| Message ID | 608846815.195543.1581952227982.JavaMail.zimbra@fora.si |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [FFmpeg-devel] dvbsub.c, change segment order to be in line with spec | expand |
| Context | Check | Description |
|---|---|---|
| andriy/ffmpeg-patchwork | success | Make fate finished |
> Am 17.02.2020 um 16:10 schrieb Alen Vrečko <alen.vrecko@fora.si>: > > Hello, everyone. > > I'd like to request a change in order of DVB segments to be in line with ETSI 300 743. Basically Region Composition Segment should come before CLUT segment. This caused problems on legacy STBs namely Amino A110. > > In the ETSI EN 300 743 V1.6.1 (2018-10) section 4.4 it is written: > > The order of their listing here matches their ordering, when present, in a display set: > display definition segment ... > page composition segment ... > region composition segment ... > disparity signalling segment ... > CLUT definition segment ... > alternative CLUT segment ... > object data segment ... > end of display set segment ... > > It makes sense for this order to be implemented by ffmpeg. > > I attached a patch, made using git format-patch. This should really include a micro library version bump. Thank you, Carl Eugen
Sure thing. I attached a new patch. Thank you. Alen ----- Original Message ----- From: "Carl Eugen Hoyos" <ceffmpeg@gmail.com> To: "ffmpeg-devel" <ffmpeg-devel@ffmpeg.org> Sent: Monday, February 17, 2020 6:28:17 PM Subject: Re: [FFmpeg-devel] dvbsub.c, change segment order to be in line with spec > Am 17.02.2020 um 16:10 schrieb Alen Vrečko <alen.vrecko@fora.si>: > > Hello, everyone. > > I'd like to request a change in order of DVB segments to be in line with ETSI 300 743. Basically Region Composition Segment should come before CLUT segment. This caused problems on legacy STBs namely Amino A110. > > In the ETSI EN 300 743 V1.6.1 (2018-10) section 4.4 it is written: > > The order of their listing here matches their ordering, when present, in a display set: > display definition segment ... > page composition segment ... > region composition segment ... > disparity signalling segment ... > CLUT definition segment ... > alternative CLUT segment ... > object data segment ... > end of display set segment ... > > It makes sense for this order to be implemented by ffmpeg. > > I attached a patch, made using git format-patch. This should really include a micro library version bump. Thank you, Carl Eugen
On https://patchwork.ffmpeg.org/project/ffmpeg/list/ the new patch doesn't show up. I used the same filename as before. Just in case sending it again with changed filename (and also changed commit message). Sorry for the extra message. Alen ----- Original Message ----- From: "Alen Vrečko" <alen.vrecko@fora.si> To: "ffmpeg-devel" <ffmpeg-devel@ffmpeg.org> Sent: Tuesday, February 18, 2020 11:15:04 AM Subject: Re: [FFmpeg-devel] dvbsub.c, change segment order to be in line with spec Sure thing. I attached a new patch. Thank you. Alen ----- Original Message ----- From: "Carl Eugen Hoyos" <ceffmpeg@gmail.com> To: "ffmpeg-devel" <ffmpeg-devel@ffmpeg.org> Sent: Monday, February 17, 2020 6:28:17 PM Subject: Re: [FFmpeg-devel] dvbsub.c, change segment order to be in line with spec > Am 17.02.2020 um 16:10 schrieb Alen Vrečko <alen.vrecko@fora.si>: > > Hello, everyone. > > I'd like to request a change in order of DVB segments to be in line with ETSI 300 743. Basically Region Composition Segment should come before CLUT segment. This caused problems on legacy STBs namely Amino A110. > > In the ETSI EN 300 743 V1.6.1 (2018-10) section 4.4 it is written: > > The order of their listing here matches their ordering, when present, in a display set: > display definition segment ... > page composition segment ... > region composition segment ... > disparity signalling segment ... > CLUT definition segment ... > alternative CLUT segment ... > object data segment ... > end of display set segment ... > > It makes sense for this order to be implemented by ffmpeg. > > I attached a patch, made using git format-patch. This should really include a micro library version bump. Thank you, Carl Eugen
