February 2026 - ffmpeg-devel - lists.ffmpeg.org

[PATCH] avformat/demux: avoid unconditional ID3v2 tag consumption
by Nil Fons Miret 10 Jul '26

10 Jul '26

I have found a few instances of rawvideo files that coincidentally start with a header matching the ID3v2 header format, and ffmpeg consumes the header before demuxing, causing a decoding error. This happens even if "-f rawvideo" is specified. This patch limits the formats for which this automatic consuming of ID3v2 headers is done. Note: I am not an expert in ID3v2 by any means, and I'm happy to accept suggestions, both on the scope and mechanism of the solution. To reproduce, you can easily generate an example as follows: ``` ( echo "49 44 33 2B 98 3A 49 44 33 2B 98 3A 6A 44 54 2B" | xxd -r -p dd if=/dev/zero bs=1 count=115184 2>/dev/null ) > id3v2_320x240_yuv420p.yuv ffmpeg -f rawvideo -s 320x240 -pix_fmt yuv420p -i id3v2_320x240_yuv420p.yuv -f null - ``` This generates a 320x240 yuv420p file with a 16-byte header that matches the ID3v2 header format (this was copied from a real case I saw) and completed with zeroes for a total of 115200 bytes, the right size for this resolution and pixel format. On the master branch, I see: ``` [...] ID3v2.43 tag skipped, cannot handle version [...] [out#0/null @ 0x15700a6e0] Output file is empty, nothing was encoded(check -ss / -t / -frames parameters if used) frame= 0 fps=0.0 q=0.0 Lsize=N/A time=N/A bitrate=N/A speed=N/A ``` After the changes: it decodes as expected. ``` [...] frame= 1 fps=0.0 q=-0.0 Lsize=N/A time=00:00:00.04 bitrate=N/A speed=16.5x elapsed=0:00:00.00 ``` Thanks in advance, Nil

3 6

[PATCH] ffbuild/common.mak: ensure target directories are created before running shell redirects into them
by Alexander Kanavin 19 Jun '26

19 Jun '26

Otherwise, occasional build races have been observed: https://autobuilder.yoctoproject.org/valkyrie/#/builders/37/builds/3001/ste… /bin/sh: 4: cannot create fftools/resources/graph.css.min: Directory nonexistent mkdir -p fftools/graph /bin/sh: 1: cannot create fftools/resources/graph.html.gz: Directory nonexistent make: *** [/srv/pokybuild/.../ffmpeg-8.0.1/ffbuild/common.mak:165: fftools/resources/graph.html.gz] Error 2 make: *** Waiting for unfinished jobs.... make: *** [/srv/pokybuild/.../ffmpeg-8.0.1/ffbuild/common.mak:145: fftools/resources/graph.css.min] Error 2 There's a separate rule for making those directories, but unfortunately it's racing with the rules that expect the directories to exist. Rather than add a Makefile dependency, I've injected the dir creation directly in front of commands that can otherwise fail - a proper fix would probably add the rule rather. Signed-off-by: Alexander Kanavin <alex(a)linutronix.de> --- ffbuild/common.mak | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ffbuild/common.mak b/ffbuild/common.mak index 89c0c413e1..7f223297b8 100644 --- a/ffbuild/common.mak +++ b/ffbuild/common.mak @@ -125,8 +125,8 @@ $(BIN2CEXE): ffbuild/bin2c_host.o $(HOSTLD) $(HOSTLDFLAGS) $(HOSTLD_O) $^ $(HOSTEXTRALIBS) RUN_BIN2C = $(BIN2C) $(patsubst $(SRC_PATH)/%,$(SRC_LINK)/%,$<) $@ $(subst .,_,$(basename $(notdir $@))) -RUN_GZIP = $(M)gzip -nc9 $(patsubst $(SRC_PATH)/%,$(SRC_LINK)/%,$<) >$@ -RUN_MINIFY = $(M)sed 's!/\\*.*\\*/!!g' $< | tr '\n' ' ' | tr -s ' ' | sed 's/^ //; s/ $$//' > $@ +RUN_GZIP = mkdir -p $(dir $@) && $(M)gzip -nc9 $(patsubst $(SRC_PATH)/%,$(SRC_LINK)/%,$<) >$@ +RUN_MINIFY = mkdir -p $(dir $@) && $(M)sed 's!/\\*.*\\*/!!g' $< | tr '\n' ' ' | tr -s ' ' | sed 's/^ //; s/ $$//' > $@ %.gz: TAG = GZIP %.min: TAG = MINIFY -- 2.47.3

