[FFmpeg-devel] [PATCH] avformat/movenc: Support alpha channel for AVIF

Mon Jun 13 19:30:49 EEST 2022

On Mon, Jun 6, 2022 at 9:29 AM Vignesh Venkatasubramanian
<vigneshv at google.com> wrote:
>
> On Wed, Jun 1, 2022 at 12:47 PM Vignesh Venkatasubramanian
> <vigneshv at google.com> wrote:
> >
> > On Wed, Jun 1, 2022 at 11:52 AM xyesbee <xyesbee at protonmail.com> wrote:
> > >
> > > Why can it not accept it as one but two distinct streams? Why the extra hurdle of alphaextract? YUVA pixel format exists though.
> >
> > Yes, while the YUVA pixel formats exist, AVIF specification requires
> > alpha channel to be a separate track (i.e.) we would need two AV1
> > encoders to make an AVIF image with alpha channel (one for the YUV
> > planes and one for the alpha plane encoded as a monochrome frame). The
> > underlying AV1 codec does not support the YUVA pixel format. So, in
> > order to be able to encode an AVIF image with alpha channel, we would
> > have to separate the alpha channel with the alphaextract filter and
> > pass it to a separate AV1 encoder.
> >
> > If you already have the alpha channel as a separate input, then you
> > can simply pass it along as a second input stream and there will be no
> > need to use the alphaextract filter. I pointed out the use of
> > alphaextract filter simply because it is the simplest way to re-encode
> > existing PNG images with alpha channel into AVIF.
> >
> > An alternative solution is to use the YUVA pixel format and internally
> > use two AV1 encoders in the libaom wrapper. But that requires more
> > changes and does not really make things any better. If you have a
> > better solution, please let me know and I can update this.
> >
> > > -------- Original Message --------
> > > On Jun 1, 2022, 11:35 PM, Vignesh Venkatasubramanian wrote:
> > >
> > > > AVIF specification allows for alpha channel as an auxiliary item (in
> > > > case of still images) or as an auxiliary track (in case of animated
> > > > images). Add support for both of these. The AVIF muxer will take
> > > > exactly two streams (when alpha is present) as input (first one being
> > > > the YUV planes and the second one being the alpha plane).
> > > >
> > > > The input has to come from two different images (one of it color and
> > > > the other one being alpha), or it can come from a single file
> > > > source with the alpha channel extracted using the "alphaextract"
> > > > filter.
> > > >
> > > > Example using alphaextract:
> > > > ffmpeg -i rgba.png -filter_complex "[0:v]alphaextract[a]" -map 0 -map "[a]" -still-picture 1 avif_with_alpha.avif
> > > >
> > > > Example using two sources (first source can be in any pixel format and
> > > > the second source has to be in monochrome grey pixel format):
> > > > ffmpeg -i color.avif -i grey.avif -map 0 -map 1 -c copy avif_with_alpha.avif
> > > >
> > > > The generated files pass the compliance checks in Compliance Warden:
> > > > https://github.com/gpac/ComplianceWarden
> > > >
> > > > libavif (the reference avif library) is able to decode the files
> > > > generated using this patch.
> > > >
> > > > They also play back properly (with transparent background) in:
> > > > 1) Chrome
> > > > 2) Firefox (only still AVIF, no animation support)
> > > >
> > > > Signed-off-by: Vignesh Venkatasubramanian <vigneshv at google.com>
> > > > ---
> > > > libavformat/movenc.c | 188 +++++++++++++++++++++++++++++--------------
> > > > libavformat/movenc.h | 4 +-
> > > > 2 files changed, 130 insertions(+), 62 deletions(-)
> > > >
> > > > diff --git a/libavformat/movenc.c b/libavformat/movenc.c
> > > > index de971f94e8..02eb4d0e9a 100644
> > > > --- a/libavformat/movenc.c
> > > > +++ b/libavformat/movenc.c
> > > > @@ -2852,7 +2852,7 @@ static int mov_write_hdlr_tag(AVFormatContext *s, AVIOContext *pb, MOVTrack *tra
> > > > hdlr = (track->mode == MODE_MOV) ? "mhlr" : "\0\0\0\0";
> > > > if (track->par->codec_type == AVMEDIA_TYPE_VIDEO) {
> > > > if (track->mode == MODE_AVIF) {
> > > > - hdlr_type = "pict";
> > > > + hdlr_type = (track == &mov->tracks[0]) ? "pict" : "auxv";
> > > > descr = "PictureHandler";
> > > > } else {
> > > > hdlr_type = "vide";
> > > > @@ -2940,57 +2940,83 @@ static int mov_write_iloc_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatConte
> > > > avio_wb32(pb, 0); /* Version & flags */
> > > > avio_w8(pb, (4 << 4) + 4); /* offset_size(4) and length_size(4) */
> > > > avio_w8(pb, 0); /* base_offset_size(4) and reserved(4) */
> > > > - avio_wb16(pb, 1); /* item_count */
> > > > + avio_wb16(pb, s->nb_streams); /* item_count */
> > > >
> > > > - avio_wb16(pb, 1); /* item_id */
> > > > - avio_wb16(pb, 0); /* data_reference_index */
> > > > - avio_wb16(pb, 1); /* extent_count */
> > > > - mov->avif_extent_pos = avio_tell(pb);
> > > > - avio_wb32(pb, 0); /* extent_offset (written later) */
> > > > - // For animated AVIF, we simply write the first packet's size.
