[FFmpeg-cvslog] fftools/ffmpeg: support applying container level cropping
James Almer
git at videolan.org
Mon Jul 8 19:31:10 EEST 2024
ffmpeg | branch: master | James Almer <jamrial at gmail.com> | Tue May 28 16:37:44 2024 -0300| [bf87688a9fb8afdd9fd9150fb6c7d2b4ca10f48c] | committer: James Almer
fftools/ffmpeg: support applying container level cropping
Signed-off-by: James Almer <jamrial at gmail.com>
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=bf87688a9fb8afdd9fd9150fb6c7d2b4ca10f48c
---
doc/ffmpeg.texi | 15 +++++++++++++++
fftools/ffmpeg.h | 14 ++++++++++++++
fftools/ffmpeg_demux.c | 43 +++++++++++++++++++++++++++++++++++++++++++
fftools/ffmpeg_filter.c | 11 +++++++++++
fftools/ffmpeg_opt.c | 3 +++
5 files changed, 86 insertions(+)
diff --git a/doc/ffmpeg.texi b/doc/ffmpeg.texi
index f25f6192eb..877edae3b3 100644
--- a/doc/ffmpeg.texi
+++ b/doc/ffmpeg.texi
@@ -1379,6 +1379,21 @@ Note that forcing too many keyframes is very harmful for the lookahead
algorithms of certain encoders: using fixed-GOP options or similar
would be more efficient.
+ at item -apply_cropping[:@var{stream_specifier}] @var{source} (@emph{input,per-stream})
+Automatically crop the video after decoding according to file metadata.
+Default is @emph{all}.
+
+ at table @option
+ at item none (0)
+Don't apply any cropping metadata.
+ at item all (1)
+Apply both codec and container level croppping. This is the default mode.
+ at item codec (2)
+Apply codec level croppping.
+ at item container (3)
+Apply container level croppping.
+ at end table
+
@item -copyinkf[:@var{stream_specifier}] (@emph{output,per-stream})
When doing stream copy, copy also non-key frames found at the
beginning.
diff --git a/fftools/ffmpeg.h b/fftools/ffmpeg.h
index d173475621..d0298d53cf 100644
--- a/fftools/ffmpeg.h
+++ b/fftools/ffmpeg.h
@@ -155,6 +155,7 @@ typedef struct OptionsContext {
SpecifierOptList hwaccel_devices;
SpecifierOptList hwaccel_output_formats;
SpecifierOptList autorotate;
+ SpecifierOptList apply_cropping;
/* output options */
StreamMap *stream_maps;
@@ -239,6 +240,7 @@ enum IFilterFlags {
IFILTER_FLAG_AUTOROTATE = (1 << 0),
IFILTER_FLAG_REINIT = (1 << 1),
IFILTER_FLAG_CFR = (1 << 2),
+ IFILTER_FLAG_CROP = (1 << 3),
};
typedef struct InputFilterOptions {
@@ -254,6 +256,11 @@ typedef struct InputFilterOptions {
* accurate */
AVRational framerate;
+ unsigned crop_top;
+ unsigned crop_bottom;
+ unsigned crop_left;
+ unsigned crop_right;
+
int sub2video_width;
int sub2video_height;
@@ -539,6 +546,13 @@ typedef struct KeyframeForceCtx {
typedef struct Encoder Encoder;
+enum CroppingType {
+ CROP_DISABLED = 0,
+ CROP_ALL,
+ CROP_CODEC,
+ CROP_CONTAINER,
+};
+
typedef struct OutputStream {
const AVClass *class;
diff --git a/fftools/ffmpeg_demux.c b/fftools/ffmpeg_demux.c
index 3762d589e3..528e02149f 100644
--- a/fftools/ffmpeg_demux.c
+++ b/fftools/ffmpeg_demux.c
@@ -66,6 +66,7 @@ typedef struct DemuxStream {
int have_sub2video;
int reinit_filters;
int autorotate;
+ int apply_cropping;
int wrap_correction_done;
@@ -1000,11 +1001,22 @@ int ist_filter_add(InputStream *ist, InputFilter *ifilter, int is_simple,
ist->filters[ist->nb_filters - 1] = ifilter;
if (ist->par->codec_type == AVMEDIA_TYPE_VIDEO) {
+ const AVPacketSideData *sd = av_packet_side_data_get(ist->par->coded_side_data,
+ ist->par->nb_coded_side_data,
+ AV_PKT_DATA_FRAME_CROPPING);
if (ist->framerate.num > 0 && ist->framerate.den > 0) {
opts->framerate = ist->framerate;
opts->flags |= IFILTER_FLAG_CFR;
} else
opts->framerate = av_guess_frame_rate(d->f.ctx, ist->st, NULL);
+ if (sd && sd->size >= sizeof(uint32_t) * 4) {
+ opts->crop_top = AV_RL32(sd->data + 0);
+ opts->crop_bottom = AV_RL32(sd->data + 4);
+ opts->crop_left = AV_RL32(sd->data + 8);
+ opts->crop_right = AV_RL32(sd->data + 12);
+ if (ds->apply_cropping && ds->apply_cropping != CROP_CODEC)
+ opts->flags |= IFILTER_FLAG_CROP;
+ }
} else if (ist->par->codec_type == AVMEDIA_TYPE_SUBTITLE) {
/* Compute the size of the canvas for the subtitles stream.