2 1

[PATCH] avformat/mpjpegdec: add support for X-Timestamp and X-Framerate headers
by Vladimir Sobolev 11 Jun '26

11 Jun '26

From 31d73c6774c1ea6d621db57f26439e297cc23c3d Mon Sep 17 00:00:00 2001 From: Vladimir Sobolev <v.sobolev(a)gmail.com> Date: Sun, 9 Nov 2025 02:28:13 +0200 Subject: [PATCH] avformat/mpjpegdec: add support for X-Timestamp and X-Framerate headers Add support for parsing X-Timestamp and X-Framerate headers from HTTP multipart MJPEG streams. These headers allow servers to provide accurate timestamps and framerate information for each frame. Changes: - Parse X-Timestamp header (in seconds) and set packet PTS/DTS - Parse X-Framerate/X-FrameRate header and update stream framerate - Maintain backward compatibility (defaults to 25 fps if not provided) - Add debug logging for parsed header values This enables proper timestamp handling for MJPEG streams that provide timing information in HTTP headers, improving synchronization accuracy. --- libavformat/mpjpegdec.c | 69 ++++++++++++++++++++++++++++++++++++++--- 1 file changed, 65 insertions(+), 4 deletions(-) diff --git a/libavformat/mpjpegdec.c b/libavformat/mpjpegdec.c index 125b17585e..c90d7a2ad4 100644 --- a/libavformat/mpjpegdec.c +++ b/libavformat/mpjpegdec.c @@ -22,6 +22,9 @@ #include "libavutil/avstring.h" #include "libavutil/mem.h" #include "libavutil/opt.h" +#include "libavutil/parseutils.h" +#include "libavutil/eval.h" +#include "libavutil/intfloat.h" #include "avformat.h" #include "demux.h" @@ -34,6 +37,11 @@ typedef struct MPJPEGDemuxContext { char *searchstr; int searchstr_len; int strict_mime_boundary; + AVRational framerate; /* framerate from X-Framerate header */ + int64_t timestamp; /* timestamp from X-Timestamp header */ + int has_timestamp; /* flag indicating if timestamp was set */ + int framerate_set; /* flag indicating if framerate was set in header */ + int framerate_applied; /* flag indicating if framerate was applied to stream */ } MPJPEGDemuxContext; static void trim_right(char *p) @@ -97,7 +105,8 @@ static int split_tag_value(char **tag, char **value, char *line) static int parse_multipart_header(AVIOContext *pb, int* size, const char* expected_boundary, - void *log_ctx); + void *log_ctx, + MPJPEGDemuxContext *mpjpeg); static int mpjpeg_read_close(AVFormatContext *s) { @@ -118,7 +127,7 @@ static int mpjpeg_read_probe(const AVProbeData *p) ffio_init_read_context(&pb, p->buf, p->buf_size); - ret = (parse_multipart_header(&pb.pub, &size, "--", NULL) >= 0) ? AVPROBE_SCORE_MAX : 0; + ret = (parse_multipart_header(&pb.pub, &size, "--", NULL, NULL) >= 0) ? AVPROBE_SCORE_MAX : 0; return ret; } @@ -146,6 +155,12 @@ static int mpjpeg_read_header(AVFormatContext *s) st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO; st->codecpar->codec_id = AV_CODEC_ID_MJPEG; + /* Default framerate is 25 fps, will be updated from headers if available */ + MPJPEGDemuxContext *mpjpeg = s->priv_data; + mpjpeg->framerate = (AVRational){25, 1}; + mpjpeg->framerate_set = 0; + mpjpeg->framerate_applied = 0; + mpjpeg->has_timestamp = 0; avpriv_set_pts_info(st, 60, 1, 25); avio_seek(s->pb, pos, SEEK_SET); @@ -167,7 +182,8 @@ static int parse_content_length(const char *value) static int parse_multipart_header(AVIOContext *pb, int* size, const char* expected_boundary, - void *log_ctx) + void *log_ctx, + MPJPEGDemuxContext *mpjpeg) { char line[128]; int found_content_type = 0; @@ -235,6 +251,33 @@ static int parse_multipart_header(AVIOContext *pb, av_log(log_ctx, AV_LOG_WARNING, "Invalid Content-Length value : %s\n", value); + } else if (mpjpeg && !av_strcasecmp(tag, "X-Timestamp")) { + double ts = av_strtod(value, NULL); + if (!isnan(ts) && isfinite(ts)) { + /* X-Timestamp is in seconds, convert to AV_TIME_BASE */ + mpjpeg->timestamp = (int64_t)(ts * AV_TIME_BASE); + mpjpeg->has_timestamp = 1; + if (log_ctx) + av_log(log_ctx, AV_LOG_DEBUG, + "Parsed X-Timestamp: %s -> %"PRId64" (%.6f seconds)\n", + value, mpjpeg->timestamp, ts); + } else if (log_ctx) { + av_log(log_ctx, AV_LOG_WARNING, + "Invalid X-Timestamp value : %s\n", value); + } + } else if (mpjpeg && (!av_strcasecmp(tag, "X-Framerate") || !av_strcasecmp(tag, "X-FrameRate"))) { + AVRational fps = {0}; + if (av_parse_video_rate(&fps, value) >= 0 && fps.num > 0 && fps.den > 0) { + mpjpeg->framerate = fps; + mpjpeg->framerate_set = 1; + if (log_ctx) + av_log(log_ctx, AV_LOG_DEBUG, + "Parsed X-Framerate: %s -> %d/%d fps\n", + value, fps.num, fps.den); + } else if (log_ctx) { + av_log(log_ctx, AV_LOG_WARNING, + "Invalid X-Framerate value : %s\n", value); + } } } @@ -311,10 +354,21 @@ static int mpjpeg_read_packet(AVFormatContext *s, AVPacket *pkt) mpjpeg->searchstr_len = strlen(mpjpeg->searchstr); } - ret = parse_multipart_header(s->pb, &size, mpjpeg->boundary, s); + /* Reset timestamp flag for each packet */ + mpjpeg->has_timestamp = 0; + + ret = parse_multipart_header(s->pb, &size, mpjpeg->boundary, s, mpjpeg); if (ret < 0) return ret; + /* Update framerate if it was set in header and hasn't been applied yet */ + if (mpjpeg->framerate_set && !mpjpeg->framerate_applied && s->nb_streams > 0) { + AVStream *st = s->streams[0]; + st->avg_frame_rate = mpjpeg->framerate; + avpriv_set_pts_info(st, 60, mpjpeg->framerate.den, mpjpeg->framerate.num); + mpjpeg->framerate_applied = 1; + } + if (size > 0) { /* size has been provided to us in MIME header */ ret = av_get_packet(s->pb, pkt, size); @@ -353,6 +407,13 @@ static int mpjpeg_read_packet(AVFormatContext *s, AVPacket *pkt) } } + /* Set timestamp from X-Timestamp header if available */ + if (ret >= 0 && mpjpeg->has_timestamp && s->nb_streams > 0) { + AVStream *st = s->streams[0]; + pkt->pts = av_rescale_q(mpjpeg->timestamp, AV_TIME_BASE_Q, st->time_base); + pkt->dts = pkt->pts; + } + return ret; } -- 2.50.1 (Apple Git-155)

