Usage: %s <infile> <outfile>\n", exec_name);
+ exit(EXIT_FAILURE);
+}
+
+int main(int argc, char **argv) {
+ int frame_cnt = 0;
+ FILE *outfile = NULL;
+ aom_codec_ctx_t codec;
+ AvxVideoReader *reader = NULL;
+ const AvxInterface *decoder = NULL;
+ const AvxVideoInfo *info = NULL;
+
+ exec_name = argv[0];
+
+ if (argc != 3) die("Invalid number of arguments.");
+
+ reader = aom_video_reader_open(argv[1]);
+ if (!reader) die("Failed to open %s for reading.", argv[1]);
+
+ if (!(outfile = fopen(argv[2], "wb")))
+ die("Failed to open %s for writing.", argv[2]);
+
+ info = aom_video_reader_get_info(reader);
+
+ decoder = get_aom_decoder_by_fourcc(info->codec_fourcc);
+ if (!decoder) die("Unknown input codec.");
+
+ printf("Using %s\n", aom_codec_iface_name(decoder->codec_interface()));
+
+ if (aom_codec_dec_init(&codec, decoder->codec_interface(), NULL, 0))
+ die_codec(&codec, "Failed to initialize decoder.");
+
+ while (aom_video_reader_read_frame(reader)) {
+ aom_codec_iter_t iter = NULL;
+ aom_image_t *img = NULL;
+ size_t frame_size = 0;
+ const unsigned char *frame =
+ aom_video_reader_get_frame(reader, &frame_size);
+ if (aom_codec_decode(&codec, frame, frame_size, NULL))
+ die_codec(&codec, "Failed to decode frame.");
+
+ while ((img = aom_codec_get_frame(&codec, &iter)) != NULL) {
+ aom_img_write(img, outfile);
+ ++frame_cnt;
+ }
+ }
+
+ printf("Processed %d frames.\n", frame_cnt);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
+
+ printf("Play: ffplay -f rawvideo -pix_fmt yuv420p -s %dx%d %s\n",
+ info->frame_width, info->frame_height, argv[2]);
+
+ aom_video_reader_close(reader);
+
+ fclose(outfile);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/simple_encoder.c b/media/libaom/src/examples/simple_encoder.c
new file mode 100644
index 000000000..01a37cf0c
--- /dev/null
+++ b/media/libaom/src/examples/simple_encoder.c
@@ -0,0 +1,249 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Simple Encoder
+// ==============
+//
+// This is an example of a simple encoder loop. It takes an input file in
+// YV12 format, passes it through the encoder, and writes the compressed
+// frames to disk in IVF format. Other decoder examples build upon this
+// one.
+//
+// The details of the IVF format have been elided from this example for
+// simplicity of presentation, as IVF files will not generally be used by
+// your application. In general, an IVF file consists of a file header,
+// followed by a variable number of frames. Each frame consists of a frame
+// header followed by a variable length payload. The length of the payload
+// is specified in the first four bytes of the frame header. The payload is
+// the raw compressed data.
+//
+// Standard Includes
+// -----------------
+// For encoders, you only have to include `aom_encoder.h` and then any
+// header files for the specific codecs you use. In this case, we're using
+// aom.
+//
+// Getting The Default Configuration
+// ---------------------------------
+// Encoders have the notion of "usage profiles." For example, an encoder
+// may want to publish default configurations for both a video
+// conferencing application and a best quality offline encoder. These
+// obviously have very different default settings. Consult the
+// documentation for your codec to see if it provides any default
+// configurations. All codecs provide a default configuration, number 0,
+// which is valid for material in the vacinity of QCIF/QVGA.
+//
+// Updating The Configuration
+// ---------------------------------
+// Almost all applications will want to update the default configuration
+// with settings specific to their usage. Here we set the width and height
+// of the video file to that specified on the command line. We also scale
+// the default bitrate based on the ratio between the default resolution
+// and the resolution specified on the command line.
+//
+// Initializing The Codec
+// ----------------------
+// The encoder is initialized by the following code.