> > > > - avio_wb32(pb, mov->avif_extent_length); /* extent_length */
> > > > + for (int i = 0; i < s->nb_streams; i++) {
> > > > + avio_wb16(pb, i + 1); /* item_id */
> > > > + avio_wb16(pb, 0); /* data_reference_index */
> > > > + avio_wb16(pb, 1); /* extent_count */
> > > > + mov->avif_extent_pos[i] = avio_tell(pb);
> > > > + avio_wb32(pb, 0); /* extent_offset (written later) */
> > > > + // For animated AVIF, we simply write the first packet's size.
> > > > + avio_wb32(pb, mov->avif_extent_length[i]); /* extent_length */
> > > > + }
> > > >
> > > > return update_size(pb, pos);
> > > > }
> > > >
> > > > static int mov_write_iinf_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatContext *s)
> > > > {
> > > > - int64_t infe_pos;
> > > > int64_t iinf_pos = avio_tell(pb);
> > > > avio_wb32(pb, 0); /* size */
> > > > ffio_wfourcc(pb, "iinf");
> > > > avio_wb32(pb, 0); /* Version & flags */
> > > > - avio_wb16(pb, 1); /* entry_count */
> > > > + avio_wb16(pb, s->nb_streams); /* entry_count */
> > > >
> > > > - infe_pos = avio_tell(pb);
> > > > - avio_wb32(pb, 0); /* size */
> > > > - ffio_wfourcc(pb, "infe");
> > > > - avio_w8(pb, 0x2); /* Version */
> > > > - avio_wb24(pb, 0); /* flags */
> > > > - avio_wb16(pb, 1); /* item_id */
> > > > - avio_wb16(pb, 0); /* item_protection_index */
> > > > - avio_write(pb, "av01", 4); /* item_type */
> > > > - avio_write(pb, "Color\0", 6); /* item_name */
> > > > - update_size(pb, infe_pos);
> > > > + for (int i = 0; i < s->nb_streams; i++) {
> > > > + int64_t infe_pos = avio_tell(pb);
> > > > + avio_wb32(pb, 0); /* size */
> > > > + ffio_wfourcc(pb, "infe");
> > > > + avio_w8(pb, 0x2); /* Version */
> > > > + avio_wb24(pb, 0); /* flags */
> > > > + avio_wb16(pb, i + 1); /* item_id */
> > > > + avio_wb16(pb, 0); /* item_protection_index */
> > > > + avio_write(pb, "av01", 4); /* item_type */
> > > > + avio_write(pb, !i ? "Color\0" : "Alpha\0", 6); /* item_name */
> > > > + update_size(pb, infe_pos);
> > > > + }
> > > >
> > > > return update_size(pb, iinf_pos);
> > > > }
> > > >
> > > > -static int mov_write_ispe_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatContext *s)
> > > > +
> > > > +static int mov_write_iref_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatContext *s)
> > > > +{
> > > > + int64_t auxl_pos;
> > > > + int64_t iref_pos = avio_tell(pb);
> > > > + avio_wb32(pb, 0); /* size */
> > > > + ffio_wfourcc(pb, "iref");
> > > > + avio_wb32(pb, 0); /* Version & flags */
> > > > +
> > > > + auxl_pos = avio_tell(pb);
> > > > + avio_wb32(pb, 0); /* size */
> > > > + ffio_wfourcc(pb, "auxl");
> > > > + avio_wb16(pb, 2); /* from_item_ID */
> > > > + avio_wb16(pb, 1); /* reference_count */
> > > > + avio_wb16(pb, 1); /* to_item_ID */
> > > > + update_size(pb, auxl_pos);
> > > > +
> > > > + return update_size(pb, iref_pos);
> > > > +}
> > > > +
> > > > +static int mov_write_ispe_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatContext *s,
> > > > + int stream_index)
> > > > {
> > > > int64_t pos = avio_tell(pb);
> > > > avio_wb32(pb, 0); /* size */
> > > > ffio_wfourcc(pb, "ispe");