2 11

[PATCH] avfilter: enhance command processing with chain propagation and direction control. (PR #20731)
by cenzhanquan1 05 Jun '26

05 Jun '26

PR #20731 opened by cenzhanquan1 URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20731 Patch URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20731.patch The existing avfilter_process_command and avfilter_graph_send_command havelimitations in command propagation: they only handle commands for individualfilters and lack the ability to propagate commands through the entire filterchain, nor do they support directional traversal of the filter graph. Thismakes it difficult to control multiple filters in a chain (e.g., adjustingvolume for all related filters or enabling/disabling a series of filters)with a single command. This patch enhances the command processing logic to address these issues: Add two new flags to control command propagation: AVFILTER_CMD_FLAG_CHAIN: Enables command propagation through the entire filter chain. After processing the current filter, it traverses all associated links (inputs/outputs based on direction) and recursively forwards the command to subsequent filters, covering the full filter topology. AVFILTER_CMD_FLAG_REVERSE: Works with AVFILTER_CMD_FLAG_CHAIN to control traversal direction. Default (forward) follows data flow (source → destination filters), while reverse traversal goes against data flow (destination → source filters). Refactor avfilter_process_command to integrate chain propagation logic: Process the command for the current filter first, then check if chain propagation is enabled. Traverse all relevant links (inputs for reverse, outputs for forward) and recursively forward the command to next-level filters. Track processing status (processed) to determine if any filter in the chain handled the command. Respect AVFILTER_CMD_FLAG_ONE to stop propagation once a filter processes the command, and propagate critical errors. Improve command handling for built-in commands ("ping" and "enable") to workseamlessly with the new propagation logic, ensuring consistent behavioracross the chain. These changes enable flexible command control over entire filter chains,supporting use cases like batch adjustment of filters, topology-wide statuschecks (via "ping"), and coordinated enable/disable operations, whilemaintaining compatibility with existing filter command implementations. Signed-off-by: cenzhanquan1 <cenzhanquan1(a)xiaomi.com> From 9dba682222bd55c952e14e71789a7084529a74f7 Mon Sep 17 00:00:00 2001 From: cenzhanquan1 <cenzhanquan1(a)xiaomi.com> Date: Tue, 21 Oct 2025 17:34:22 +0800 Subject: [PATCH] avfilter: enhance command processing with chain propagation and direction control. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The existing avfilter_process_command and avfilter_graph_send_command havelimitations in command propagation: they only handle commands for individualfilters and lack the ability to propagate commands through the entire filterchain, nor do they support directional traversal of the filter graph. Thismakes it difficult to control multiple filters in a chain (e.g., adjustingvolume for all related filters or enabling/disabling a series of filters)with a single command. This patch enhances the command processing logic to address these issues: Add two new flags to control command propagation: AVFILTER_CMD_FLAG_CHAIN: Enables command propagation through the entire filter chain. After processing the current filter, it traverses all associated links (inputs/outputs based on direction) and recursively forwards the command to subsequent filters, covering the full filter topology. AVFILTER_CMD_FLAG_REVERSE: Works with AVFILTER_CMD_FLAG_CHAIN to control traversal direction. Default (forward) follows data flow (source → destination filters), while reverse traversal goes against data flow (destination → source filters). Refactor avfilter_process_command to integrate chain propagation logic: Process the command for the current filter first, then check if chain propagation is enabled. Traverse all relevant links (inputs for reverse, outputs for forward) and recursively forward the command to next-level filters. Track processing status (processed) to determine if any filter in the chain handled the command. Respect AVFILTER_CMD_FLAG_ONE to stop propagation once a filter processes the command, and propagate critical errors. Improve command handling for built-in commands ("ping" and "enable") to workseamlessly with the new propagation logic, ensuring consistent behavioracross the chain. These changes enable flexible command control over entire filter chains,supporting use cases like batch adjustment of filters, topology-wide statuschecks (via "ping"), and coordinated enable/disable operations, whilemaintaining compatibility