+//
+// Encoding A Frame
+// ----------------
+// The frame is read as a continuous block (size width * height * 3 / 2)
+// from the input file. If a frame was read (the input file has not hit
+// EOF) then the frame is passed to the encoder. Otherwise, a NULL
+// is passed, indicating the End-Of-Stream condition to the encoder. The
+// `frame_cnt` is reused as the presentation time stamp (PTS) and each
+// frame is shown for one frame-time in duration. The flags parameter is
+// unused in this example.
+
+// Forced Keyframes
+// ----------------
+// Keyframes can be forced by setting the AOM_EFLAG_FORCE_KF bit of the
+// flags passed to `aom_codec_control()`. In this example, we force a
+// keyframe every <keyframe-interval> frames. Note, the output stream can
+// contain additional keyframes beyond those that have been forced using the
+// AOM_EFLAG_FORCE_KF flag because of automatic keyframe placement by the
+// encoder.
+//
+// Processing The Encoded Data
+// ---------------------------
+// Each packet of type `AOM_CODEC_CX_FRAME_PKT` contains the encoded data
+// for this frame. We write a IVF frame header, followed by the raw data.
+//
+// Cleanup
+// -------
+// The `aom_codec_destroy` call frees any memory allocated by the codec.
+//
+// Error Handling
+// --------------
+// This example does not special case any error return codes. If there was
+// an error, a descriptive message is printed and the program exits. With
+// few exeptions, aom_codec functions return an enumerated error status,
+// with the value `0` indicating success.
+//
+// Error Resiliency Features
+// -------------------------
+// Error resiliency is controlled by the g_error_resilient member of the
+// configuration structure. Use the `decode_with_drops` example to decode with
+// frames 5-10 dropped. Compare the output for a file encoded with this example
+// versus one encoded with the `simple_encoder` example.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_encoder.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr,
+ "Usage: %s <codec> <width> <height> <infile> <outfile> "
+ "<keyframe-interval> <error-resilient> <frames to encode>\n"
+ "See comments in simple_encoder.c for more information.\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static int encode_frame(aom_codec_ctx_t *codec, aom_image_t *img,
+ int frame_index, int flags, AvxVideoWriter *writer) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res =
+ aom_codec_encode(codec, img, frame_index, 1, flags);
+ if (res != AOM_CODEC_OK) die_codec(codec, "Failed to encode frame");
+
+ while ((pkt = aom_codec_get_cx_data(codec, &iter)) != NULL) {
+ got_pkts = 1;
+
+ if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
+ const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
+ if (!aom_video_writer_write_frame(writer, pkt->data.frame.buf,
+ pkt->data.frame.sz,
+ pkt->data.frame.pts)) {
+ die_codec(codec, "Failed to write compressed frame");
+ }
+ printf(keyframe ? "K" : ".");
+ fflush(stdout);
+ }
+ }
+
+ return got_pkts;
+}
+
+// TODO(tomfinegan): Improve command line parsing and add args for bitrate/fps.
+int main(int argc, char **argv) {
+ FILE *infile = NULL;
+ aom_codec_ctx_t codec;
+ aom_codec_enc_cfg_t cfg;
+ int frame_count = 0;
+ aom_image_t raw;
+ aom_codec_err_t res;
+ AvxVideoInfo info;
+ AvxVideoWriter *writer = NULL;
+ const AvxInterface *encoder = NULL;
+ const int fps = 30;
+ const int bitrate = 200;
+ int keyframe_interval = 0;
+ int max_frames = 0;
+ int frames_encoded = 0;
+ const char *codec_arg = NULL;
+ const char *width_arg = NULL;
+ const char *height_arg = NULL;
+ const char *infile_arg = NULL;
+ const char *outfile_arg = NULL;
+ const char *keyframe_interval_arg = NULL;
+
+ exec_name = argv[0];
+
+ // Clear explicitly, as simply assigning "{ 0 }" generates
+ // "missing-field-initializers" warning in some compilers.
