From 21520e59d165b9f614d312db4114053d06d04750 Mon Sep 17 00:00:00 2001 From: Terra Packaging Team Date: Sun, 31 May 2026 03:05:02 -0500 Subject: [PATCH] Patch for HandBrake Patch based off a very similar patch by Negativo17 but formatted as a Git patch. See: https://github.com/HandBrake/HandBrake/tree/2f464fcf93d411ebdd969b39d67739ed658c5e58 --- libavcodec/amfenc.c | 4 - libavcodec/av1dec.c | 14 + libavcodec/ccaption_dec.c | 9 +- libavcodec/dvdsubdec.c | 42 ++- libavcodec/libdav1d.c | 14 + libavcodec/mfenc.c | 13 +- libavcodec/nvenc.c | 4 +- libavcodec/pgssubdec.c | 554 ++++++++++++++++++----------- libavcodec/qsvenc.c | 49 +++ libavcodec/qsvenc.h | 8 +- libavcodec/qsvenc_av1.c | 4 + libavcodec/videotoolbox.c | 22 +- libavfilter/vf_vpp_qsv.c | 6 +- libavformat/isom.h | 3 + libavformat/matroskaenc.c | 10 + libavformat/mov.c | 204 ++++++++++- libavformat/movenc.c | 231 +++++++++--- libavutil/frame.h | 6 + libavutil/hwcontext_d3d11va.c | 45 ++- libavutil/hwcontext_videotoolbox.c | 12 +- libavutil/side_data.c | 1 + 21 files changed, 963 insertions(+), 292 deletions(-) diff --git a/libavcodec/amfenc.c b/libavcodec/amfenc.c index 329ce29..86fa5ec 100644 --- a/libavcodec/amfenc.c +++ b/libavcodec/amfenc.c @@ -173,10 +173,6 @@ static int amf_init_encoder(AVCodecContext *avctx) else pix_fmt = avctx->pix_fmt; - if (pix_fmt == AV_PIX_FMT_P010) { - AMF_RETURN_IF_FALSE(ctx, amf_device_ctx->version >= AMF_MAKE_FULL_VERSION(1, 4, 32, 0), AVERROR_UNKNOWN, "10-bit encoder is not supported by AMD GPU drivers versions lower than 23.30.\n"); - } - ctx->format = av_av_to_amf_format(pix_fmt); AMF_RETURN_IF_FALSE(ctx, ctx->format != AMF_SURFACE_UNKNOWN, AVERROR(EINVAL), "Format %s is not supported\n", av_get_pix_fmt_name(pix_fmt)); diff --git a/libavcodec/av1dec.c b/libavcodec/av1dec.c index 2b3a36a..5dc813d 100644 --- a/libavcodec/av1dec.c +++ b/libavcodec/av1dec.c @@ -1019,6 +1019,8 @@ FF_ENABLE_DEPRECATION_WARNINGS if (ret < 0) return ret; } else if (country_code == ITU_T_T35_COUNTRY_CODE_US && provider_code == ITU_T_T35_PROVIDER_CODE_DOLBY) { + AVBufferRef *rpu_buf; + AVFrameSideData *rpu; int provider_oriented_code = bytestream2_get_be32(&gb); if (provider_oriented_code != 0x800) return 0; // ignore @@ -1030,6 +1032,18 @@ FF_ENABLE_DEPRECATION_WARNINGS return 0; // ignore } + rpu_buf = av_buffer_alloc(itut_t35->payload_size); + if (rpu_buf) { + memcpy(rpu_buf->data, itut_t35->payload, itut_t35->payload_size); + rpu = av_frame_new_side_data_from_buf(frame, AV_FRAME_DATA_DOVI_RPU_BUFFER_T35, rpu_buf); + if (!rpu) { + av_buffer_unref(&rpu_buf); + return AVERROR(ENOMEM); + } + } else { + return AVERROR(ENOMEM); + } + ret = ff_dovi_attach_side_data(&s->dovi, frame); if (ret < 0) return ret; diff --git a/libavcodec/ccaption_dec.c b/libavcodec/ccaption_dec.c index 9b6c766..399296e 100644 --- a/libavcodec/ccaption_dec.c +++ b/libavcodec/ccaption_dec.c @@ -893,12 +893,13 @@ static int decode(AVCodecContext *avctx, AVSubtitle *sub, if (ctx->buffer[bidx].str[0] || ctx->real_time) { ff_dlog(avctx, "cdp writing data (%s)\n", ctx->buffer[bidx].str); - start_time = ctx->buffer_time[0]; - sub->pts = start_time; - end_time = ctx->buffer_time[1]; - if (!ctx->real_time) + if (!ctx->real_time) { + start_time = ctx->buffer_time[0]; + sub->pts = start_time; + end_time = ctx->buffer_time[1]; sub->end_display_time = av_rescale_q(end_time - start_time, AV_TIME_BASE_Q, ms_tb); + } else sub->end_display_time = -1; ret = ff_ass_add_rect2(sub, ctx->buffer[bidx].str, ctx->readorder++, 0, NULL, NULL, &nb_rect_allocated); diff --git a/libavcodec/dvdsubdec.c b/libavcodec/dvdsubdec.c index 9f9845b..1beb1fc 100644 --- a/libavcodec/dvdsubdec.c +++ b/libavcodec/dvdsubdec.c @@ -45,6 +45,8 @@ typedef struct DVDSubContext int buf_size; int forced_subs_only; uint8_t used_color[256]; + int64_t pts; + int output_empty_rects; } DVDSubContext; static void yuv_a_to_rgba(const uint8_t *ycbcr, const uint8_t *alpha, uint32_t *rgba, int num_values) @@ -230,7 +232,10 @@ static int decode_dvd_subtitles(void *logctx, DVDSubContext *ctx, uint32_t size; int64_t offset1, offset2; - if (buf_size < 10) + if (buf_size < 2) + return AVERROR(EAGAIN); + + if (buf_size == 2 && AV_RB16(buf) == 0) return -1; if (AV_RB16(buf) == 0) { /* HD subpicture with 4-byte offsets */ @@ -243,15 +248,21 @@ static int decode_dvd_subtitles(void *logctx, DVDSubContext *ctx, cmd_pos = 2; } + if (big_offsets && buf_size < 6) + return AVERROR(EAGAIN); + size = READ_OFFSET(buf + (big_offsets ? 2 : 0)); - cmd_pos = READ_OFFSET(buf + cmd_pos); + if (size == 0) + return -1; - if (cmd_pos < 0 || cmd_pos > buf_size - 2 - offset_size) { - if (cmd_pos > size) { - av_log(logctx, AV_LOG_ERROR, "Discarding invalid packet\n"); - return 0; - } + if (buf_size < size) return AVERROR(EAGAIN); + + cmd_pos = READ_OFFSET(buf + cmd_pos); + + if (cmd_pos < 0 || cmd_pos > size) { + av_log(ctx, AV_LOG_ERROR, "Discarding invalid packet\n"); + return AVERROR_INVALIDDATA; } while (cmd_pos > 0 && cmd_pos < buf_size - 2 - offset_size) { @@ -524,10 +535,13 @@ static int dvdsub_decode(AVCodecContext *avctx, AVSubtitle *sub, int appended = 0; int is_menu; + if (ctx->pts == AV_NOPTS_VALUE && sub->pts != AV_NOPTS_VALUE) + ctx->pts = sub->pts; if (ctx->buf_size) { int ret = append_to_cached_buf(avctx, buf, buf_size); if (ret < 0) { *data_size = 0; + ctx->pts = AV_NOPTS_VALUE; return ret; } buf = ctx->buf; @@ -538,7 +552,12 @@ static int dvdsub_decode(AVCodecContext *avctx, AVSubtitle *sub, is_menu = decode_dvd_subtitles(avctx, ctx, sub, buf, buf_size); if (is_menu == AVERROR(EAGAIN)) { *data_size = 0; - return appended ? 0 : append_to_cached_buf(avctx, buf, buf_size); + int ret = appended ? 0 : append_to_cached_buf(avctx, buf, buf_size); + if (ret < 0) { + ctx->pts = AV_NOPTS_VALUE; + return ret; + } + return buf_size; } if (is_menu < 0) { @@ -547,9 +566,10 @@ static int dvdsub_decode(AVCodecContext *avctx, AVSubtitle *sub, reset_rects(sub); *data_size = 0; + ctx->pts = AV_NOPTS_VALUE; return buf_size; } - if (!is_menu && find_smallest_bounding_rectangle(ctx, sub) == 0) + if (!is_menu && !ctx->output_empty_rects && find_smallest_bounding_rectangle(ctx, sub) == 0) goto no_subtitle; if (ctx->forced_subs_only && !(sub->rects[0]->flags & AV_SUBTITLE_FLAG_FORCED)) @@ -557,6 +577,8 @@ static int dvdsub_decode(AVCodecContext *avctx, AVSubtitle *sub, ctx->buf_size = 0; *data_size = 1; + sub->pts = ctx->pts; + ctx->pts = AV_NOPTS_VALUE; return buf_size; } @@ -683,6 +705,7 @@ static av_cold int dvdsub_init(AVCodecContext *avctx) av_log(avctx, AV_LOG_DEBUG, " 0x%06"PRIx32, ctx->palette[i]); av_log(avctx, AV_LOG_DEBUG, "\n"); } + ctx->pts = AV_NOPTS_VALUE; return 0; } @@ -699,6 +722,7 @@ static const AVOption options[] = { { "palette", "set the global palette", OFFSET(palette_str), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, SD }, { "ifo_palette", "obtain the global palette from .IFO file", OFFSET(ifo_str), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, SD }, { "forced_subs_only", "Only show forced subtitles", OFFSET(forced_subs_only), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, SD}, + { "output_empty_rects", "Output subtitles with empty or fully transparent rects", OFFSET(output_empty_rects), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, SD}, { NULL } }; static const AVClass dvdsub_class = { diff --git a/libavcodec/libdav1d.c b/libavcodec/libdav1d.c index 14ec984..90aa2ba 100644 --- a/libavcodec/libdav1d.c +++ b/libavcodec/libdav1d.c @@ -415,6 +415,8 @@ FF_ENABLE_DEPRECATION_WARNINGS if (res < 0) return res; } else if (country_code == ITU_T_T35_COUNTRY_CODE_US && provider_code == ITU_T_T35_PROVIDER_CODE_DOLBY) { + AVBufferRef *rpu_buf; + AVFrameSideData *rpu; int provider_oriented_code = bytestream2_get_be32(&gb); if (provider_oriented_code != 0x800) return 0; // ignore @@ -426,6 +428,18 @@ FF_ENABLE_DEPRECATION_WARNINGS return 0; // ignore } + rpu_buf = av_buffer_alloc(itut_t35->payload_size); + if (rpu_buf) { + memcpy(rpu_buf->data, itut_t35->payload, itut_t35->payload_size); + rpu = av_frame_new_side_data_from_buf(frame, AV_FRAME_DATA_DOVI_RPU_BUFFER_T35, rpu_buf); + if (!rpu) { + av_buffer_unref(&rpu_buf); + return AVERROR(ENOMEM); + } + } else { + return AVERROR(ENOMEM); + } + res = ff_dovi_attach_side_data(&dav1d->dovi, frame); if (res < 0) return res; diff --git a/libavcodec/mfenc.c b/libavcodec/mfenc.c index 541f7fb..b517352 100644 --- a/libavcodec/mfenc.c +++ b/libavcodec/mfenc.c @@ -904,6 +904,8 @@ static int64_t mf_encv_input_score(AVCodecContext *avctx, IMFMediaType *type) static int mf_encv_input_adjust(AVCodecContext *avctx, IMFMediaType *type) { enum AVPixelFormat pix_fmt = ff_media_type_to_pix_fmt((IMFAttributes *)type); + AVRational framerate; + if (avctx->pix_fmt == AV_PIX_FMT_D3D11) { if (pix_fmt != AV_PIX_FMT_NV12 && pix_fmt != AV_PIX_FMT_D3D11) { av_log(avctx, AV_LOG_ERROR, "unsupported input pixel format set\n"); @@ -916,7 +918,16 @@ static int mf_encv_input_adjust(AVCodecContext *avctx, IMFMediaType *type) } } - //ff_MFSetAttributeSize((IMFAttributes *)type, &MF_MT_FRAME_SIZE, avctx->width, avctx->height); + ff_MFSetAttributeSize((IMFAttributes *)type, &MF_MT_FRAME_SIZE, avctx->width, avctx->height); + IMFAttributes_SetUINT32(type, &MF_MT_INTERLACE_MODE, MFVideoInterlace_Progressive); + + if (avctx->framerate.num > 0 && avctx->framerate.den > 0) { + framerate = avctx->framerate; + } else { + framerate = av_inv_q(avctx->time_base); + } + + ff_MFSetAttributeRatio((IMFAttributes *)type, &MF_MT_FRAME_RATE, framerate.num, framerate.den); return 0; } diff --git a/libavcodec/nvenc.c b/libavcodec/nvenc.c index 3922305..8c02815 100644 --- a/libavcodec/nvenc.c +++ b/libavcodec/nvenc.c @@ -680,7 +680,9 @@ static int nvenc_check_capabilities(AVCodecContext *avctx) #ifdef NVENC_HAVE_MVHEVC ctx->multiview_supported = nvenc_check_cap(avctx, NV_ENC_CAPS_SUPPORT_MVHEVC_ENCODE) > 0; - if(ctx->profile == NV_ENC_HEVC_PROFILE_MULTIVIEW_MAIN && !ctx->multiview_supported) { + if (avctx->codec_id == AV_CODEC_ID_HEVC && + ctx->profile == NV_ENC_HEVC_PROFILE_MULTIVIEW_MAIN && + !ctx->multiview_supported) { av_log(avctx, AV_LOG_WARNING, "Multiview not supported by the device\n"); return AVERROR(ENOSYS); } diff --git a/libavcodec/pgssubdec.c b/libavcodec/pgssubdec.c index 20583c9..b43ce23 100644 --- a/libavcodec/pgssubdec.c +++ b/libavcodec/pgssubdec.c @@ -35,9 +35,11 @@ #include "libavutil/opt.h" #define RGBA(r,g,b,a) (((unsigned)(a) << 24) | ((r) << 16) | ((g) << 8) | (b)) -#define MAX_EPOCH_PALETTES 8 // Max 8 allowed per PGS epoch -#define MAX_EPOCH_OBJECTS 64 // Max 64 allowed per PGS epoch -#define MAX_OBJECT_REFS 2 // Max objects per display set +#define MAX_EPOCH_PALETTES 8 // Max 8 allowed per PGS epoch +#define MAX_EPOCH_OBJECTS 64 // Max 64 allowed per PGS epoch +#define MAX_OBJECT_REFS 2 // Max objects per display set +#define MAX_OBJECT_WH 4096 // Max object width/height + enum SegmentType { PALETTE_SEGMENT = 0x14, @@ -48,57 +50,80 @@ enum SegmentType { }; typedef struct PGSSubObjectRef { - int id; - int window_id; - uint8_t composition_flag; - int x; - int y; - int crop_x; - int crop_y; - int crop_w; - int crop_h; + uint16_t id; + uint8_t window_id; + uint8_t composition_flag; + uint16_t x; + uint16_t y; + uint16_t crop_x; + uint16_t crop_y; + uint16_t crop_w; + uint16_t crop_h; } PGSSubObjectRef; typedef struct PGSSubPresentation { - int id_number; - int palette_id; - int object_count; + uint8_t palette_flag; + uint8_t palette_id; + uint8_t object_count; PGSSubObjectRef objects[MAX_OBJECT_REFS]; - int64_t pts; + int64_t pts; } PGSSubPresentation; typedef struct PGSSubObject { - int id; - int w; - int h; - uint8_t *rle; - unsigned int rle_buffer_size, rle_data_len; - unsigned int rle_remaining_len; + uint16_t id; + uint16_t w; + uint16_t h; + uint8_t *rle; + uint8_t *bitmap; + uint32_t rle_buffer_size; + uint32_t rle_data_len; + uint32_t rle_remaining_len; + uint32_t bitmap_buffer_size; + uint32_t bitmap_size; } PGSSubObject; typedef struct PGSSubObjects { - int count; + uint8_t count; PGSSubObject object[MAX_EPOCH_OBJECTS]; } PGSSubObjects; typedef struct PGSSubPalette { - int id; - uint32_t clut[256]; + uint8_t id; + uint32_t clut[AVPALETTE_COUNT]; } PGSSubPalette; typedef struct PGSSubPalettes { - int count; + uint8_t count; PGSSubPalette palette[MAX_EPOCH_PALETTES]; } PGSSubPalettes; +typedef struct PGSGraphicPlane { + uint8_t count; + uint8_t writable; + AVSubtitleRect visible_rect[MAX_OBJECT_REFS]; +} PGSGraphicPlane; + typedef struct PGSSubContext { AVClass *class; PGSSubPresentation presentation; PGSSubPalettes palettes; PGSSubObjects objects; + PGSGraphicPlane plane; int forced_subs_only; } PGSSubContext; +static void clear_graphic_plane(PGSSubContext *ctx) +{ + int i; + + for (i = 0; i < ctx->plane.count; i++) { + av_freep(&ctx->plane.visible_rect[i].data[0]); + memset(&ctx->plane.visible_rect[i], 0, sizeof(ctx->plane.visible_rect[i])); + } + ctx->plane.writable = 0; + ctx->plane.count = 0; +} + static void flush_cache(AVCodecContext *avctx) { PGSSubContext *ctx = avctx->priv_data; @@ -106,8 +131,11 @@ static void flush_cache(AVCodecContext *avctx) for (i = 0; i < ctx->objects.count; i++) { av_freep(&ctx->objects.object[i].rle); - ctx->objects.object[i].rle_buffer_size = 0; + ctx->objects.object[i].rle_buffer_size = 0; ctx->objects.object[i].rle_remaining_len = 0; + av_freep(&ctx->objects.object[i].bitmap); + ctx->objects.object[i].bitmap_buffer_size = 0; + ctx->objects.object[i].bitmap_size = 0; } ctx->objects.count = 0; ctx->palettes.count = 0; @@ -144,6 +172,7 @@ static av_cold int init_decoder(AVCodecContext *avctx) static av_cold int close_decoder(AVCodecContext *avctx) { + clear_graphic_plane((PGSSubContext *)avctx->priv_data); flush_cache(avctx); return 0; @@ -159,48 +188,51 @@ static av_cold int close_decoder(AVCodecContext *avctx) * @param buf pointer to the RLE data to process * @param buf_size size of the RLE data to process */ -static int decode_rle(AVCodecContext *avctx, AVSubtitleRect *rect, - const uint8_t *buf, unsigned int buf_size) +static int decode_object_rle(AVCodecContext *avctx, PGSSubObject *object) { - const uint8_t *rle_bitmap_end; + const uint8_t *rle_buf; + const uint8_t *rle_end; int pixel_count, line_count; + rle_buf = object->rle; + rle_end = object->rle + object->rle_data_len; - rle_bitmap_end = buf + buf_size; + object->bitmap_size = object->w * object->h; + av_fast_padded_malloc(&object->bitmap, &object->bitmap_buffer_size, + object->bitmap_size); - rect->data[0] = av_malloc_array(rect->w, rect->h); - - if (!rect->data[0]) + if (!object->bitmap) return AVERROR(ENOMEM); pixel_count = 0; line_count = 0; - while (buf < rle_bitmap_end && line_count < rect->h) { + while (rle_buf < rle_end && line_count < object->h) { uint8_t flags, color; int run; - color = bytestream_get_byte(&buf); + color = bytestream_get_byte(&rle_buf); run = 1; if (color == 0x00) { - flags = bytestream_get_byte(&buf); + flags = bytestream_get_byte(&rle_buf); run = flags & 0x3f; if (flags & 0x40) - run = (run << 8) + bytestream_get_byte(&buf); - color = flags & 0x80 ? bytestream_get_byte(&buf) : 0; + run = (run << 8) + bytestream_get_byte(&rle_buf); + color = flags & 0x80 ? bytestream_get_byte(&rle_buf) : 0; } - if (run > 0 && pixel_count + run <= rect->w * rect->h) { - memset(rect->data[0] + pixel_count, color, run); + if (run > 0 && pixel_count + run <= object->w * object->h) { + memset(object->bitmap + pixel_count, color, run); pixel_count += run; } else if (!run) { /* * New Line. Check if correct pixels decoded, if not display warning * and adjust bitmap pointer to correct new line position. */ - if (pixel_count % rect->w > 0) { - av_log(avctx, AV_LOG_ERROR, "Decoded %d pixels, when line should be %d pixels\n", - pixel_count % rect->w, rect->w); + if (pixel_count % object->w > 0) { + av_log(avctx, AV_LOG_ERROR, + "Decoded %d pixels, when object line should be %d pixels\n", + pixel_count % object->w, object->w); if (avctx->err_recognition & AV_EF_EXPLODE) { return AVERROR_INVALIDDATA; } @@ -209,13 +241,11 @@ static int decode_rle(AVCodecContext *avctx, AVSubtitleRect *rect, } } - if (pixel_count < rect->w * rect->h) { - av_log(avctx, AV_LOG_ERROR, "Insufficient RLE data for subtitle\n"); + if (pixel_count < object->w * object->h) { + av_log(avctx, AV_LOG_ERROR, "Insufficient RLE data for object\n"); return AVERROR_INVALIDDATA; } - - ff_dlog(avctx, "Pixel Count = %d, Area = %d\n", pixel_count, rect->w * rect->h); - + ff_dlog(avctx, "Pixel Count = %d, Area = %d\n", pixel_count, object->w * object->h); return 0; } @@ -237,7 +267,7 @@ static int parse_object_segment(AVCodecContext *avctx, uint8_t sequence_desc; unsigned int rle_bitmap_len, width, height; - int id; + int id, ret; if (buf_size <= 4) return AVERROR_INVALIDDATA; @@ -260,57 +290,71 @@ static int parse_object_segment(AVCodecContext *avctx, /* Read the Sequence Description to determine if start of RLE data or appended to previous RLE */ sequence_desc = bytestream_get_byte(&buf); - if (!(sequence_desc & 0x80)) { - /* Additional RLE data */ - if (buf_size > object->rle_remaining_len) + /* First in sequence object definition segment */ + if (sequence_desc & 0x80) { + if (buf_size <= 7) return AVERROR_INVALIDDATA; + buf_size -= 7; - memcpy(object->rle + object->rle_data_len, buf, buf_size); - object->rle_data_len += buf_size; - object->rle_remaining_len -= buf_size; - - return 0; - } - - if (buf_size <= 7) - return AVERROR_INVALIDDATA; - buf_size -= 7; + /* Decode rle bitmap length, stored size includes width/height data */ + rle_bitmap_len = bytestream_get_be24(&buf) - 2*2; - /* Decode rle bitmap length, stored size includes width/height data */ - rle_bitmap_len = bytestream_get_be24(&buf) - 2*2; + if (buf_size > rle_bitmap_len) { + av_log(avctx, AV_LOG_ERROR, + "Buffer dimension %d larger than the expected RLE data %d\n", + buf_size, rle_bitmap_len); + return AVERROR_INVALIDDATA; + } - if (buf_size > rle_bitmap_len) { - av_log(avctx, AV_LOG_ERROR, - "Buffer dimension %d larger than the expected RLE data %d\n", - buf_size, rle_bitmap_len); - return AVERROR_INVALIDDATA; - } + /* Get bitmap dimensions from data */ + width = bytestream_get_be16(&buf); + height = bytestream_get_be16(&buf); - /* Get bitmap dimensions from data */ - width = bytestream_get_be16(&buf); - height = bytestream_get_be16(&buf); + /* Make sure the bitmap is not too large */ + if (MAX_OBJECT_WH < width || MAX_OBJECT_WH < height || !width || !height) { + av_log(avctx, AV_LOG_ERROR, "Bitmap dimensions (%dx%d) invalid.