with existing filter command implementations. Signed-off-by: cenzhanquan1 <cenzhanquan1(a)xiaomi.com> --- libavfilter/avfilter.c | 75 +++++++++++++++++++++++++++++++++--------- libavfilter/avfilter.h | 10 ++++-- 2 files changed, 67 insertions(+), 18 deletions(-) diff --git a/libavfilter/avfilter.c b/libavfilter/avfilter.c index 169c2baa42..68ec9f6f6a 100644 --- a/libavfilter/avfilter.c +++ b/libavfilter/avfilter.c @@ -607,25 +607,68 @@ static int set_enable_expr(FFFilterContext *ctxi, const char *expr) return 0; } -int avfilter_process_command(AVFilterContext *filter, const char *cmd, const char *arg, char *res, int res_len, int flags) +int avfilter_process_command(AVFilterContext *filter, const char *cmd, const char *arg, + char *res, int res_len, int flags) { - if(!strcmp(cmd, "ping")){ - char local_res[256] = {0}; + int direction = flags & AVFILTER_CMD_FLAG_REVERSE; + int ret = AVERROR(ENOSYS); + int processed = 0; - if (!res) { - res = local_res; - res_len = sizeof(local_res); - } - av_strlcatf(res, res_len, "pong from:%s %s\n", filter->filter->name, filter->name); - if (res == local_res) - av_log(filter, AV_LOG_INFO, "%s", res); - return 0; - }else if(!strcmp(cmd, "enable")) { - return set_enable_expr(fffilterctx(filter), arg); - }else if (fffilter(filter->filter)->process_command) { - return fffilter(filter->filter)->process_command(filter, cmd, arg, res, res_len, flags); + int process_flags = flags & ~AVFILTER_CMD_FLAG_CHAIN; + if (!strcmp(cmd, "ping")) { + char local_res[256] = {0}; + char *res_buf = res ? res : local_res; + size_t buf_len = res ? res_len : sizeof(local_res); + av_strlcatf(res_buf, buf_len, "pong from:%s %s\n", + filter->filter->name, filter->name ? filter->name : "unknown"); + if (!res) + av_log(filter, AV_LOG_INFO, "%s", res_buf); + ret = 0; + } else if (!strcmp(cmd, "enable")) { + ret = set_enable_expr(fffilterctx(filter), arg); + } else if (fffilter(filter->filter)->process_command) { + ret = fffilter(filter->filter)->process_command(filter, cmd, arg, res, res_len, process_flags); + } else { + ret = AVERROR(ENOSYS); } - return AVERROR(ENOSYS); + + if (ret != AVERROR(ENOSYS)) { + processed = 1; + if ((flags & AVFILTER_CMD_FLAG_ONE) || ret < 0) { + return ret; + } + } + + if (!(flags & AVFILTER_CMD_FLAG_CHAIN)) { + return processed ? 0 : AVERROR(ENOSYS); + } + + av_log(filter, AV_LOG_DEBUG, + "cmd_chain: [%s] dir:%s -> '%s' '%s' (forwarding)\n", + filter->name ? filter->name : "unknown", + direction ? "reverse" : "forward", + cmd, arg ? arg : ""); + + unsigned nb_links = direction ? filter->nb_inputs : filter->nb_outputs; + for (int i = 0; i < nb_links; i++) { + AVFilterLink *link = direction ? filter->inputs[i] : filter->outputs[i]; + AVFilterContext *next_filter = direction ? (link ? link->src : NULL) : (link ? link->dst : NULL); + + if (!link || !next_filter) { + av_log(filter, AV_LOG_DEBUG, "Invalid %s link at pad %d\n", + direction ? "input" : "output", i); + continue; + } + + ret = avfilter_process_command(next_filter, cmd, arg, res, res_len, flags); + if (ret >= 0) { + processed = 1; + } else if (ret != AVERROR(ENOSYS)) { + return ret; + } + } + + return processed ? 0 : AVERROR(ENOSYS); } unsigned avfilter_filter_pad_count(const AVFilter *filter, int is_output) diff --git a/libavfilter/avfilter.h b/libavfilter/avfilter.h index 02b58c42c2..fde3811dc3 100644 --- a/libavfilter/avfilter.h +++ b/libavfilter/avfilter.h @@ -466,8 +466,14 @@ struct AVFilterLink { int avfilter_link(AVFilterContext *src, unsigned srcpad, AVFilterContext *dst, unsigned dstpad); -#define AVFILTER_CMD_FLAG_ONE 1 ///< Stop once a filter understood the command (for target=all for example), fast filters are favored automatically -#define AVFILTER_CMD_FLAG_FAST 2 ///< Only execute command when its fast (like a video out that supports contrast adjustment in hw) +#define AVFILTER_CMD_FLAG_ONE 1 ///< Stop once a filter understood the command (for target=all for example), fast filters are favored automatically +#define AVFILTER_CMD_FLAG_FAST 2 ///< Only execute command when its fast (like a video out that supports contrast adjustment in hw) +#define AVFILTER_CMD_FLAG_CHAIN 4 ///< Propagate the command through the entire filter chain. After processing the current filter, + /// traverse all its associated links (inputs or outputs, based on direction) and recursively + /// forward the command to subsequent filters, covering the full filter topology. +#define AVFILTER_CMD_FLAG_REVERSE 8 ///< Only effective when paired with AVFILTER_CMD_FLAG_CHAIN. Changes the command's traversal + /// direction in the chain: default (forward) follows data flow (source → destination filters), + /// while reverse traversal opposes data flow (destination → source filters). /** * Make the filter instance process a command. -- 2.49.1