+ memset(&info, 0, sizeof(info));
+
+ if (argc != 9) die("Invalid number of arguments");
+
+ codec_arg = argv[1];
+ width_arg = argv[2];
+ height_arg = argv[3];
+ infile_arg = argv[4];
+ outfile_arg = argv[5];
+ keyframe_interval_arg = argv[6];
+ max_frames = (int)strtol(argv[8], NULL, 0);
+
+ encoder = get_aom_encoder_by_name(codec_arg);
+ if (!encoder) die("Unsupported codec.");
+
+ info.codec_fourcc = encoder->fourcc;
+ info.frame_width = (int)strtol(width_arg, NULL, 0);
+ info.frame_height = (int)strtol(height_arg, NULL, 0);
+ info.time_base.numerator = 1;
+ info.time_base.denominator = fps;
+
+ if (info.frame_width <= 0 || info.frame_height <= 0 ||
+ (info.frame_width % 2) != 0 || (info.frame_height % 2) != 0) {
+ die("Invalid frame size: %dx%d", info.frame_width, info.frame_height);
+ }
+
+ if (!aom_img_alloc(&raw, AOM_IMG_FMT_I420, info.frame_width,
+ info.frame_height, 1)) {
+ die("Failed to allocate image.");
+ }
+
+ keyframe_interval = (int)strtol(keyframe_interval_arg, NULL, 0);
+ if (keyframe_interval < 0) die("Invalid keyframe interval value.");
+
+ printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
+
+ res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
+ if (res) die_codec(&codec, "Failed to get default codec config.");
+
+ cfg.g_w = info.frame_width;
+ cfg.g_h = info.frame_height;
+ cfg.g_timebase.num = info.time_base.numerator;
+ cfg.g_timebase.den = info.time_base.denominator;
+ cfg.rc_target_bitrate = bitrate;
+ cfg.g_error_resilient = (aom_codec_er_flags_t)strtoul(argv[7], NULL, 0);
+
+ writer = aom_video_writer_open(outfile_arg, kContainerIVF, &info);
+ if (!writer) die("Failed to open %s for writing.", outfile_arg);
+
+ if (!(infile = fopen(infile_arg, "rb")))
+ die("Failed to open %s for reading.", infile_arg);
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), &cfg, 0))
+ die_codec(&codec, "Failed to initialize encoder");
+
+ // Encode frames.
+ while (aom_img_read(&raw, infile)) {
+ int flags = 0;
+ if (keyframe_interval > 0 && frame_count % keyframe_interval == 0)
+ flags |= AOM_EFLAG_FORCE_KF;
+ encode_frame(&codec, &raw, frame_count++, flags, writer);
+ frames_encoded++;
+ if (max_frames > 0 && frames_encoded >= max_frames) break;
+ }
+
+ // Flush encoder.
+ while (encode_frame(&codec, NULL, -1, 0, writer)) continue;
+
+ printf("\n");
+ fclose(infile);
+ printf("Processed %d frames.\n", frame_count);
+
+ aom_img_free(&raw);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ aom_video_writer_close(writer);
+
+ return EXIT_SUCCESS;
+}
diff --git a/media/libaom/src/examples/twopass_encoder.c b/media/libaom/src/examples/twopass_encoder.c
new file mode 100644
index 000000000..a03bc6cc2
--- /dev/null
+++ b/media/libaom/src/examples/twopass_encoder.c
@@ -0,0 +1,250 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+// Two Pass Encoder
+// ================
+//
+// This is an example of a two pass encoder loop. It takes an input file in
+// YV12 format, passes it through the encoder twice, and writes the compressed
+// frames to disk in IVF format. It builds upon the simple_encoder example.
+//
+// Twopass Variables
+// -----------------
+// Twopass mode needs to track the current pass number and the buffer of
+// statistics packets.
+//
+// Updating The Configuration
+// ---------------------------------
+// In two pass mode, the configuration has to be updated on each pass. The
+// statistics buffer is passed on the last pass.