\n", width, height); + return AVERROR_INVALIDDATA; + } - /* Make sure the bitmap is not too large */ - if (avctx->width < width || avctx->height < height || !width || !height) { - av_log(avctx, AV_LOG_ERROR, "Bitmap dimensions (%dx%d) invalid.\n", width, height); - return AVERROR_INVALIDDATA; - } + object->rle_data_len = 0; + object->w = width; + object->h = height; + /* Dimensions against video are checked at decode after cropping. */ + av_fast_padded_malloc(&object->rle, &object->rle_buffer_size, rle_bitmap_len); - object->w = width; - object->h = height; + if (!object->rle) { + object->rle_remaining_len = 0; + return AVERROR(ENOMEM); + } - av_fast_padded_malloc(&object->rle, &object->rle_buffer_size, rle_bitmap_len); + memcpy(object->rle, buf, buf_size); + object->rle_remaining_len = rle_bitmap_len; + } else { + /* Additional RLE data */ + if (buf_size > object->rle_remaining_len) + return AVERROR_INVALIDDATA; - if (!object->rle) { - object->rle_data_len = 0; - object->rle_remaining_len = 0; - return AVERROR(ENOMEM); + memcpy(object->rle + object->rle_data_len, buf, buf_size); + } + object->rle_data_len += buf_size; + object->rle_remaining_len -= buf_size; + + /* Last in sequence object definition (can be both first and last) */ + if (sequence_desc & 0x40) { + /* Attempt decoding if data is valid */ + if (0 == object->rle_remaining_len) { + ret = decode_object_rle(avctx, object); + if (ret < 0 && (avctx->err_recognition & AV_EF_EXPLODE || ret == AVERROR(ENOMEM))) { + return ret; + } + } else { + av_log(avctx, AV_LOG_ERROR, + "RLE data length %u is %u bytes shorter than expected\n", + object->rle_data_len, object->rle_remaining_len); + if (avctx->err_recognition & AV_EF_EXPLODE) + return AVERROR_INVALIDDATA; + } } - - memcpy(object->rle, buf, buf_size); - object->rle_data_len = buf_size; - object->rle_remaining_len = rle_bitmap_len - buf_size; - return 0; } @@ -318,7 +362,7 @@ static int parse_object_segment(AVCodecContext *avctx, * Parse the palette segment packet. * * The palette segment contains details of the palette, - * a maximum of 256 colors can be defined. + * a maximum of 256 colors (AVPALETTE_COUNT) can be defined. * * @param avctx contains the current codec context * @param buf pointer to the packet to process @@ -391,13 +435,17 @@ static int parse_presentation_segment(AVCodecContext *avctx, int64_t pts) { PGSSubContext *ctx = avctx->priv_data; - int i, state, ret; + int ret; + uint8_t i, state; const uint8_t *buf_end = buf + buf_size; // Video descriptor int w = bytestream_get_be16(&buf); int h = bytestream_get_be16(&buf); + // On a new display set, reset writability of the graphic plane + ctx->plane.writable = 0; + ctx->presentation.pts = pts; ff_dlog(avctx, "Video Dimensions %dx%d\n", @@ -406,88 +454,121 @@ static int parse_presentation_segment(AVCodecContext *avctx, if (ret < 0) return ret; - /* Skip 1 bytes of unknown, frame rate */ - buf++; + /* Skip 3 bytes: framerate (1), presentation id number (2) */ + buf+=3; - // Composition descriptor - ctx->presentation.id_number = bytestream_get_be16(&buf); /* - * state is a 2 bit field that defines pgs epoch boundaries + * State is a 2 bit field that defines pgs epoch boundaries * 00 - Normal, previously defined objects and palettes are still valid * 01 - Acquisition point, previous objects and palettes can be released * 10 - Epoch start, previous objects and palettes can be released * 11 - Epoch continue, previous objects and palettes can be released * - * reserved 6 bits discarded + * Reserved 6 bits discarded */ state = bytestream_get_byte(&buf) >> 6; if (state != 0) { + /* Epoch start always wipes the graphic plane. Epoch continue does only if + * playback is not seamless, which should not happen with a proper stream. + */ + if (0b10 == state) + clear_graphic_plane((PGSSubContext *)avctx->priv_data); flush_cache(avctx); } + /* Reserved 7 bits discarded. */ + ctx->presentation.palette_flag = bytestream_get_byte(&buf) & 0x80; + ctx->presentation.palette_id = bytestream_get_byte(&buf); + /* - * skip palette_update_flag (0x80), + * On palette update, don't parse the compositions references, + * just evaluate the existing graphic plane with the new palette. */ - buf += 1; - ctx->presentation.palette_id = bytestream_get_byte(&buf); - ctx->presentation.object_count = bytestream_get_byte(&buf); - if (ctx->presentation.object_count > MAX_OBJECT_REFS) { - av_log(avctx, AV_LOG_ERROR, - "Invalid number of presentation objects %d\n", - ctx->presentation.object_count); - ctx->presentation.object_count = 2; - if (avctx->err_recognition & AV_EF_EXPLODE) { - return AVERROR_INVALIDDATA; + if (!ctx->presentation.palette_flag) { + ctx->presentation.object_count = bytestream_get_byte(&buf); + if (ctx->presentation.object_count > MAX_OBJECT_REFS) { + av_log(avctx, AV_LOG_ERROR, + "Invalid number of presentation objects %d\n", + ctx->presentation.object_count); + ctx->presentation.object_count = 2; + if (avctx->err_recognition & AV_EF_EXPLODE) { + return AVERROR_INVALIDDATA; + } } - } + for (i = 0; i < ctx->presentation.object_count; i++) { + PGSSubObjectRef *const object = &ctx->presentation.objects[i]; - for (i = 0; i < ctx->presentation.object_count; i++) - { - PGSSubObjectRef *const object = &ctx->presentation.objects[i]; + if (buf_end - buf < 8) { + av_log(avctx, AV_LOG_ERROR, "Insufficient space for object\n"); + ctx->presentation.object_count = i; + return AVERROR_INVALIDDATA; + } - if (buf_end - buf < 8) { - av_log(avctx, AV_LOG_ERROR, "Insufficient space for object\n"); - ctx->presentation.object_count = i; - return AVERROR_INVALIDDATA; - } + object->id = bytestream_get_be16(&buf); + object->window_id = bytestream_get_byte(&buf); + object->composition_flag = bytestream_get_byte(&buf); - object->id = bytestream_get_be16(&buf); - object->window_id = bytestream_get_byte(&buf); - object->composition_flag = bytestream_get_byte(&buf); + object->x = bytestream_get_be16(&buf); + object->y = bytestream_get_be16(&buf); - object->x = bytestream_get_be16(&buf); - object->y = bytestream_get_be16(&buf); + // If cropping + if (object->composition_flag & 0x80) { + object->crop_x = bytestream_get_be16(&buf); + object->crop_y = bytestream_get_be16(&buf); + object->crop_w = bytestream_get_be16(&buf); + object->crop_h = bytestream_get_be16(&buf); + } - // If cropping - if (object->composition_flag & 0x80) { - object->crop_x = bytestream_get_be16(&buf); - object->crop_y = bytestream_get_be16(&buf); - object->crop_w = bytestream_get_be16(&buf); - object->crop_h = bytestream_get_be16(&buf); + /* Placement is checked at decode after cropping. */ + ff_dlog(avctx, "Subtitle Placement x=%d, y=%d\n", + object->x, object->y); } + } + return 0; +} - ff_dlog(avctx, "Subtitle Placement x=%d, y=%d\n", - object->x, object->y); +/** + * Parse the window segment packet. + * + * The window segment instructs the decoder to redraw the graphic plane + * with the composition references provided in the presentation segment + * + * @param avctx contains the current codec context + */ +static int parse_window_segment(AVCodecContext *avctx, const uint8_t *buf, + int buf_size) +{ + PGSSubContext *ctx = (PGSSubContext *)avctx->priv_data; - if (object->x > avctx->width || object->y > avctx->height) { - av_log(avctx, AV_LOG_ERROR, "Subtitle out of video bounds. x = %d, y = %d, video width = %d, video height = %d.\n", - object->x, object->y, - avctx->width, avctx->height); - object->y = object->x = 0; - if (avctx->err_recognition & AV_EF_EXPLODE) { - return AVERROR_INVALIDDATA; - } - } + // 1 byte: number of windows defined + if (bytestream_get_byte(&buf) > MAX_OBJECT_REFS) { + av_log(avctx, AV_LOG_ERROR, "Too many windows defined.\n"); + return AVERROR_INVALIDDATA; } + /* TODO: mask objects with windows when transfering to the graphic plane + * Window Segment Structure + * { + * 1 byte : window id, + * 2 bytes: X position of window, + * 2 bytes: Y position of window, + * 2 bytes: Width of window, + * 2 bytes: Height of window. + * } + */ + // Flush the graphic plane, it will be redrawn. + clear_graphic_plane(ctx); + ctx->plane.writable = 1; + ctx->plane.count = ctx->presentation.object_count; return 0; } /** * Parse the display segment packet. * - * The display segment controls the updating of the display. + * The display segment closes the display set. The inferred data is used + * to decide if the display should be updated. * * @param avctx contains the current codec context * @param data pointer to the data pertaining the subtitle to display @@ -500,26 +581,33 @@ static int display_end_segment(AVCodecContext *avctx, AVSubtitle *sub, PGSSubContext *ctx = avctx->priv_data; int64_t pts; PGSSubPalette *palette; - int i, ret; + int i; pts = ctx->presentation.pts != AV_NOPTS_VALUE ? ctx->presentation.pts : sub->pts; memset(sub, 0, sizeof(*sub)); sub->pts = pts; ctx->presentation.pts = AV_NOPTS_VALUE; - sub->start_display_time = 0; // There is no explicit end time for PGS subtitles. The end time // is defined by the start of the next sub which may contain no // objects (i.e. clears the previous sub) sub->end_display_time = UINT32_MAX; - sub->format = 0; - // Blank if