5 11

[PATCH] avcodec/x86/h264_intrapred: add AVX2 for 10-bit pred16x16 functions
by ZaneHam 12 May '26

12 May '26

Add AVX2 implementations for 10-bit H.264 16x16 intra prediction: - pred16x16_vertical_10 - pred16x16_horizontal_10 - pred16x16_dc_10 - pred16x16_top_dc_10 - pred16x16_left_dc_10 - pred16x16_128_dc_10 10-bit 16x16 blocks are 32 bytes per row, perfectly matching AVX2's 256-bit YMM registers, allowing single-instruction row operations versus two XMM operations with SSE2. checkasm benchmarks on Zen3 (cycles, lower is better): C SSE2 AVX2 pred16x16_dc_10 65.7 40.3 27.3 (1.48x vs SSE2) pred16x16_128_dc_10 31.1 28.1 21.4 (1.31x vs SSE2) pred16x16_horizontal 67.8 28.1 21.6 (1.30x vs SSE2) pred16x16_left_dc_10 55.6 35.0 22.9 (1.53x vs SSE2) pred16x16_top_dc_10 49.5 32.3 21.8 (1.48x vs SSE2) pred16x16_vertical_10 32.3 28.3 24.1 (1.17x vs SSE2) --- libavcodec/x86/h264_intrapred_10bit.asm | 186 ++++++++++++++++++++++++ libavcodec/x86/h264_intrapred_init.c | 14 ++ 2 files changed, 200 insertions(+) diff --git a/libavcodec/x86/h264_intrapred_10bit.asm b/libavcodec/x86/h264_intrapred_10bit.asm index 2f30807332..78e2f263bc 100644 --- a/libavcodec/x86/h264_intrapred_10bit.asm +++ b/libavcodec/x86/h264_intrapred_10bit.asm @@ -1117,3 +1117,189 @@ cglobal pred16x16_128_dc_10, 2,3 dec r2d jg .loop RET + +;----------------------------------------------------------------------------- +; AVX2 versions of pred16x16 10-bit functions +; For 10-bit: 16 pixels * 2 bytes = 32 bytes = 1 YMM register (perfect match\!) +;----------------------------------------------------------------------------- + +%if HAVE_AVX2_EXTERNAL + +;----------------------------------------------------------------------------- +; void ff_pred16x16_vertical_10_avx2(pixel *src, ptrdiff_t stride) +;----------------------------------------------------------------------------- +INIT_YMM avx2 +cglobal pred16x16_vertical_10, 2, 4 + sub r0, r1 + movu m0, [r0] ; Load all 16 pixels (32 bytes) from top row + mov r2d, 4 + lea r3, [r1*3] +.loop: + movu [r0+r1*1], m0 + movu [r0+r1*2], m0 + movu [r0+r3 ], m0 + lea r0, [r0+r1*2] + movu [r0+r1*2], m0 + lea r0, [r0+r1*2] + dec r2d + jg .loop + RET + +;----------------------------------------------------------------------------- +; void ff_pred16x16_horizontal_10_avx2(pixel *src, ptrdiff_t stride) +;----------------------------------------------------------------------------- +INIT_YMM avx2 +cglobal pred16x16_horizontal_10, 2, 4 + lea r2, [r1*3] + mov r3d, 4 +.loop: + vpbroadcastw m0, [r0-2] + movu [r0], m0 + vpbroadcastw m0, [r0+r1-2] + movu [r0+r1], m0 + vpbroadcastw m0, [r0+r1*2-2] + movu [r0+r1*2], m0 + vpbroadcastw m0, [r0+r2-2] + movu [r0+r2], m0 + lea r0, [r0+r1*4] + dec r3d + jg .loop + RET + +;----------------------------------------------------------------------------- +; void ff_pred16x16_dc_10_avx2(pixel *src, ptrdiff_t stride) +; DC = (sum of 16 top pixels + sum of 16 left pixels + 16) >> 5 +;----------------------------------------------------------------------------- +INIT_YMM avx2 +cglobal pred16x16_dc_10, 2, 6 + mov r5, r0 ; Save dest pointer + sub r0, r1 + movu m0, [r0] ; Load top row (32 bytes) + vextracti128 xm1, m0, 1 ; Get high 128 bits + paddw xm0, xm1 ; Sum to 8 words + phaddw xm0, xm0 ; 4 words + phaddw xm0, xm0 ; 2 words + phaddw xm0, xm0 ; 1 word (top sum in low word) + movd r3d, xm0 + and r3d, 0xFFFF ; Keep only low 16 bits + + ; Sum left column using lea-based pointer advancement + lea r0, [r0+r1-2] ; Point to left pixel of row 0 + movzx r4d, word [r0] + add r3d, r4d + movzx r4d, word [r0+r1] + add r3d, r4d +%rep 7 + lea r0, [r0+r1*2] + movzx r4d, word [r0] + add r3d, r4d + movzx r4d, word [r0+r1] + add r3d, r4d +%endrep + add r3d, 16 ; Rounding + shr