+//
+// Encoding A Frame
+// ----------------
+// Encoding a frame in two pass mode is identical to the simple encoder
+// example.
+//
+// Processing Statistics Packets
+// -----------------------------
+// Each packet of type `AOM_CODEC_CX_FRAME_PKT` contains the encoded data
+// for this frame. We write a IVF frame header, followed by the raw data.
+//
+//
+// Pass Progress Reporting
+// -----------------------------
+// It's sometimes helpful to see when each pass completes.
+//
+//
+// Clean-up
+// -----------------------------
+// Destruction of the encoder instance must be done on each pass. The
+// raw image should be destroyed at the end as usual.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "aom/aom_encoder.h"
+#include "common/tools_common.h"
+#include "common/video_writer.h"
+
+static const char *exec_name;
+
+void usage_exit(void) {
+ fprintf(stderr,
+ "Usage: %s <codec> <width> <height> <infile> <outfile> "
+ "<limit(optional)>\n",
+ exec_name);
+ exit(EXIT_FAILURE);
+}
+
+static int get_frame_stats(aom_codec_ctx_t *ctx, const aom_image_t *img,
+ aom_codec_pts_t pts, unsigned int duration,
+ aom_enc_frame_flags_t flags,
+ aom_fixed_buf_t *stats) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res = aom_codec_encode(ctx, img, pts, duration, flags);
+ if (res != AOM_CODEC_OK) die_codec(ctx, "Failed to get frame stats.");
+
+ while ((pkt = aom_codec_get_cx_data(ctx, &iter)) != NULL) {
+ got_pkts = 1;
+
+ if (pkt->kind == AOM_CODEC_STATS_PKT) {
+ const uint8_t *const pkt_buf = pkt->data.twopass_stats.buf;
+ const size_t pkt_size = pkt->data.twopass_stats.sz;
+ stats->buf = realloc(stats->buf, stats->sz + pkt_size);
+ memcpy((uint8_t *)stats->buf + stats->sz, pkt_buf, pkt_size);
+ stats->sz += pkt_size;
+ }
+ }
+
+ return got_pkts;
+}
+
+static int encode_frame(aom_codec_ctx_t *ctx, const aom_image_t *img,
+ aom_codec_pts_t pts, unsigned int duration,
+ aom_enc_frame_flags_t flags, AvxVideoWriter *writer) {
+ int got_pkts = 0;
+ aom_codec_iter_t iter = NULL;
+ const aom_codec_cx_pkt_t *pkt = NULL;
+ const aom_codec_err_t res = aom_codec_encode(ctx, img, pts, duration, flags);
+ if (res != AOM_CODEC_OK) die_codec(ctx, "Failed to encode frame.");
+
+ while ((pkt = aom_codec_get_cx_data(ctx, &iter)) != NULL) {
+ got_pkts = 1;
+ if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
+ const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
+
+ if (!aom_video_writer_write_frame(writer, pkt->data.frame.buf,
+ pkt->data.frame.sz,
+ pkt->data.frame.pts))
+ die_codec(ctx, "Failed to write compressed frame.");
+ printf(keyframe ? "K" : ".");
+ fflush(stdout);
+ }
+ }
+
+ return got_pkts;
+}
+
+static aom_fixed_buf_t pass0(aom_image_t *raw, FILE *infile,
+ const AvxInterface *encoder,
+ const aom_codec_enc_cfg_t *cfg, int limit) {
+ aom_codec_ctx_t codec;
+ int frame_count = 0;
+ aom_fixed_buf_t stats = { NULL, 0 };
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), cfg, 0))
+ die_codec(&codec, "Failed to initialize encoder");
+
+ // Calculate frame statistics.
+ while (aom_img_read(raw, infile) && frame_count < limit) {
+ ++frame_count;
+ get_frame_stats(&codec, raw, frame_count, 1, 0, &stats);
+ }
+
+ // Flush encoder.