From eaa9addf0bcab40e17d67c699919e5610b341eab Mon Sep 17 00:00:00 2001 From: Alen Vrecko <alen.vrecko@fora.si> Date: Mon, 17 Feb 2020 15:47:05 +0100 Subject: [PATCH] Change Dvb segment order to be the same as in ETSI EN 300 743 spec. Having CLUT before Region Composition Segment caused problems on legacy STBs i.e. Amino A110. In any case, the spec assumes an order. In section 4.4. It is written: "The order of their listing (referring to segment types) here matches their ordering, when present, in a display set". I think ffmpeg should match the order. --- libavcodec/dvbsub.c | 80 ++++++++++++++++++++++----------------------- 1 file changed, 40 insertions(+), 40 deletions(-) diff --git a/libavcodec/dvbsub.c b/libavcodec/dvbsub.c index a8d43d81d6..e76d3044b8 100644 --- a/libavcodec/dvbsub.c +++ b/libavcodec/dvbsub.c @@ -296,6 +296,46 @@ static int encode_dvb_subtitles(AVCodecContext *avctx, bytestream_put_be16(&pseg_len, q - pseg_len - 2); + for (region_id = 0; region_id < h->num_rects; region_id++) { + + /* region composition segment */ + + if (h->rects[region_id]->nb_colors <= 4) { + /* 2 bpp, some decoders do not support it correctly */ + bpp_index = 0; + } else if (h->rects[region_id]->nb_colors <= 16) { + /* 4 bpp, standard encoding */ + bpp_index = 1; + } else if (h->rects[region_id]->nb_colors <= 256) { + /* 8 bpp, standard encoding */ + bpp_index = 2; + } else { + return -1; + } + + *q++ = 0x0f; /* sync_byte */ + *q++ = 0x11; /* segment_type */ + bytestream_put_be16(&q, page_id); + pseg_len = q; + q += 2; /* segment length */ + *q++ = region_id; + *q++ = (s->object_version << 4) | (0 << 3) | 0x07; /* version , no fill */ + bytestream_put_be16(&q, h->rects[region_id]->w); /* region width */ + bytestream_put_be16(&q, h->rects[region_id]->h); /* region height */ + *q++ = ((1 + bpp_index) << 5) | ((1 + bpp_index) << 2) | 0x03; + *q++ = region_id; /* clut_id == region_id */ + *q++ = 0; /* 8 bit fill colors */ + *q++ = 0x03; /* 4 bit and 2 bit fill colors */ + + bytestream_put_be16(&q, region_id); /* object_id == region_id */ + *q++ = (0 << 6) | (0 << 4); + *q++ = 0; + *q++ = 0xf0; + *q++ = 0; + + bytestream_put_be16(&pseg_len, q - pseg_len - 2); + } + if (h->num_rects) { for (clut_id = 0; clut_id < h->num_rects; clut_id++) { @@ -346,46 +386,6 @@ static int encode_dvb_subtitles(AVCodecContext *avctx, } } - for (region_id = 0; region_id < h->num_rects; region_id++) { - - /* region composition segment */ - - if (h->rects[region_id]->nb_colors <= 4) { - /* 2 bpp, some decoders do not support it correctly */ - bpp_index = 0; - } else if (h->rects[region_id]->nb_colors <= 16) { - /* 4 bpp, standard encoding */ - bpp_index = 1; - } else if (h->rects[region_id]->nb_colors <= 256) { - /* 8 bpp, standard encoding */ - bpp_index = 2; - } else { - return -1; - } - - *q++ = 0x0f; /* sync_byte */ - *q++ = 0x11; /* segment_type */ - bytestream_put_be16(&q, page_id); - pseg_len = q; - q += 2; /* segment length */ - *q++ = region_id; - *q++ = (s->object_version << 4) | (0 << 3) | 0x07; /* version , no fill */ - bytestream_put_be16(&q, h->rects[region_id]->w); /* region width */ - bytestream_put_be16(&q, h->rects[region_id]->h); /* region height */ - *q++ = ((1 + bpp_index) << 5) | ((1 + bpp_index) << 2) | 0x03; - *q++ = region_id; /* clut_id == region_id */ - *q++ = 0; /* 8 bit fill colors */ - *q++ = 0x03; /* 4 bit and 2 bit fill colors */ - - bytestream_put_be16(&q, region_id); /* object_id == region_id */ - *q++ = (0 << 6) | (0 << 4); - *q++ = 0; - *q++ = 0xf0; - *q++ = 0; - - bytestream_put_be16(&pseg_len, q - pseg_len - 2); - } - if (h->num_rects) { for (object_id = 0; object_id < h->num_rects; object_id++) { -- 2.21.1