r3d, 5 ; Divide by 32 + + movd xm0, r3d + vpbroadcastw m0, xm0 ; Broadcast to all 16 words + + ; Fill all 16 rows + mov r3d, 4 + lea r4, [r1*3] +.loop: + movu [r5+r1*0], m0 + movu [r5+r1*1], m0 + movu [r5+r1*2], m0 + movu [r5+r4 ], m0 + lea r5, [r5+r1*4] + dec r3d + jg .loop + RET + +;----------------------------------------------------------------------------- +; void ff_pred16x16_top_dc_10_avx2(pixel *src, ptrdiff_t stride) +; DC = (sum of 16 top pixels + 8) >> 4 +;----------------------------------------------------------------------------- +INIT_YMM avx2 +cglobal pred16x16_top_dc_10, 2, 4 + sub r0, r1 + movu m0, [r0] ; Load top row + vextracti128 xm1, m0, 1 + paddw xm0, xm1 + phaddw xm0, xm0 + phaddw xm0, xm0 + phaddw xm0, xm0 + paddw xm0, [pw_8] ; Add 8 for rounding + psrlw xm0, 4 ; Divide by 16 + vpbroadcastw m0, xm0 + + mov r2d, 4 + lea r3, [r1*3] +.loop: + movu [r0+r1*1], m0 + movu [r0+r1*2], m0 + movu [r0+r3 ], m0 + lea r0, [r0+r1*2] + movu [r0+r1*2], m0 + lea r0, [r0+r1*2] + dec r2d + jg .loop + RET + +;----------------------------------------------------------------------------- +; void ff_pred16x16_left_dc_10_avx2(pixel *src, ptrdiff_t stride) +; DC = (sum of 16 left pixels + 8) >> 4 +;----------------------------------------------------------------------------- +INIT_YMM avx2 +cglobal pred16x16_left_dc_10, 2, 5 + mov r4, r0 ; Save dest pointer + + ; Sum left column using lea-based pointer advancement + sub r0, 2 ; Point to left pixel of row 0 + movzx r2d, word [r0] + movzx r3d, word [r0+r1] +%rep 7 + lea r0, [r0+r1*2] + movzx eax, word [r0] + add r2d, eax + movzx eax, word [r0+r1] + add r3d, eax +%endrep + lea r2d, [r2+r3+8] ; Sum with rounding + shr r2d, 4 ; Divide by 16 + + movd xm0, r2d + vpbroadcastw m0, xm0 + + ; Fill all 16 rows + mov r2d, 4 + lea r3, [r1*3] +.loop: + movu [r4+r1*0], m0 + movu [r4+r1*1], m0 + movu [r4+r1*2], m0 + movu [r4+r3 ], m0 + lea r4, [r4+r1*4] + dec r2d + jg .loop + RET + +;----------------------------------------------------------------------------- +; void ff_pred16x16_128_dc_10_avx2(pixel *src, ptrdiff_t stride) +; Fill with constant 512 (1 << 9 for 10-bit midpoint) +;----------------------------------------------------------------------------- +INIT_YMM avx2 +cglobal pred16x16_128_dc_10, 2, 4 + vpbroadcastw m0, [pw_512] + mov r2d, 4 + lea r3, [r1*3] +.loop: + movu [r0+r1*0], m0 + movu [r0+r1*1], m0 + movu [r0+r1*2], m0 + movu [r0+r3 ], m0 + lea r0, [r0+r1*4] + dec r2d + jg .loop + RET + +%endif ; HAVE_AVX2_EXTERNAL diff --git a/libavcodec/x86/h264_intrapred_init.c b/libavcodec/x86/h264_intrapred_init.c index aa9bc721f0..6918c7f985 100644 --- a/libavcodec/x86/h264_intrapred_init.c +++ b/libavcodec/x86/h264_intrapred_init.c @@ -97,6 +97,12 @@ PRED16x16(128_dc, 10, sse2) PRED16x16(left_dc, 10, sse2) PRED16x16(vertical, 10, sse2) PRED16x16(horizontal, 10, sse2) +PRED16x16(dc, 10, avx2) +PRED16x16(top_dc, 10, avx2) +PRED16x16(128_dc, 10, avx2) +PRED16x16(left_dc, 10, avx2) +PRED16x16(vertical, 10, avx2) +PRED16x16(horizontal, 10, avx2) /* 8-bit versions */ PRED16x16(vertical, 8, sse) @@ -328,5 +334,13 @@ av_cold void ff_h264_pred_init_x86(H264PredContext *h, int codec_id, h->pred8x8l[VERT_RIGHT_PRED ] = ff_pred8x8l_vertical_right_10_avx; h->pred8x8l[HOR_UP_PRED ] = ff_pred8x8l_horizontal_up_10_avx; } + if (EXTERNAL_AVX2(cpu_flags)) { + h->pred16x16[DC_PRED8x8 ] = ff_pred16x16_dc_10_avx2; + h->pred16x16[TOP_DC_PRED8x8 ] = ff_pred16x16_top_dc_10_avx2; + h->pred16x16[DC_128_PRED8x8 ] = ff_pred16x16_128_dc_10_avx2; + h->pred16x16[LEFT_DC_PRED8x8 ] = ff_pred16x16_left_dc_10_avx2; + h->pred16x16[VERT_PRED8x8 ] = ff_pred16x16_vertical_10_avx2; + h->pred16x16[HOR_PRED8x8 ] = ff_pred16x16_horizontal_10_avx2; + } } } -- 2.51.0.windows.2