last object_count was 0. - if (!ctx->presentation.object_count) + // Object count is zero only on an epoch start with no WDS + // or the last DS with a WDS had no presentation object. + if (!ctx->plane.count) { return 1; - sub->rects = av_calloc(ctx->presentation.object_count, sizeof(*sub->rects)); - if (!sub->rects) { - return AVERROR(ENOMEM); } + + if (!ctx->presentation.palette_flag && !ctx->plane.writable) { + // This display set does not perform a display update + // E.g. it only defines new objects or palettes for future usage. + return 0; + } + + sub->rects = av_calloc(ctx->plane.count, sizeof(*sub->rects)); + if (!sub->rects) + return AVERROR(ENOMEM); + palette = find_palette(ctx->presentation.palette_id, &ctx->palettes); if (!palette) { // Missing palette. Should only happen with damaged streams. @@ -528,57 +616,128 @@ static int display_end_segment(AVCodecContext *avctx, AVSubtitle *sub, avsubtitle_free(sub); return AVERROR_INVALIDDATA; } - for (i = 0; i < ctx->presentation.object_count; i++) { - AVSubtitleRect *const rect = av_mallocz(sizeof(*rect)); - PGSSubObject *object; - if (!rect) - return AVERROR(ENOMEM); - sub->rects[sub->num_rects++] = rect; - rect->type = SUBTITLE_BITMAP; - - /* Process bitmap */ - object = find_object(ctx->presentation.objects[i].id, &ctx->objects); - if (!object) { - // Missing object. Should only happen with damaged streams. - av_log(avctx, AV_LOG_ERROR, "Invalid object id %d\n", - ctx->presentation.objects[i].id); - if (avctx->err_recognition & AV_EF_EXPLODE) - return AVERROR_INVALIDDATA; - // Leaves rect empty with 0 width and height. - continue; - } - if (ctx->presentation.objects[i].composition_flag & 0x40) - rect->flags |= AV_SUBTITLE_FLAG_FORCED; + for (i = 0; i < ctx->plane.count; i++) { + const PGSSubObjectRef *sub_object = &ctx->presentation.objects[i]; + AVSubtitleRect *const gp_rect = &ctx->plane.visible_rect[i]; + AVSubtitleRect *rect; + gp_rect->type = SUBTITLE_BITMAP; + + // Compose the graphic plane if a window segment has been provided + if (ctx->plane.writable) { + PGSSubObject *object; + + // Process bitmap + object = find_object(sub_object->id, &ctx->objects); + if (!object) { + // Missing object. Should only happen with damaged streams. + av_log(avctx, AV_LOG_ERROR, "Invalid object id %d\n", sub_object->id); + if (avctx->err_recognition & AV_EF_EXPLODE) + return AVERROR_INVALIDDATA; + // Leaves rect empty with 0 width and height. + continue; + } + if (sub_object->composition_flag & 0x40) + gp_rect->flags |= AV_SUBTITLE_FLAG_FORCED; + + gp_rect->x = sub_object->x; + gp_rect->y = sub_object->y; + + if (object->rle) { + int out_of_picture = 0; + gp_rect->w = object->w; + gp_rect->h = object->h; + + gp_rect->linesize[0] = object->w; + + // Check for cropping. + if (sub_object->composition_flag & 0x80) { + int out_of_object = 0; + + if (object->w < sub_object->crop_x + sub_object->crop_w) + out_of_object = 1; + if (object->h < sub_object->crop_y + sub_object->crop_h) + out_of_object = 1; + + if (out_of_object) { + av_log(avctx, AV_LOG_ERROR, + "Subtitle cropping values are out of object. " + "obj_w = %d, obj_h = %d, crop_x = %d, crop_y = %d, " + "crop_w = %d, crop_h = %d.\n", + object->w, + object->h, + sub_object->crop_x, + sub_object->crop_y, + sub_object->crop_w, + sub_object->crop_h); + if (avctx->err_recognition & AV_EF_EXPLODE) + return AVERROR_INVALIDDATA; + } else { + // Replace subtitle dimensions with cropping ones. + gp_rect->w = sub_object->crop_w; + gp_rect->h = sub_object->crop_h; + gp_rect->linesize[0] = sub_object->crop_w; + } + } + + /* Make sure the subtitle is not out of picture. */ + if (avctx->width < gp_rect->x + gp_rect->w || !gp_rect->w) + out_of_picture = 1; + if (avctx->height < gp_rect->y + gp_rect->h || !gp_rect->h) + out_of_picture = 1; + if (out_of_picture) { + av_log(avctx, AV_LOG_ERROR, + "Subtitle out of video bounds. " + "x = %d, y = %d, width = %d, height = %d.\n", + gp_rect->x, gp_rect->y, gp_rect->w, gp_rect->h); + if (avctx->err_recognition & AV_EF_EXPLODE) + return AVERROR_INVALIDDATA; + gp_rect->w = 0; + gp_rect->h = 0; + continue; + } - rect->x = ctx->presentation.objects[i].x; - rect->y = ctx->presentation.objects[i].y; + if (!object->bitmap_size || object->rle_remaining_len) { + gp_rect->w = 0; + gp_rect->h = 0; + continue; + } - if (object->rle) { - rect->w = object->w; - rect->h = object->h; + gp_rect->data[0] = av_malloc_array(gp_rect->w, gp_rect->h); + if (!gp_rect->data[0]) + return AVERROR(ENOMEM); - rect->linesize[0] = object->w; + if (sub_object->composition_flag & 0x80) { + /* Copy cropped bitmap. */ + int y; - if (object->rle_remaining_len) { - av_log(avctx, AV_LOG_ERROR, "RLE data length %u is %u bytes shorter than expected\n", - object->rle_data_len, object->rle_remaining_len); - if (avctx->err_recognition & AV_EF_EXPLODE) - return AVERROR_INVALIDDATA; - } - ret = decode_rle(avctx, rect, object->rle, object->rle_data_len); - if (ret < 0) { - if ((avctx->err_recognition & AV_EF_EXPLODE) || - ret == AVERROR(ENOMEM)) { - return ret; + for (y = 0; y < sub_object->crop_h; y++) { + memcpy(&gp_rect->data[0][y * sub_object->crop_w], + &object->bitmap[(sub_object->crop_y + y) * + object->w + sub_object->crop_x], + sub_object->crop_w); + } + } + else { + /* copy full object */ + memcpy(gp_rect->data[0], object->bitmap, object->bitmap_size); } - rect->w = 0; - rect->h = 0; - continue; } } - /* Allocate memory for colors */ - rect->nb_colors = 256; + // Export graphic plane content with latest palette + rect = av_memdup(gp_rect, sizeof(*gp_rect)); + if (!rect) + return AVERROR(ENOMEM); + + sub->rects[sub->num_rects++] = rect; + if (gp_rect->data[0]) { + rect->data[0] = av_memdup(gp_rect->data[0], rect->w*rect->h); + if (!rect->data[0]) + return AVERROR(ENOMEM); + } + + // Allocate memory for colors + rect->nb_colors = AVPALETTE_COUNT; rect->data[1] = av_mallocz(AVPALETTE_SIZE); if (!rect->data[1]) return AVERROR(ENOMEM); @@ -641,14 +800,7 @@ static int decode(AVCodecContext *avctx, AVSubtitle *sub, ret = parse_presentation_segment(avctx, buf, segment_length, sub->pts); break; case WINDOW_SEGMENT: - /* - * Window Segment Structure (No new information provided): - * 2 bytes: Unknown, - * 2 bytes: X position of subtitle, - * 2 bytes: Y position of subtitle, - * 2 bytes: Width of subtitle, - * 2 bytes: Height of subtitle. - */ + ret = parse_window_segment(avctx, buf, segment_length); break; case DISPLAY_SEGMENT: if (*got_sub_ptr) { diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index fc1b879..eb0c116 100644 --- a/libavcodec/qsvenc.c +++ b/libavcodec/qsvenc.c @@ -492,6 +492,9 @@ static void dump_video_av1_param(AVCodecContext *avctx, QSVEncContext *q, mfxExtAV1BitstreamParam *av1_bs_param = (mfxExtAV1BitstreamParam *)coding_opts[1]; mfxExtCodingOption2 *co2 = (mfxExtCodingOption2*)coding_opts[2]; mfxExtCodingOption3 *co3 = (mfxExtCodingOption3*)coding_opts[3]; +#if QSV_HAVE_EXT_AV1_SCC + mfxExtAV1ScreenContentTools *scc = (mfxExtAV1ScreenContentTools*)coding_opts[4]; +#endif av_log(avctx, AV_LOG_VERBOSE, "profile: %s; level: %"PRIu16"\n", print_profile(avctx->codec_id, info->CodecProfile), info->CodecLevel); @@ -564,6 +567,13 @@ static void dump_video_av1_param(AVCodecContext *avctx, QSVEncContext *q, print_threestate(av1_bs_param->WriteIVFHeaders)); av_log(avctx, AV_LOG_VERBOSE, "LowDelayBRC: %s\n", print_threestate(co3->LowDelayBRC)); av_log(avctx, AV_LOG_VERBOSE, "MaxFrameSize: %d;\n", co2->MaxFrameSize); +#if QSV_HAVE_EXT_AV1_SCC + if (scc) { + av_log(avctx, AV_LOG_VERBOSE, + "Palette: %s; IntraBlockCopy: %s\n", + print_threestate(scc->Palette), print_threestate(scc->IntraBlockCopy)); + } +#endif } #endif @@ -1280,6 +1290,28 @@ static int init_video_param(AVCodecContext *avctx, QSVEncContext *q) } #endif +#if QSV_HAVE_EXT_AV1_SCC + if (q->palette_mode || q->intrabc) { + if (QSV_RUNTIME_VERSION_ATLEAST(q->ver, 2, 13)) { + if (q->param.mfx.CodecId != MFX_CODEC_AV1) { + av_log(avctx, AV_LOG_ERROR, "Not supported encoder for Screen Content Tool Encode. " + "Supported: av1_qsv \n"); + return AVERROR_UNKNOWN; + } + + q->extsccparam.Header.BufferId = MFX_EXTBUFF_AV1_SCREEN_CONTENT_TOOLS; + q->extsccparam.Header.BufferSz = sizeof(q->extsccparam); + q->extsccparam.Palette = q->palette_mode ? MFX_CODINGOPTION_ON : MFX_CODINGOPTION_OFF; + q->extsccparam.IntraBlockCopy = q->intrabc ? MFX_CODINGOPTION_ON : MFX_CODINGOPTION_OFF; + q->extparam_internal[q->nb_extparam_internal++] = (mfxExtBuffer *)&q->extsccparam; + } else { + av_log(avctx, AV_LOG_ERROR, + "This version of runtime doesn't support Screen Content Tool Encode\n"); + return AVERROR_UNKNOWN; + } + } +#endif + if (!check_enc_param(avctx,q)) { av_log(avctx, AV_LOG_ERROR, "some encoding parameters are not supported by the QSV " @@ -1387,11 +1419,21 @@ static int qsv_retrieve_enc_av1_params(AVCodecContext *avctx, QSVEncContext *q) .Header.BufferSz = sizeof(co3), }; +#if QSV_HAVE_EXT_AV1_SCC + mfxExtAV1ScreenContentTools scc_buf = { + .Header.BufferId = MFX_EXTBUFF_AV1_SCREEN_CONTENT_TOOLS, + .Header.BufferSz = sizeof(scc_buf), + }; +#endif + mfxExtBuffer *ext_buffers[] = { (mfxExtBuffer*)&av1_extend_tile_buf, (mfxExtBuffer*)&av1_bs_param, (mfxExtBuffer*)&co2, (mfxExtBuffer*)&co3, +#if