> > > > avio_wb32(pb, 0); /* Version & flags */
> > > > - avio_wb32(pb, s->streams[0]->codecpar->width); /* image_width */
> > > > - avio_wb32(pb, s->streams[0]->codecpar->height); /* image_height */
> > > > + avio_wb32(pb, s->streams[stream_index]->codecpar->width); /* image_width */
> > > > + avio_wb32(pb, s->streams[stream_index]->codecpar->height); /* image_height */
> > > > return update_size(pb, pos);
> > > > }
> > > >
> > > > -static int mov_write_pixi_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatContext *s)
> > > > +static int mov_write_pixi_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatContext *s,
> > > > + int stream_index)
> > > > {
> > > > int64_t pos = avio_tell(pb);
> > > > - const AVPixFmtDescriptor *pixdesc = av_pix_fmt_desc_get(s->streams[0]->codecpar->format);
> > > > + const AVPixFmtDescriptor *pixdesc =
> > > > + av_pix_fmt_desc_get(s->streams[stream_index]->codecpar->format);
> > > > avio_wb32(pb, 0); /* size */
> > > > ffio_wfourcc(pb, "pixi");
> > > > avio_wb32(pb, 0); /* Version & flags */
> > > > @@ -3001,15 +3027,30 @@ static int mov_write_pixi_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatConte
> > > > return update_size(pb, pos);
> > > > }
> > > >
> > > > +static int mov_write_auxC_tag(AVIOContext *pb)
> > > > +{
> > > > + int64_t pos = avio_tell(pb);
> > > > + avio_wb32(pb, 0); /* size */
> > > > + ffio_wfourcc(pb, "auxC");
> > > > + avio_wb32(pb, 0); /* Version & flags */
> > > > + avio_write(pb, "urn:mpeg:mpegB:cicp:systems:auxiliary:alpha\0", 44);
> > > > + return update_size(pb, pos);
> > > > +}
> > > > +
> > > > static int mov_write_ipco_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatContext *s)
> > > > {
> > > > int64_t pos = avio_tell(pb);
> > > > avio_wb32(pb, 0); /* size */
> > > > ffio_wfourcc(pb, "ipco");
> > > > - mov_write_ispe_tag(pb, mov, s);
> > > > - mov_write_pixi_tag(pb, mov, s);
> > > > - mov_write_av1c_tag(pb, &mov->tracks[0]);
> > > > - mov_write_colr_tag(pb, &mov->tracks[0], 0);
> > > > + for (int i = 0; i < s->nb_streams; i++) {
> > > > + mov_write_ispe_tag(pb, mov, s, i);
> > > > + mov_write_pixi_tag(pb, mov, s, i);
> > > > + mov_write_av1c_tag(pb, &mov->tracks[i]);
> > > > + if (!i)
> > > > + mov_write_colr_tag(pb, &mov->tracks[0], 0);
> > > > + else
> > > > + mov_write_auxC_tag(pb);
> > > > + }
> > > > return update_size(pb, pos);
> > > > }
> > > >
> > > > @@ -3019,18 +3060,21 @@ static int mov_write_ipma_tag(AVIOContext *pb, MOVMuxContext *mov, AVFormatConte
> > > > avio_wb32(pb, 0); /* size */
> > > > ffio_wfourcc(pb, "ipma");
> > > > avio_wb32(pb, 0); /* Version & flags */
> > > > - avio_wb32(pb, 1); /* entry_count */
> > > > - avio_wb16(pb, 1); /* item_ID */
> > > > - avio_w8(pb, 4); /* association_count */
> > > > -
> > > > - // ispe association.
> > > > - avio_w8(pb, 1); /* essential and property_index */
> > > > - // pixi association.
> > > > - avio_w8(pb, 2); /* essential and property_index */
> > > > - // av1C association.
> > > > - avio_w8(pb, 0x80 | 3); /* essential and property_index */
> > > > - // colr association.