If the subtitles codecpar has set a size, use it. Otherwise use the
@@ -1215,6 +1227,7 @@ static int ist_add(const OptionsContext *o, Demuxer *d, AVStream *st, AVDictiona
InputStream *ist;
char *framerate = NULL, *hwaccel_device = NULL;
const char *hwaccel = NULL;
+ const char *apply_cropping = NULL;
char *hwaccel_output_format = NULL;
char *codec_tag = NULL;
char *bsfs = NULL;
@@ -1241,6 +1254,33 @@ static int ist_add(const OptionsContext *o, Demuxer *d, AVStream *st, AVDictiona
ds->autorotate = 1;
MATCH_PER_STREAM_OPT(autorotate, i, ds->autorotate, ic, st);
+ ds->apply_cropping = CROP_ALL;
+ MATCH_PER_STREAM_OPT(apply_cropping, str, apply_cropping, ic, st);
+ if (apply_cropping) {
+ const AVOption opts[] = {
+ { "apply_cropping", NULL, 0, AV_OPT_TYPE_INT,
+ { .i64 = CROP_ALL }, CROP_DISABLED, CROP_CONTAINER, AV_OPT_FLAG_DECODING_PARAM, .unit = "apply_cropping" },
+ { "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CROP_DISABLED }, .unit = "apply_cropping" },
+ { "all", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CROP_ALL }, .unit = "apply_cropping" },
+ { "codec", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CROP_CODEC }, .unit = "apply_cropping" },
+ { "container", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CROP_CONTAINER }, .unit = "apply_cropping" },
+ { NULL },
+ };
+ const AVClass class = {
+ .class_name = "apply_cropping",
+ .item_name = av_default_item_name,
+ .option = opts,
+ .version = LIBAVUTIL_VERSION_INT,
+ };
+ const AVClass *pclass = &class;
+
+ ret = av_opt_eval_int(&pclass, opts, apply_cropping, &ds->apply_cropping);
+ if (ret < 0) {
+ av_log(ist, AV_LOG_ERROR, "Invalid apply_cropping value '%s'.\n", apply_cropping);
+ return ret;
+ }
+ }
+
MATCH_PER_STREAM_OPT(codec_tags, str, codec_tag, ic, st);
if (codec_tag) {
uint32_t tag = strtol(codec_tag, &next, 0);
@@ -1362,6 +1402,9 @@ static int ist_add(const OptionsContext *o, Demuxer *d, AVStream *st, AVDictiona
ds->dec_opts.flags |= DECODER_FLAG_BITEXACT * !!o->bitexact;
+ av_dict_set_int(&ds->decoder_opts, "apply_cropping",
+ ds->apply_cropping && ds->apply_cropping != CROP_CONTAINER, 0);
+
/* Attached pics are sparse, therefore we would not want to delay their decoding
* till EOF. */
if (ist->st->disposition & AV_DISPOSITION_ATTACHED_PIC)
diff --git a/fftools/ffmpeg_filter.c b/fftools/ffmpeg_filter.c
index 12cca684b4..097bd2ed48 100644
--- a/fftools/ffmpeg_filter.c
+++ b/fftools/ffmpeg_filter.c
@@ -1701,6 +1701,17 @@ static int configure_input_video_filter(FilterGraph *fg, AVFilterGraph *graph,
desc = av_pix_fmt_desc_get(ifp->format);
av_assert0(desc);
+ if ((ifp->opts.flags & IFILTER_FLAG_CROP)) {
+ char crop_buf[64];
+ snprintf(crop_buf, sizeof(crop_buf), "w=iw-%d-%d:h=ih-%d-%d:x=%d:y=%d",
+ ifp->opts.crop_left, ifp->opts.crop_right,
+ ifp->opts.crop_top, ifp->opts.crop_bottom,
+ ifp->opts.crop_left, ifp->opts.crop_top);
+ ret = insert_filter(&last_filter, &pad_idx, "crop", crop_buf);
+ if (ret < 0)
+ return ret;
+ }
+
// TODO: insert hwaccel enabled filters like transpose_vaapi into the graph
ifp->displaymatrix_applied = 0;
if ((ifp->opts.flags & IFILTER_FLAG_AUTOROTATE) &&
diff --git a/fftools/ffmpeg_opt.c b/fftools/ffmpeg_opt.c
index 97585e4dfa..2c201c74b2 100644
--- a/fftools/ffmpeg_opt.c
+++ b/fftools/ffmpeg_opt.c
@@ -1711,6 +1711,9 @@ const OptionDef options[] = {
{ "autoscale", OPT_TYPE_BOOL, OPT_VIDEO | OPT_PERSTREAM | OPT_EXPERT | OPT_OUTPUT,
{ .off = OFFSET(autoscale) },
"automatically insert a scale filter at the end of the filter graph" },
+ { "apply_cropping", OPT_TYPE_STRING, OPT_VIDEO | OPT_PERSTREAM | OPT_EXPERT | OPT_INPUT,
+ { .off = OFFSET(apply_cropping) },
+ "select the cropping to apply" },
{ "fix_sub_duration_heartbeat", OPT_TYPE_BOOL, OPT_VIDEO | OPT_EXPERT | OPT_PERSTREAM | OPT_OUTPUT,
{ .off = OFFSET(fix_sub_duration_heartbeat) },
"set this video output stream to be a heartbeat stream for "
More information about the ffmpeg-cvslog
mailing list