QSV_HAVE_EXT_AV1_SCC + (mfxExtBuffer*)&scc_buf, +#endif }; if (!QSV_RUNTIME_VERSION_ATLEAST(q->ver, 2, 5)) { @@ -1840,6 +1882,13 @@ int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q) return ret; } + // Update AVCodecContext with actual encoding parameters + mfxInfoMFX *info = &q->param.mfx; + avctx->has_b_frames = 0; + if (info->GopRefDist > 1) { + avctx->has_b_frames = info->GopRefDist - 1; + } + q->avctx = avctx; return 0; diff --git a/libavcodec/qsvenc.h b/libavcodec/qsvenc.h index 4bc77f2..2e0a19b 100644 --- a/libavcodec/qsvenc.h +++ b/libavcodec/qsvenc.h @@ -38,6 +38,7 @@ #define QSV_HAVE_EXT_VP9_TILES QSV_VERSION_ATLEAST(1, 29) #define QSV_HAVE_EXT_AV1_PARAM QSV_VERSION_ATLEAST(2, 5) +#define QSV_HAVE_EXT_AV1_SCC QSV_VERSION_ATLEAST(2, 13) #if defined(_WIN32) || defined(__CYGWIN__) #define QSV_HAVE_AVBR 1 @@ -188,10 +189,13 @@ typedef struct QSVEncContext { mfxFrameSurface1 **opaque_surfaces; AVBufferRef *opaque_alloc_buf; #endif +#if QSV_HAVE_EXT_AV1_SCC + mfxExtAV1ScreenContentTools extsccparam; +#endif mfxExtVideoSignalInfo extvsi; - mfxExtBuffer *extparam_internal[5 + (QSV_HAVE_MF * 2) + (QSV_HAVE_EXT_AV1_PARAM * 2) + QSV_HAVE_HE]; + mfxExtBuffer *extparam_internal[5 + (QSV_HAVE_MF * 2) + (QSV_HAVE_EXT_AV1_PARAM * 2) + QSV_HAVE_HE + QSV_HAVE_EXT_AV1_SCC]; int nb_extparam_internal; mfxExtBuffer **extparam_str; @@ -319,6 +323,8 @@ typedef struct QSVEncContext { int dual_gfx; AVDictionary *qsv_params; + int palette_mode; + int intrabc; } QSVEncContext; int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q); diff --git a/libavcodec/qsvenc_av1.c b/libavcodec/qsvenc_av1.c index c7aa6e5..5edb5f3 100644 --- a/libavcodec/qsvenc_av1.c +++ b/libavcodec/qsvenc_av1.c @@ -189,6 +189,10 @@ static const AVOption options[] = { { "tile_cols", "Number of columns for tiled encoding", OFFSET(qsv.tile_cols), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, UINT16_MAX, VE }, { "tile_rows", "Number of rows for tiled encoding", OFFSET(qsv.tile_rows), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, UINT16_MAX, VE }, { "look_ahead_depth", "Depth of look ahead in number frames, available when extbrc option is enabled", OFFSET(qsv.look_ahead_depth), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 100, VE }, +#if QSV_HAVE_EXT_AV1_SCC + { "palette_mode", "Enable palette mode of Screen Content Tool for encoding", OFFSET(qsv.palette_mode), AV_OPT_TYPE_BOOL, { .i64 = 0}, 0, 1, VE}, + { "intrabc", "Enable intra block copy of Screen Content Tool for encoding", OFFSET(qsv.intrabc), AV_OPT_TYPE_BOOL, { .i64 = 0}, 0, 1, VE}, +#endif { NULL }, }; diff --git a/libavcodec/videotoolbox.c b/libavcodec/videotoolbox.c index 2cd22cb..247ac19 100644 --- a/libavcodec/videotoolbox.c +++ b/libavcodec/videotoolbox.c @@ -132,9 +132,6 @@ static int videotoolbox_postproc_frame(void *avctx, AVFrame *frame) frame->crop_top = 0; frame->crop_bottom = 0; - if ((ret = av_vt_pixbuf_set_attachments(avctx, ref->pixbuf, frame)) < 0) - return ret; - frame->data[3] = (uint8_t*)ref->pixbuf; if (ref->hw_frames_ctx) { @@ -820,7 +817,7 @@ static CFDictionaryRef videotoolbox_buffer_attributes_create(int width, #if TARGET_OS_IPHONE CFDictionarySetValue(buffer_attributes, kCVPixelBufferOpenGLESCompatibilityKey, kCFBooleanTrue); #else - CFDictionarySetValue(buffer_attributes, kCVPixelBufferIOSurfaceOpenGLTextureCompatibilityKey, kCFBooleanTrue); + CFDictionarySetValue(buffer_attributes, kCVPixelBufferMetalCompatibilityKey, kCFBooleanTrue); #endif CFRelease(io_surface_properties); @@ -956,6 +953,23 @@ static int videotoolbox_start(AVCodecContext *avctx) break; } +#if ARCH_X86_64 + if (avctx->codec_id == AV_CODEC_ID_H264 && + avctx->sw_pix_fmt == AV_PIX_FMT_YUV420P10) + { + // 10-bit H.264 is not supported on x86_64 + return AVERROR(ENOSYS); + } +#endif + + if (avctx->codec_id == AV_CODEC_ID_H264 && + (avctx->level == 61 || avctx->level == 62)) + { + // H.264 Level 6.1 and 6.2 can't be + // decoded properly + return AVERROR(ENOSYS); + } + #if defined(MAC_OS_X_VERSION_10_9) && !TARGET_OS_IPHONE && (MAC_OS_X_VERSION_MAX_ALLOWED >= MAC_OS_X_VERSION_10_9) && AV_HAS_BUILTIN(__builtin_available) if (avctx->codec_id == AV_CODEC_ID_PRORES) { if (__builtin_available(macOS 10.9, *)) { diff --git a/libavfilter/vf_vpp_qsv.c b/libavfilter/vf_vpp_qsv.c index 4de99d6..182bbf7 100644 --- a/libavfilter/vf_vpp_qsv.c +++ b/libavfilter/vf_vpp_qsv.c @@ -494,9 +494,9 @@ static int vpp_set_frame_ext_params(AVFilterContext *ctx, const AVFrame *in, AVF outvsi_conf.Header.BufferId = MFX_EXTBUFF_VIDEO_SIGNAL_INFO_OUT; outvsi_conf.Header.BufferSz = sizeof(mfxExtVideoSignalInfo); outvsi_conf.VideoFullRange = (out->color_range == AVCOL_RANGE_JPEG); - outvsi_conf.ColourPrimaries = (out->color_primaries == AVCOL_PRI_UNSPECIFIED) ? AVCOL_PRI_BT709 : out->color_primaries; - outvsi_conf.TransferCharacteristics = (out->color_trc == AVCOL_TRC_UNSPECIFIED) ? AVCOL_TRC_BT709 : out->color_trc; - outvsi_conf.MatrixCoefficients = (out->colorspace == AVCOL_SPC_UNSPECIFIED) ? AVCOL_SPC_BT709 : out->colorspace; + outvsi_conf.ColourPrimaries = (out->color_primaries == AVCOL_PRI_UNSPECIFIED) ? invsi_conf.ColourPrimaries : out->color_primaries; + outvsi_conf.TransferCharacteristics = (out->color_trc == AVCOL_TRC_UNSPECIFIED) ? invsi_conf.TransferCharacteristics : out->color_trc; + outvsi_conf.MatrixCoefficients = (out->colorspace == AVCOL_SPC_UNSPECIFIED) ? invsi_conf.MatrixCoefficients : out->colorspace; outvsi_conf.ColourDescriptionPresent = 1; if (memcmp(&vpp->invsi_conf, &invsi_conf, sizeof(mfxExtVideoSignalInfo)) || diff --git a/libavformat/isom.h b/libavformat/isom.h index d7e1385..8fbc4d0 100644 --- a/libavformat/isom.h +++ b/libavformat/isom.h @@ -282,6 +282,9 @@ typedef struct MOVStreamContext { MOVEncryptionIndex *encryption_index; } cenc; + int has_fallback; // Audio fallback track + int fallback; + struct IAMFDemuxContext *iamf; int iamf_stream_offset; } MOVStreamContext; diff --git a/libavformat/matroskaenc.c b/libavformat/matroskaenc.c index 2c85672..8250c96 100644 --- a/libavformat/matroskaenc.c +++ b/libavformat/matroskaenc.c @@ -2994,6 +2994,16 @@ static int mkv_check_new_extra_data(AVFormatContext *s, const AVPacket *pkt) case AV_CODEC_ID_AAC: if (side_data_size && mkv->track.bc) { int output_sample_rate = 0; + if (par->extradata && par->extradata_size) { + if (par->extradata_size != side_data_size || + memcmp(par->extradata, side_data, side_data_size)) { + av_log(s, AV_LOG_ERROR, "Error, AAC extradata changed mid-stream.\n"); + return AVERROR(EINVAL); + } else { + // Already written + break; + } + } ret = get_aac_sample_rates(s, mkv, side_data, side_data_size, &track->sample_rate, &output_sample_rate); if (ret < 0) diff --git a/libavformat/mov.c b/libavformat/mov.c index 222d79e..eb0de62 100644 --- a/libavformat/mov.c +++ b/libavformat/mov.c @@ -57,6 +57,7 @@ #include "libavcodec/mpegaudiodecheader.h" #include "libavcodec/mlp_parse.h" #include "avformat.h" +#include "avlanguage.h" #include "internal.h" #include "avio_internal.h" #include "demux.h" @@ -131,6 +132,33 @@ static int mov_metadata_int8_no_padding(MOVContext *c, AVIOContext *pb, return 0; } +static int mov_metadata_int16_no_padding(MOVContext *c, AVIOContext *pb, + unsigned len, const char *key) +{ + c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED; + av_dict_set_int(&c->fc->metadata, key, avio_rb16(pb), 0); + + return 0; +} + +static int mov_metadata_int32_no_padding(MOVContext *c, AVIOContext *pb, + unsigned len, const char *key) +{ + c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED; + av_dict_set_int(&c->fc->metadata, key, avio_rb32(pb), 0); + + return 0; +} + +static int mov_metadata_int64_no_padding(MOVContext *c, AVIOContext *pb, + unsigned len, const char *key) +{ + c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED; + av_dict_set_int(&c->fc->metadata, key, avio_rb64(pb), 0); + + return 0; +} + static int mov_metadata_gnre(MOVContext *c, AVIOContext *pb, unsigned len, const char *key) { @@ -342,6 +370,73 @@ static int mov_metadata_hmmt(MOVContext *c, AVIOContext *pb, unsigned len) return 0; } +static int mov_read_3gp_udta_tag(MOVContext *c, AVIOContext *pb, MOVAtom atom) +{ + const char *key; + AVDictionary *metadata; + uint16_t langcode = 0; + char key2[32], language[4] = {0}; + uint32_t str_size, version; + char *str; + if (atom.size < 6) + return AVERROR_INVALIDDATA; + switch (atom.type) { + case MKTAG( 'a','l','b','m'): key = "album"; break; + case MKTAG( 'a','u','t','h'): key = "author"; break; + case MKTAG( 'c','p','r','t'): key = "copyright"; break; + case MKTAG( 'd','s','c','p'): key = "comment"; break; + case MKTAG( 'g','n','r','e'): key = "genre"; break; + case MKTAG( 'p','e','r','f'): key = "artist"; break; + case MKTAG( 't','i','t','l'): key = "title"; break; + case MKTAG( 'y','r','r','c'): key = "date"; break; + default: return 0; + } + version = avio_rb32(pb); // version + flags + if (version != 0) + av_log(c->fc, AV_LOG_WARNING, "udta %s unknown version number: %u\n", str, version); + if (MKTAG( 'y','r','r','c') == atom.type) { + int year; + year = avio_rb16(pb); + str = av_asprintf("%d", year); + if (!str) + return AVERROR(ENOMEM); + } else { + int ret; + const char *tmp; + langcode = avio_rb16(pb); + ff_mov_lang_to_iso639(langcode, language); + tmp = ff_convert_lang_to(language, AV_LANG_ISO639_2_BIBL); + if (!tmp) + av_log(c->fc, AV_LOG_WARNING, "udta %s unknown language code: %u\n", str, langcode); + str_size = atom.size - 6; + if (str_size <= 0 || str_size >= INT_MAX/2) + return AVERROR_INVALIDDATA; + str = av_mallocz(str_size + 1); + if (!str) + return AVERROR(ENOMEM); + ret = ffio_read_size(pb, str, str_size); + if (ret < 0) { + av_free(str); + return ret; + } + str[str_size] = 0; + } + if (c->trak_index < 0) { + metadata = c->fc->metadata; + c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED; + } + else { + metadata = c->fc->streams[c->trak_index]->metadata; + } + av_dict_set(&metadata, key, str, 0); + if (*language && strcmp(language, "und")) { + snprintf(key2, sizeof(key2), "%s-%s", key, language); + av_dict_set(&metadata, key2, str, 0); + } + av_freep(&str); + return 0; +} + static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom) { char tmp_key[AV_FOURCC_MAX_STRING_SIZE] = {0}; @@ -370,36 +465,71 @@ static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom) case MKTAG( 'a','k','I','D'): key = "account_type"; parse = mov_metadata_int8_no_padding; break; case MKTAG( 'a','p','I','D'): key = "account_id"; break; + case MKTAG( 'a','t','I','D'): key = "artist_id"; + parse = mov_metadata_int32_no_padding; break; case MKTAG( 'c','a','t','g'): key = "category"; break; + case MKTAG( 'c','m','I','D'): key = "composer_id"; + parse = mov_metadata_int32_no_padding; break; + case MKTAG( 'c','n','I','D'): key = "content_id"; + parse = mov_metadata_int32_no_padding; break; case MKTAG( 'c','p','i','l'): key = "compilation"; parse = mov_metadata_int8_no_padding; break; - case MKTAG( 'c','p','r','t'): key = "copyright"; break; + case MKTAG( 'c','p','r','t'): + key = "copyright"; + if (!c->itunes_metadata) { + int64_t pos = avio_tell(pb); + int ret = mov_read_3gp_udta_tag(c, pb, atom); + if (ret != AVERROR_INVALIDDATA) + return ret; + avio_seek(pb, pos, SEEK_SET); + } + break; case MKTAG( 'd','e','s','c'): key = "description"; break; case MKTAG( 'd','i','s','k'): key = "disc"; parse = mov_metadata_track_or_disc_number; break; case MKTAG( 'e','g','i','d'): key = "episode_uid"; parse = mov_metadata_int8_no_padding; break; case MKTAG( 'F','I','R','M'): key = "firmware"; raw = 1; break; - case MKTAG( 'g','n','r','e'): key = "genre"; - parse = mov_metadata_gnre; break; + case MKTAG( 'g','e','I','D'): key = "genre_id"; + parse = mov_metadata_int32_no_padding; break; + case MKTAG( 'g','n','r','e'): + key = "genre"; + parse = mov_metadata_gnre; + if (!c->itunes_metadata) { + int64_t pos = avio_tell(pb); + int ret = mov_read_3gp_udta_tag(c, pb, atom); + if (ret != AVERROR_INVALIDDATA) + return ret; + avio_seek(pb, pos, SEEK_SET); + } + break; case MKTAG( 'h','d','v','d'): key = "hd_video"; parse = mov_metadata_int8_no_padding; break; case MKTAG( 'H','M','M','T'): return mov_metadata_hmmt(c, pb, atom.size); + case MKTAG( 'i','t','n','u'): key = "itunes_u"; + parse = mov_metadata_int8_no_padding; break; case MKTAG( 'k','e','y','w'): key = "keywords"; break; case MKTAG( 'l','d','e','s'): key = "synopsis"; break; case MKTAG( 'l','o','c','i'): return mov_metadata_loci(c, pb, atom.size); case MKTAG( 'm','a','n','u'): key = "make"; break; case MKTAG( 'm','o','d','l'): key = "model"; break; - case MKTAG( 'n','a','m','e'): key = "name"; break; + case MKTAG( 'n','a','m','e'): key = "title"; raw = 1; break; case MKTAG( 'p','c','s','t'): key = "podcast"; parse = mov_metadata_int8_no_padding; break; case MKTAG( 'p','g','a','p'): key = "gapless_playback"; parse = mov_metadata_int8_no_padding; break; + case MKTAG( 'p','l','I','D'): key = "playlist_id"; + parse = mov_metadata_int64_no_padding; break; case MKTAG( 'p','u','r','d'): key = "purchase_date"; break; case MKTAG( 'r','t','n','g'): key = "rating"; parse = mov_metadata_int8_no_padding; break; + case MKTAG( 's','f','I','D'): key = "itunes_country"; + parse = mov_metadata_int32_no_padding; break; + case MKTAG( 's','d','e','s'): key = "series_description"; break; + case MKTAG( 's','h','w','m'): key = "show_work_and_movement"; + parse = mov_metadata_int8_no_padding; break; case MKTAG( 's','o','a','a'): key = "sort_album_artist"; break; case MKTAG( 's','o','a','l'): key = "sort_album"; break; case MKTAG( 's','o','a','r'): key = "sort_artist"; break; @@ -408,6 +538,8 @@ static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom) case MKTAG( 's','o','s','n'): key = "sort_show"; break; case MKTAG( 's','t','i','k'): key = "media_type"; parse = mov_metadata_int8_no_padding; break; + case MKTAG( 't','m','p','o'): key = "tmpo"; + parse = mov_metadata_int16_no_padding; break; case MKTAG( 't','r','k','n'): key = "track"; parse = mov_metadata_track_or_disc_number; break; case MKTAG( 't','v','e','n'): key = "episode_id"; break; @@ -417,17 +549,23 @@ static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom) case MKTAG( 't','v','s','h'): key = "show"; break; case MKTAG( 't','v','s','n'): key = "season_number"; parse = mov_metadata_int8_bypass_padding; break; + case MKTAG( 'x','i','d',' '): key = "xid"; break; case MKTAG(0xa9,'A','R','T'): key = "artist"; break; case MKTAG(0xa9,'P','R','D'): key = "producer"; break; case MKTAG(0xa9,'a','l','b'): key = "album"; break; - case MKTAG(0xa9,'a','u','t'): key = "artist"; break; + case MKTAG(0xa9,'a','r','d'): key = "art_director"; break; + case MKTAG(0xa9,'a','r','g'): key = "arranger"; break; + case MKTAG(0xa9,'a','u','t'): key = "author"; break; + case MKTAG(0xa9,'c','a','k'): key = "acknowledgement"; break; case MKTAG(0xa9,'c','h','p'): key = "chapter"; break; case MKTAG(0xa9,'c','m','t'): key = "comment"; break; case MKTAG(0xa9,'c','o','m'): key = "composer"; break; + case MKTAG(0xa9,'c','o','n'): key = "conductor"; break; case MKTAG(0xa9,'c','p','y'): key = "copyright"; break; case MKTAG(0xa9,'d','a','y'): key = "date"; break; case MKTAG(0xa9,'d','i','r'): key = "director"; break; case MKTAG(0xa9,'d','i','s'): key = "disclaimer"; break; + case MKTAG(0xa9,'d','e','s'): key = "song_description"; break; case MKTAG(0xa9,'e','d','1'): key = "edit_date"; break; case MKTAG(0xa9,'e','n','c'): key = "encoder"; break; case MKTAG(0xa9,'f','m','t'): key = "original_format"; break; @@ -435,23 +573,45 @@ static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom) case MKTAG(0xa9,'g','r','p'): key = "grouping"; break; case MKTAG(0xa9,'h','s','t'): key = "host_computer"; break; case MKTAG(0xa9,'i','n','f'): key = "comment"; break; + case MKTAG(0xa9,'l','n','t'): key = "linear_notes"; break; case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break; case MKTAG(0xa9,'m','a','k'): key = "make"; break; case MKTAG(0xa9,'m','o','d'): key = "model"; break; + case MKTAG(0xa9,'m','v','n'): key = "movement_name"; break; + case MKTAG(0xa9,'m','v','i'): key = "movement_number"; + parse = mov_metadata_int16_no_padding; break; + case MKTAG(0xa9,'m','v','c'): key = "movement_count"; + parse = mov_metadata_int16_no_padding; break; case MKTAG(0xa9,'n','a','m'): key = "title"; break; case MKTAG(0xa9,'o','p','e'): key = "original_artist"; break; + case MKTAG(0xa9,'p','h','g'): key = "phonogram_rights"; break; case MKTAG(0xa9,'p','r','d'): key = "producer"; break; case MKTAG(0xa9,'p','r','f'): key = "performers"; break; + case MKTAG(0xa9,'p','u','b'): key = "publisher"; break; case MKTAG(0xa9,'r','e','q'): key = "playback_requirements"; break; + case MKTAG(0xa9,'s','n','e'): key = "sound_engineer"; break; + case MKTAG(0xa9,'s','o','l'): key = "soloist"; break; case MKTAG(0xa9,'s','r','c'): key = "original_source"; break; case MKTAG(0xa9,'s','t','3'): key = "subtitle"; break; case MKTAG(0xa9,'s','w','r'): key = "encoder"; break; + case MKTAG(0xa9,'t','h','x'): key = "thanks"; break; case MKTAG(0xa9,'t','o','o'): key = "encoder"; break; case MKTAG(0xa9,'t','r','k'): key = "track"; break; case MKTAG(0xa9,'u','r','l'): key = "URL"; break; + case MKTAG(0xa9,'w','r','k'): key = "work_name"; break; case MKTAG(0xa9,'w','r','n'): key = "warning"; break; case MKTAG(0xa9,'w','r','t'): key = "composer"; break; + case MKTAG(0xa9,'x','p','d'): key = "executive_producer"; break; case MKTAG(0xa9,'x','y','z'): key = "location"; break; + case MKTAG( 'a','l','b','m'): + case MKTAG( 'a','u','t','h'): + case MKTAG( 'd','s','c','p'): + case MKTAG( 'p','e','r','f'): + case MKTAG( 't','i','t','l'): + case MKTAG( 'y','r','r','c'): + if (!c->itunes_metadata) { + return mov_read_3gp_udta_tag(c, pb, atom); + } } retry: if (c->itunes_metadata && atom.size > 8) { @@ -9514,6 +9674,22 @@ fail: return ret; } +static int mov_read_fall(MOVContext *c, AVIOContext *pb, MOVAtom atom) +{ + AVStream *st; + MOVStreamContext *sc; + + if (c->fc->nb_streams < 1) + return 0; + st = c->fc->streams[c->fc->nb_streams-1]; + sc = st->priv_data; + + sc->fallback = avio_rb32(pb); + sc->has_fallback = 1; + + return 0; +} + static const MOVParseTableEntry mov_default_parse_table[] = { { MKTAG('A','C','L','R'), mov_read_aclr }, { MKTAG('A','P','R','G'), mov_read_avid }, @@ -9616,6 +9792,7 @@ static const MOVParseTableEntry mov_default_parse_table[] = { { MKTAG('v','p','c','C'), mov_read_vpcc }, { MKTAG('m','d','c','v'), mov_read_mdcv }, { MKTAG('c','l','l','i'), mov_read_clli }, +{ MKTAG('f','a','l','l'), mov_read_fall }, { MKTAG('d','v','c','C'), mov_read_dvcc_dvvc }, { MKTAG('d','v','v','C'), mov_read_dvcc_dvvc }, { MKTAG('d','v','w','C'), mov_read_dvcc_dvvc }, @@ -11071,6 +11248,23 @@ static int mov_read_header(AVFormatContext *s) err = ff_replaygain_export(st, s->metadata); if (err < 0) return err; + if (sc->has_fallback) { + for (j = 0; j < s->nb_streams; j++) { + if (s->streams[j]->id == sc->fallback) { + AVPacketSideData *sd; + int *fallback; + sd = av_packet_side_data_new(&st->codecpar->coded_side_data, + &st->codecpar->nb_coded_side_data, + AV_PKT_DATA_FALLBACK_TRACK, + sizeof(int), 0); + if (!sd) + return AVERROR(ENOMEM); + fallback = (int*)sd->data; + *fallback = j; + break; + } + } + } break; case AVMEDIA_TYPE_VIDEO: if (sc->display_matrix) { diff --git a/libavformat/movenc.c b/libavformat/movenc.c index fe6b259..89ec962 100644 --- a/libavformat/movenc.c +++ b/libavformat/movenc.c @@ -28,6 +28,7 @@ #include "movenc.h" #include "avformat.h" +#include "avlanguage.h" #include "avio_internal.h" #include "dovi_isom.h" #include "riff.h" @@ -1001,6 +1002,15 @@ static int mov_write_chan_tag(AVFormatContext *s, AVIOContext *pb, MOVTrack *tra if (track->multichannel_as_mono) return 0; + // Custom channels layout is not handled properly for these formats + // unset it for now, it will fallback to the layout stored in the extradata + // TODO: find the reason + if (track->par->codec_id == AV_CODEC_ID_AAC || + track->par->codec_id == AV_CODEC_ID_ALAC) + { + return 0; + } + ret = ff_mov_get_channel_layout_tag(track->par, &layout_tag, &bitmap, &channel_desc); @@ -1373,11 +1383,13 @@ static int mov_write_audio_tag(AVFormatContext *s, AVIOContext *pb, MOVMuxContex int64_t pos = avio_tell(pb); int version = 0; uint32_t tag = track->tag; + int lpcm_flags = 0; int ret = 0; if (track->mode == MODE_MOV) { - if (track->timescale > UINT16_MAX || !track->par->ch_layout.nb_channels) { - if (mov_get_lpcm_flags(track->par->codec_id)) + lpcm_flags = mov_get_lpcm_flags(track->par->codec_id); + if (track->timescale > UINT16_MAX || track->par->ch_layout.nb_channels > 2 || lpcm_flags) { + if (lpcm_flags) tag = AV_RL32("lpcm"); version = 2; } else if (track->audio_vbr || mov_pcm_le_gt16(track->par->codec_id) || @@ -4244,6 +4256,41 @@ static int mov_write_udta_sdp(AVIOContext *pb, MOVTrack *track) return len + 24; } +static uint16_t language_code(const char *str) +{ + return (((str[0] - 0x60) & 0x1F) << 10) + + (((str[1] - 0x60) & 0x1F) << 5) + + (( str[2] - 0x60) & 0x1F); +} +static int mov_write_3gp_udta_tag(AVIOContext *pb, AVDictionary *metadata, + const char *tag, const char *str) +{ + int64_t pos = avio_tell(pb); + AVDictionaryEntry *t = av_dict_get(metadata, str, NULL, 0); + if (!t || !utf8len(t->value)) + return 0; + avio_wb32(pb, 0); /* size */ + ffio_wfourcc(pb, tag); /* type */ + avio_wb32(pb, 0); /* version + flags */ + if (!strcmp(tag, "yrrc")) + avio_wb16(pb, atoi(t->value)); + else { + int lang = 0, len; + len = strlen(t->key); + if (t->key[len - 4] == '-') { + lang = ff_mov_iso639_to_lang(&t->key[len - 3], 1); + } + if (!lang) + lang = ff_mov_iso639_to_lang("und", 1); + avio_wb16(pb, lang); /* language */ + avio_write(pb, t->value, strlen(t->value) + 1); /* UTF8 string value */ + if (!strcmp(tag, "albm") && + (t = av_dict_get(metadata, "track", NULL, 0))) + avio_w8(pb, atoi(t->value)); + } + return update_size(pb, pos); +} + static int mov_write_track_metadata(AVIOContext *pb, AVStream *st, const char *tag, const char *str) { @@ -4318,8 +4365,23 @@ static int mov_write_track_udta_tag(AVIOContext *pb, MOVMuxContext *mov, if (ret < 0) return ret; - if (mov->mode & (MODE_MP4|MODE_MOV)) + if (mov->mode & (MODE_MP4|MODE_MOV)) { + AVDictionaryEntry *t = NULL; + int und = 0; + mov_write_track_metadata(pb_buf, st, "name", "title"); + while ((t = av_dict_get(st->metadata, "title-", t, AV_DICT_IGNORE_SUFFIX))) { + int len = strlen(t->key); + if (len == 10 && + ff_convert_lang_to(&t->key[len - 3], AV_LANG_ISO639_2_BIBL)) { + mov_write_3gp_udta_tag(pb_buf, st->metadata, "titl", t->key); + if (!strcmp("und", &t->key[len - 3])) + und = 1; + } + } + if (!und) + mov_write_3gp_udta_tag(pb_buf, st->metadata, "titl", "title"); + } if (mov->mode & MODE_MP4) { if ((ret = mov_write_track_kinds(pb_buf, st)) < 0) @@ -4694,20 +4756,66 @@ static int mov_write_trkn_tag(AVIOContext *pb, MOVMuxContext *mov, return size; } +/* iTunes iTMF metadata */ +static int mov_write_itmf_tag(AVIOContext *pb, AVFormatContext *s, + const char *name, const char *tag) +{ + AVDictionaryEntry *t = av_dict_get(s->metadata, tag, NULL, 0); + uint64_t size = 0; + if (t) { + size_t name_len = strlen(name); + size_t value_len = strlen(t->value); + size = 8 + 28 + 12 + name_len + 16 + value_len; + + if (size > INT_MAX) { + av_log(s, AV_LOG_WARNING, + "Unable to write %s, invalid value len %ld\n", + name, value_len); + return 0; + } + + avio_wb32(pb, size); // size + ffio_wfourcc(pb, "----"); // type + + avio_wb32(pb, 28); // size + ffio_wfourcc(pb, "mean"); // type + avio_wb32(pb, 0); // version + flags + avio_write(pb, "com.apple.iTunes", 16); + + avio_wb32(pb, 12 + name_len); // size + ffio_wfourcc(pb, "name"); // type + avio_wb32(pb, 0); // version + flags + avio_write(pb, name, name_len); + + avio_wb32(pb, 16 + value_len); // size + ffio_wfourcc(pb, "data"); // type + avio_wb16(pb, 0); // typeReserved + avio_w8(pb, 0); // typeSetIdentifier + avio_w8(pb, 1); // typeCode = UTF-8 + avio_wb32(pb, 0); // locale + avio_write(pb, t->value, value_len); + } + return size; +} + static int mov_write_int8_metadata(AVFormatContext *s, AVIOContext *pb, const char *name, const char *tag, int len) { AVDictionaryEntry *t = NULL; - uint8_t num; + uint64_t num; int size = 24 + len; - if (len != 1 && len != 4) + if (len != 1 && len != 4 && + len != 2 && len != 8) return -1; if (!(t = av_dict_get(s->metadata, tag, NULL, 0))) return 0; - num = atoi(t->value); + if (len <= 4) + num = atoi(t->value); + else + num = atol(t->value); avio_wb32(pb, size); ffio_wfourcc(pb, name); @@ -4715,7 +4823,9 @@ static int mov_write_int8_metadata(AVFormatContext *s, AVIOContext *pb, ffio_wfourcc(pb, "data"); avio_wb32(pb, 0x15); avio_wb32(pb, 0); - if (len==4) avio_wb32(pb, num); + if (len==8) avio_wb64(pb, num); + else if (len==4) avio_wb32(pb, num); + else if (len==2) avio_wb16(pb, num); else avio_w8 (pb, num); return size; @@ -4771,6 +4881,8 @@ static int mov_write_ilst_tag(AVIOContext *pb, MOVMuxContext *mov, mov_write_string_metadata(s, pb, "\251lyr", "lyrics" , 1); mov_write_string_metadata(s, pb, "desc", "description",1); mov_write_string_metadata(s, pb, "ldes", "synopsis" , 1); + mov_write_string_metadata(s, pb, "sdes", "series_description", 1); + mov_write_string_metadata(s, pb, "rtng", "rating", 1); mov_write_string_metadata(s, pb, "tvsh", "show" , 1); mov_write_string_metadata(s, pb, "tven", "episode_id",1); mov_write_string_metadata(s, pb, "tvnn", "network" , 1); @@ -4781,10 +4893,67 @@ static int mov_write_ilst_tag(AVIOContext *pb, MOVMuxContext *mov, mov_write_int8_metadata (s, pb, "hdvd", "hd_video", 1); mov_write_int8_metadata (s, pb, "pgap", "gapless_playback",1); mov_write_int8_metadata (s, pb, "cpil", "compilation", 1); + + mov_write_string_metadata(s, pb, "\251st3", "subtitle" , 1); + mov_write_string_metadata(s, pb, "\251des", "song_description", 1); + mov_write_string_metadata(s, pb, "\251dir", "director" , 1); + mov_write_string_metadata(s, pb, "\251ard", "art_director" , 1); + mov_write_string_metadata(s, pb, "\251arg", "arranger" , 1); + mov_write_string_metadata(s, pb, "\251aut", "author" , 1); + mov_write_string_metadata(s, pb, "\251cak", "acknowledgement" , 1); + mov_write_string_metadata(s, pb, "\251con", "conductor" , 1); + + mov_write_string_metadata(s, pb, "\251wrk", "work_name" , 1); + mov_write_string_metadata(s, pb, "\251mvn", "movement_name" , 1); + mov_write_int8_metadata (s, pb, "\251mvi", "movement_number", 2); + mov_write_int8_metadata (s, pb, "\251mvc", "movement_count" , 2); + mov_write_int8_metadata (s, pb, "shwm", "show_work_and_movement", 1); + + mov_write_string_metadata(s, pb, "\251lnt", "linear_notes" , 1); + mov_write_string_metadata(s, pb, "\251mak", "make" , 1); // Record company + mov_write_string_metadata(s, pb, "\251ope", "original_artist" , 1); + mov_write_string_metadata(s, pb, "\251phg", "phonogram_rights" , 1); + mov_write_string_metadata(s, pb, "\251prd", "producer" , 1); + mov_write_string_metadata(s, pb, "\251prf", "performers" , 1); + mov_write_string_metadata(s, pb, "\251pub", "publisher" , 1); + mov_write_string_metadata(s, pb, "\251sne", "sound_engineer" , 1); + mov_write_string_metadata(s, pb, "\251sol", "soloist" , 1); + mov_write_string_metadata(s, pb, "\251src", "original_source" , 1); // Credits + mov_write_string_metadata(s, pb, "\251thx", "thanks" , 1); + mov_write_string_metadata(s, pb, "\251url", "URL" , 1); // Online extras + mov_write_string_metadata(s, pb, "\251xpd", "executive_producer", 1); + + mov_write_string_metadata(s, pb, "sonm", "sort_name" , 1); + mov_write_string_metadata(s, pb, "soar", "sort_artist" , 1); + mov_write_string_metadata(s, pb, "soaa", "sort_album_artist", 1); + mov_write_string_metadata(s, pb, "soal", "sort_album" , 