> > > > - avio_w8(pb, 4); /* essential and property_index */
> > > > + avio_wb32(pb, s->nb_streams); /* entry_count */
> > > > +
> > > > + for (int i = 0, index = 1; i < s->nb_streams; i++) {
> > > > + avio_wb16(pb, i + 1); /* item_ID */
> > > > + avio_w8(pb, 4); /* association_count */
> > > > +
> > > > + // ispe association.
> > > > + avio_w8(pb, index++); /* essential and property_index */
> > > > + // pixi association.
> > > > + avio_w8(pb, index++); /* essential and property_index */
> > > > + // av1C association.
> > > > + avio_w8(pb, 0x80 | index++); /* essential and property_index */
> > > > + // colr/auxC association.
> > > > + avio_w8(pb, index++); /* essential and property_index */
> > > > + }
> > > > return update_size(pb, pos);
> > > > }
> > > >
> > > > @@ -4112,6 +4156,8 @@ static int mov_write_meta_tag(AVIOContext *pb, MOVMuxContext *mov,
> > > > mov_write_pitm_tag(pb, 1);
> > > > mov_write_iloc_tag(pb, mov, s);
> > > > mov_write_iinf_tag(pb, mov, s);
> > > > + if (s->nb_streams > 1)
> > > > + mov_write_iref_tag(pb, mov, s);
> > > > mov_write_iprp_tag(pb, mov, s);
> > > > } else {
> > > > /* iTunes metadata tag */
> > > > @@ -6040,8 +6086,8 @@ int ff_mov_write_packet(AVFormatContext *s, AVPacket *pkt)
> > > > avio_write(pb, reformatted_data, size);
> > > > } else {
> > > > size = ff_av1_filter_obus(pb, pkt->data, pkt->size);
> > > > - if (trk->mode == MODE_AVIF && !mov->avif_extent_length) {
> > > > - mov->avif_extent_length = size;
> > > > + if (trk->mode == MODE_AVIF && !mov->avif_extent_length[pkt->stream_index]) {
> > > > + mov->avif_extent_length[pkt->stream_index] = size;
> > > > }
> > > > }
> > > >
> > > > @@ -6872,16 +6918,26 @@ static int mov_init(AVFormatContext *s)
> > > > return AVERROR(EINVAL);
> > > > }
> > > >
> > > > - /* AVIF output must have exactly one video stream */
> > > > + /* AVIF output must have at most two video streams (one for YUV and one for
> > > > + * alpha). */
> > > > if (mov->mode == MODE_AVIF) {
> > > > - if (s->nb_streams > 1) {
> > > > - av_log(s, AV_LOG_ERROR, "AVIF output requires exactly one stream\n");
> > > > + if (s->nb_streams > 2) {
> > > > + av_log(s, AV_LOG_ERROR, "AVIF output requires exactly one or two streams\n");
> > > > return AVERROR(EINVAL);
> > > > }
> > > > - if (s->streams[0]->codecpar->codec_type != AVMEDIA_TYPE_VIDEO) {
> > > > - av_log(s, AV_LOG_ERROR, "AVIF output requires one video stream\n");
> > > > + if (s->streams[0]->codecpar->codec_type != AVMEDIA_TYPE_VIDEO &&
> > > > + (s->nb_streams > 1 && s->streams[1]->codecpar->codec_type != AVMEDIA_TYPE_VIDEO)) {
> > > > + av_log(s, AV_LOG_ERROR, "AVIF output supports only video streams\n");
> > > > return AVERROR(EINVAL);
> > > > }
> > > > + if (s->nb_streams > 1) {
> > > > + const AVPixFmtDescriptor *pixdesc =
> > > > + av_pix_fmt_desc_get(s->streams[1]->codecpar->format);
> > > > + if (pixdesc->nb_components != 1) {
> > > > + av_log(s, AV_LOG_ERROR, "Second stream for AVIF (alpha) output must have exactly one plane\n");
> > > > + return AVERROR(EINVAL);
> > > > + }
> > > > + }
> > > > s->streams[0]->disposition |= AV_DISPOSITION_DEFAULT;
> > > > }
> > > >
> > > > @@ -7543,18 +7599,25 @@ static int avif_write_trailer(AVFormatContext *s)
> > > > {
> > > > AVIOContext *pb = s->pb;
> > > > MOVMuxContext *mov = s->priv_data;
> > > > - int64_t pos_backup, mdat_pos;
> > > > + int64_t pos_backup, extent_offsets[2];
> > > > uint8_t *buf;
> > > > - int buf_size, moov_size;
> > > > + int buf_size, moov_size, i;
> > > >
> > > > if (mov->moov_written) return 0;
> > > >
> > > > mov->is_animated_avif = s->streams[0]->nb_frames > 1;
> > > > + if (mov->is_animated_avif && s->nb_streams > 1) {
> > > > + // For animated avif with alpha channel, we need to write a tref tag
> > > > + // with type "auxl".