+ while (get_frame_stats(&codec, NULL, frame_count, 1, 0, &stats)) {
+ }
+
+ printf("Pass 0 complete. Processed %d frames.\n", frame_count);
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ return stats;
+}
+
+static void pass1(aom_image_t *raw, FILE *infile, const char *outfile_name,
+ const AvxInterface *encoder, const aom_codec_enc_cfg_t *cfg,
+ int limit) {
+ AvxVideoInfo info = { encoder->fourcc,
+ cfg->g_w,
+ cfg->g_h,
+ { cfg->g_timebase.num, cfg->g_timebase.den },
+ 0 };
+ AvxVideoWriter *writer = NULL;
+ aom_codec_ctx_t codec;
+ int frame_count = 0;
+
+ writer = aom_video_writer_open(outfile_name, kContainerIVF, &info);
+ if (!writer) die("Failed to open %s for writing", outfile_name);
+
+ if (aom_codec_enc_init(&codec, encoder->codec_interface(), cfg, 0))
+ die_codec(&codec, "Failed to initialize encoder");
+
+ // Encode frames.
+ while (aom_img_read(raw, infile) && frame_count < limit) {
+ ++frame_count;
+ encode_frame(&codec, raw, frame_count, 1, 0, writer);
+ }
+
+ // Flush encoder.
+ while (encode_frame(&codec, NULL, -1, 1, 0, writer)) {
+ }
+
+ printf("\n");
+
+ if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec.");
+
+ aom_video_writer_close(writer);
+
+ printf("Pass 1 complete. Processed %d frames.\n", frame_count);
+}
+
+int main(int argc, char **argv) {
+ FILE *infile = NULL;
+ int w, h;
+ aom_codec_ctx_t codec;
+ aom_codec_enc_cfg_t cfg;
+ aom_image_t raw;
+ aom_codec_err_t res;
+ aom_fixed_buf_t stats;
+
+ const AvxInterface *encoder = NULL;
+ const int fps = 30; // TODO(dkovalev) add command line argument
+ const int bitrate = 200; // kbit/s TODO(dkovalev) add command line argument
+ const char *const codec_arg = argv[1];
+ const char *const width_arg = argv[2];
+ const char *const height_arg = argv[3];
+ const char *const infile_arg = argv[4];
+ const char *const outfile_arg = argv[5];
+ int limit = 0;
+ exec_name = argv[0];
+
+ if (argc < 6) die("Invalid number of arguments");
+
+ if (argc > 6) limit = (int)strtol(argv[6], NULL, 0);
+
+ if (limit == 0) limit = 100;
+
+ encoder = get_aom_encoder_by_name(codec_arg);
+ if (!encoder) die("Unsupported codec.");
+
+ w = (int)strtol(width_arg, NULL, 0);
+ h = (int)strtol(height_arg, NULL, 0);
+
+ if (w <= 0 || h <= 0 || (w % 2) != 0 || (h % 2) != 0)
+ die("Invalid frame size: %dx%d", w, h);
+
+ if (!aom_img_alloc(&raw, AOM_IMG_FMT_I420, w, h, 1))
+ die("Failed to allocate image", w, h);
+
+ printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
+
+ // Configuration
+ res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
+ if (res) die_codec(&codec, "Failed to get default codec config.");
+
+ cfg.g_w = w;
+ cfg.g_h = h;
+ cfg.g_timebase.num = 1;
+ cfg.g_timebase.den = fps;
+ cfg.rc_target_bitrate = bitrate;
+
+ if (!(infile = fopen(infile_arg, "rb")))
+ die("Failed to open %s for reading", infile_arg);
+
+ // Pass 0
+ cfg.g_pass = AOM_RC_FIRST_PASS;
+ stats = pass0(&raw, infile, encoder, &cfg, limit);
+
+ // Pass 1
+ rewind(infile);
+ cfg.g_pass = AOM_RC_LAST_PASS;
+ cfg.rc_twopass_stats_in = stats;
+ pass1(&raw, infile, outfile_arg, encoder, &cfg, limit);
+ free(stats.buf);
+
+ aom_img_free(&raw);
+ fclose(infile);
+
+ return EXIT_SUCCESS;
+}