1); + mov_write_string_metadata(s, pb, "soco", "sort_composer" , 1); + mov_write_string_metadata(s, pb, "sosn", "sort_show" , 1); + + mov_write_string_metadata(s, pb, "\251enc", "encoder" , 1); // Encoded by + mov_write_string_metadata(s, pb, "purd", "purchase_date" , 1); + + mov_write_int8_metadata (s, pb, "itnu", "itunes_u" , 1); + mov_write_int8_metadata (s, pb, "pcst", "podcast" , 1); + mov_write_string_metadata(s, pb, "catg", "category" , 1); + + mov_write_string_metadata(s, pb, "apID", "account_id" , 1); + mov_write_int8_metadata (s, pb, "akID", "account_type" , 1); + mov_write_int8_metadata (s, pb, "sfID", "itunes_country", 4); + mov_write_int8_metadata (s, pb, "cnID", "content_id" , 4); + mov_write_int8_metadata (s, pb, "atID", "artist_id" , 4); + mov_write_int8_metadata (s, pb, "plID", "playlist_id" , 8); + mov_write_int8_metadata (s, pb, "geID", "genre_id" , 4); + mov_write_int8_metadata (s, pb, "cmID", "composer_id" , 4); + mov_write_string_metadata(s, pb, "xid ", "xid" , 1); + mov_write_covr(pb, s); mov_write_trkn_tag(pb, mov, s, 0); // track number mov_write_trkn_tag(pb, mov, s, 1); // disc number mov_write_tmpo_tag(pb, s); + mov_write_itmf_tag(pb, s, "iTunEXTC", "iTunEXTC"); + mov_write_itmf_tag(pb, s, "iTunMOVI", "iTunMOVI"); + return update_size(pb, pos); } @@ -4916,35 +5085,6 @@ static int ascii_to_wc(AVIOContext *pb, const uint8_t *b) return 0; } -static uint16_t language_code(const char *str) -{ - return (((str[0] - 0x60) & 0x1F) << 10) + - (((str[1] - 0x60) & 0x1F) << 5) + - (( str[2] - 0x60) & 0x1F); -} - -static int mov_write_3gp_udta_tag(AVIOContext *pb, AVFormatContext *s, - const char *tag, const char *str) -{ - int64_t pos = avio_tell(pb); - AVDictionaryEntry *t = av_dict_get(s->metadata, str, NULL, 0); - if (!t || !utf8len(t->value)) - return 0; - avio_wb32(pb, 0); /* size */ - ffio_wfourcc(pb, tag); /* type */ - avio_wb32(pb, 0); /* version + flags */ - if (!strcmp(tag, "yrrc")) - avio_wb16(pb, atoi(t->value)); - else { - avio_wb16(pb, language_code("eng")); /* language */ - avio_write(pb, t->value, strlen(t->value) + 1); /* UTF8 string value */ - if (!strcmp(tag, "albm") && - (t = av_dict_get(s->metadata, "track", NULL, 0))) - avio_w8(pb, atoi(t->value)); - } - return update_size(pb, pos); -} - static int mov_write_chpl_tag(AVIOContext *pb, AVFormatContext *s) { int64_t pos = avio_tell(pb); @@ -4983,14 +5123,14 @@ static int mov_write_udta_tag(AVIOContext *pb, MOVMuxContext *mov, return ret; if (mov->mode & MODE_3GP) { - mov_write_3gp_udta_tag(pb_buf, s, "perf", "artist"); - mov_write_3gp_udta_tag(pb_buf, s, "titl", "title"); - mov_write_3gp_udta_tag(pb_buf, s, "auth", "author"); - mov_write_3gp_udta_tag(pb_buf, s, "gnre", "genre"); - mov_write_3gp_udta_tag(pb_buf, s, "dscp", "comment"); - mov_write_3gp_udta_tag(pb_buf, s, "albm", "album"); - mov_write_3gp_udta_tag(pb_buf, s, "cprt", "copyright"); - mov_write_3gp_udta_tag(pb_buf, s, "yrrc", "date"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "perf", "artist"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "titl", "title"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "auth", "author"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "gnre", "genre"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "dscp", "comment"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "albm", "album"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "cprt", "copyright"); + mov_write_3gp_udta_tag(pb_buf, s->metadata, "yrrc", "date"); mov_write_loci_tag(s, pb_buf); } else if (mov->mode == MODE_MOV && !(mov->flags & FF_MOV_FLAG_USE_MDTA)) { // the title field breaks gtkpod with mp4 and my suspicion is that stuff is not valid in mp4 mov_write_string_metadata(s, pb_buf, "\251ART", "artist", 0); @@ -7542,6 +7682,7 @@ static int mov_create_chapter_track(AVFormatContext *s, int tracknum) track->tag = MKTAG('t','e','x','t'); track->timescale = mov->movie_timescale; track->par = avcodec_parameters_alloc(); + track->language = mov->tracks[0].language; if (!track->par) return AVERROR(ENOMEM); track->par->codec_type = AVMEDIA_TYPE_SUBTITLE; @@ -7779,7 +7920,7 @@ static int mov_create_dvd_sub_decoder_specific_info(MOVTrack *track, int i, width = 720, height = 480; int have_palette = 0, have_size = 0; uint32_t palette[16]; - char *cur = track->extradata[track->last_stsd_index]; + char *cur = st->codecpar->extradata; while (cur && *cur) { if (strncmp("palette:", cur, 8) == 0) { diff --git a/libavutil/frame.h b/libavutil/frame.h index 771c9ce..4aba0c8 100644 --- a/libavutil/frame.h +++ b/libavutil/frame.h @@ -200,6 +200,12 @@ enum AVFrameSideDataType { */ AV_FRAME_DATA_DOVI_RPU_BUFFER, + /** + * Dolby Vision RPU ITU T35 raw data, suitable for passing to SVT-AV1 + * or other libraries. Array of uint8_t. + */ + AV_FRAME_DATA_DOVI_RPU_BUFFER_T35, + /** * Parsed Dolby Vision metadata, suitable for passing to a software * implementation. The payload is the AVDOVIMetadata struct defined in diff --git a/libavutil/hwcontext_d3d11va.c b/libavutil/hwcontext_d3d11va.c index 418fe9d..4369d80 100644 --- a/libavutil/hwcontext_d3d11va.c +++ b/libavutil/hwcontext_d3d11va.c @@ -85,6 +85,9 @@ typedef struct D3D11VAFramesContext { int nb_surfaces; int nb_surfaces_used; + int max_retries; + int retries; + DXGI_FORMAT format; ID3D11Texture2D *staging_texture; @@ -260,7 +263,9 @@ static AVBufferRef *d3d11va_pool_alloc(void *opaque, size_t size) ID3D11Texture2D_GetDesc(hwctx->texture, &texDesc); if (s->nb_surfaces_used >= texDesc.ArraySize) { - av_log(ctx, AV_LOG_ERROR, "Static surface pool size exceeded.\n"); + if (s->retries >= s->max_retries) { + av_log(ctx, AV_LOG_ERROR, "Static surface pool size exceeded.\n"); + } return NULL; } @@ -346,20 +351,34 @@ static int d3d11va_frames_init(AVHWFramesContext *ctx) static int d3d11va_get_buffer(AVHWFramesContext *ctx, AVFrame *frame) { AVD3D11FrameDescriptor *desc; + D3D11VAFramesContext *s = ctx->hwctx; + s->retries = 0; + s->max_retries = 50; - frame->buf[0] = av_buffer_pool_get(ctx->pool); - if (!frame->buf[0]) - return AVERROR(ENOMEM); - - desc = (AVD3D11FrameDescriptor *)frame->buf[0]->data; - - frame->data[0] = (uint8_t *)desc->texture; - frame->data[1] = (uint8_t *)desc->index; - frame->format = AV_PIX_FMT_D3D11; - frame->width = ctx->width; - frame->height = ctx->height; + /** + * Loop until a buffer becomes available from the pool. + * In a full hardware pipeline, all buffers may be temporarily in use by + * other modules (encoder/filter/decoder). Rather than immediately failing + * with ENOMEM, we wait for a buffer to be released back to the pool, which + * maintains pipeline flow and prevents unnecessary allocation failures + * during normal operation. + */ + while (s->retries < s->max_retries) { + frame->buf[0] = av_buffer_pool_get(ctx->pool); + if (frame->buf[0]) { + desc = (AVD3D11FrameDescriptor *)frame->buf[0]->data; + frame->data[0] = (uint8_t *)desc->texture; + frame->data[1] = (uint8_t *)desc->index; + frame->format = AV_PIX_FMT_D3D11; + frame->width = ctx->width; + frame->height = ctx->height; + return 0; + } - return 0; + av_usleep(1000); + } + + return AVERROR(ENOMEM); } static int d3d11va_transfer_get_formats(AVHWFramesContext *ctx, diff --git a/libavutil/hwcontext_videotoolbox.c b/libavutil/hwcontext_videotoolbox.c index 2764097..9c5c999 100644 --- a/libavutil/hwcontext_videotoolbox.c +++ b/libavutil/hwcontext_videotoolbox.c @@ -191,7 +191,8 @@ static int vt_pool_alloc(AVHWFramesContext *ctx) VTFramesContext *fctx = ctx->hwctx; AVVTFramesContext *hw_ctx = &fctx->p; CVReturn err; - CFNumberRef w, h, pixfmt; + CFNumberRef w, h, extend_w_num, extend_h_num, pixfmt; + int extend_w, extend_h; uint32_t cv_pixfmt; CFMutableDictionaryRef attributes, iosurface_properties; @@ -224,6 +225,15 @@ static int vt_pool_alloc(AVHWFramesContext *ctx) CFRelease(w); CFRelease(h); + extend_w = FFALIGN(ctx->width, 16) - ctx->width; + extend_h = FFALIGN(ctx->height, 16) - ctx->height; + extend_w_num = CFNumberCreate(NULL, kCFNumberSInt32Type, &extend_w); + extend_h_num = CFNumberCreate(NULL, kCFNumberSInt32Type, &extend_h); + CFDictionarySetValue(attributes, kCVPixelBufferExtendedPixelsRightKey, extend_w_num); + CFDictionarySetValue(attributes, kCVPixelBufferExtendedPixelsBottomKey, extend_h_num); + CFRelease(extend_w_num); + CFRelease(extend_h_num); + err = CVPixelBufferPoolCreate( NULL, NULL, diff --git a/libavutil/side_data.c b/libavutil/side_data.c index bbbeb70..c635365 100644 --- a/libavutil/side_data.c +++ b/libavutil/side_data.c @@ -42,6 +42,7 @@ static const AVSideDataDescriptor sd_props[] = { [AV_FRAME_DATA_DETECTION_BBOXES] = { "Bounding boxes for object detection and classification", AV_SIDE_DATA_PROP_SIZE_DEPENDENT }, [AV_FRAME_DATA_DOVI_RPU_BUFFER] = { "Dolby Vision RPU Data", AV_SIDE_DATA_PROP_COLOR_DEPENDENT }, [AV_FRAME_DATA_DOVI_METADATA] = { "Dolby Vision Metadata", AV_SIDE_DATA_PROP_COLOR_DEPENDENT }, + [AV_FRAME_DATA_DOVI_RPU_BUFFER_T35] = { "Dolby Vision RPU ITU T35 Data", AV_SIDE_DATA_PROP_COLOR_DEPENDENT }, [AV_FRAME_DATA_LCEVC] = { "LCEVC NAL data", AV_SIDE_DATA_PROP_SIZE_DEPENDENT }, [AV_FRAME_DATA_VIEW_ID] = { "View ID" }, [AV_FRAME_DATA_STEREO3D] = { "Stereo 3D", AV_SIDE_DATA_PROP_GLOBAL }, -- 2.54.0