> > > > + mov->tracks[1].tref_tag = MKTAG('a', 'u', 'x', 'l');
> > > > + mov->tracks[1].tref_id = 1;
> > > > + }
> > > > mov_write_identification(pb, s);
> > > > mov_write_meta_tag(pb, mov, s);
> > > >
> > > > moov_size = get_moov_size(s);
> > > > - mov->tracks[0].data_offset = avio_tell(pb) + moov_size + 8;
> > > > + for (i = 0; i < s->nb_streams; i++)
> > > > + mov->tracks[i].data_offset = avio_tell(pb) + moov_size + 8;
> > > >
> > > > if (mov->is_animated_avif) {
> > > > int ret;
> > > > @@ -7565,19 +7628,24 @@ static int avif_write_trailer(AVFormatContext *s)
> > > > buf_size = avio_get_dyn_buf(mov->mdat_buf, &buf);
> > > > avio_wb32(pb, buf_size + 8);
> > > > ffio_wfourcc(pb, "mdat");
> > > > - mdat_pos = avio_tell(pb);
> > > >
> > > > - if (mdat_pos != (uint32_t)mdat_pos) {
> > > > - av_log(s, AV_LOG_ERROR, "mdat offset does not fit in 32 bits\n");
> > > > - return AVERROR_INVALIDDATA;
> > > > - }
> > > > + // The offset for the YUV planes is the starting position of mdat.
> > > > + extent_offsets[0] = avio_tell(pb);
> > > > + // The offset for alpha plane is YUV offset + YUV size.
> > > > + extent_offsets[1] = extent_offsets[0] + mov->avif_extent_length[0];
> > > >
> > > > avio_write(pb, buf, buf_size);
> > > >
> > > > - // write extent offset.
> > > > + // write extent offsets.
> > > > pos_backup = avio_tell(pb);
> > > > - avio_seek(pb, mov->avif_extent_pos, SEEK_SET);
> > > > - avio_wb32(pb, mdat_pos); /* rewrite offset */
> > > > + for (i = 0; i < s->nb_streams; i++) {
> > > > + if (extent_offsets[i] != (uint32_t)extent_offsets[i]) {
> > > > + av_log(s, AV_LOG_ERROR, "extent offset does not fit in 32 bits\n");
> > > > + return AVERROR_INVALIDDATA;
> > > > + }
> > > > + avio_seek(pb, mov->avif_extent_pos[i], SEEK_SET);
> > > > + avio_wb32(pb, extent_offsets[i]); /* rewrite offset */
> > > > + }
> > > > avio_seek(pb, pos_backup, SEEK_SET);
> > > >
> > > > return 0;
> > > > diff --git a/libavformat/movenc.h b/libavformat/movenc.h
> > > > index 281576cc66..e4550f7900 100644
> > > > --- a/libavformat/movenc.h
> > > > +++ b/libavformat/movenc.h
> > > > @@ -246,8 +246,8 @@ typedef struct MOVMuxContext {
> > > > int empty_hdlr_name;
> > > > int movie_timescale;
> > > >
> > > > - int64_t avif_extent_pos;
> > > > - int avif_extent_length;
> > > > + int64_t avif_extent_pos[2]; // index 0 is YUV and 1 is Alpha.
> > > > + int avif_extent_length[2]; // index 0 is YUV and 1 is Alpha.
> > > > int is_animated_avif;
> > > > } MOVMuxContext;
> > > >
> > > > --
> > > > 2.36.1.255.ge46751e96f-goog
> > > >
> > > > _______________________________________________
> > > > ffmpeg-devel mailing list
> > > > ffmpeg-devel at ffmpeg.org
> > > > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> > > >
> > > > To unsubscribe, visit link above, or email
> > > > ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
> > > _______________________________________________
> > > ffmpeg-devel mailing list
> > > ffmpeg-devel at ffmpeg.org
> > > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> > >
> > > To unsubscribe, visit link above, or email
> > > ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
> >
> >
> >
> > --
> > Vignesh
>
> Any comments on this one? If not can this be merged please? :)
>

Another ping on this please?

> --
> Vignesh

--
Vignesh