2 1

Re: [FFmpeg-devel] [PATCH 2/3] libavcodec: add NETINT Quadra HW decoders & encoders
by Desmond Liu 04 May '26

04 May '26

Hi Kieran, > Roughly one year ago I tried to buy one of these and I was asked to sign an > NDA. That was a "bug". You do not need to sign an NDA. If you have difficulty obtaining our hardware, contact me. Desmond

2 10

[PATCH] MAINTAINERS: add myself for libavfilter
by Nicolas George 25 Apr '26

25 Apr '26

I have been maintaining the framework and orphaned filters for ages but apparently it was never recorded in this file. Signed-off-by: Nicolas George <george(a)nsup.org> --- MAINTAINERS | 4 ++++ 1 file changed, 4 insertions(+) … but it was common knowledge. diff --git a/MAINTAINERS b/MAINTAINERS index fe4bcc69b0..1d74bcb512 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -316,6 +316,10 @@ libavfilter =========== Generic parts: + + Famework and orphaned filters Nicolas George + (except hardware acceleration) + graphdump.c Nicolas George motion_estimation.c Davinder Singh -- 2.47.2

3 4

[PATCH] configure: use ./src instead of src
by Nicolas George 25 Apr '26

25 Apr '26

Fix “do "src/doc/t2h.pm" failed, '.' is no longer in @INC; did you mean do "./src/doc/t2h.pm"?” when we have such a symlink. Signed-off-by: Nicolas George <george(a)nsup.org> --- configure | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/configure b/configure index 998b0292b8..46ae0297a3 100755 --- a/configure +++ b/configure @@ -4392,7 +4392,7 @@ mkdir -p ffbuild if test -f configure; then source_path=. elif test -f src/configure; then - source_path=src + source_path=./src else source_path=$(cd $(dirname "$0"); pwd) case "$source_path" in @@ -6220,7 +6220,7 @@ link_name=$(mktemp -u $TMPDIR/name_XXXXXXXX) mkdir "$link_dest" $ln_s "$link_dest" "$link_name" touch "$link_dest/test_file" -if [ "$source_path" != "." ] && [ "$source_path" != "src" ] && ([ ! -d src ] || [ -L src ]) && [ -e "$link_name/test_file" ]; then +if [ "$source_path" != "." ] && [ "$source_path" != "./src" ] && ([ ! -d src ] || [ -L src ]) && [ -e "$link_name/test_file" ]; then # create link to source path [ -e src ] && rm src $ln_s "$source_path" src -- 2.51.0

2 2

[PATCH] avcodec/libvpxenc: fix sRGB colorspace for non-RGB pixel formats
by Guangyu Sun 17 Mar '26

17 Mar '26

When encoding VP9 with a YUV pixel format (e.g. yuv420p) and AVCOL_SPC_RGB colorspace metadata, libvpxenc unconditionally set VPX_CS_SRGB. This produced a spec-violating bitstream: Profile 0 (4:2:0) with sRGB colorspace, which is only valid for Profile 1/3 (4:4:4). The resulting file is undecodable. RGB pixel formats (GBRP, GBRP10, GBRP12) already correctly set ctx->vpx_cs to VPX_CS_SRGB in set_pix_fmt() and bypass the colorspace switch in set_colorspace(). So reaching the AVCOL_SPC_RGB case means the pixel format is YUV with incorrect RGB metadata. Log a warning and fall back to VPX_CS_BT_709. To reproduce: # generate a bad source ffmpeg -f lavfi -i testsrc=s=64x64:d=1:r=1 -pix_fmt yuv420p -colorspace rgb bad.mp4 # transcode with default parameters ffmpeg -i bad.mp4 bad.webm # check decoding ffmpeg -i bad.webm -f null - # -> 0 frames decoded, error Signed-off-by: Guangyu Sun <gsun(a)roblox.com> --- libavcodec/libvpxenc.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/libavcodec/libvpxenc.c b/libavcodec/libvpxenc.c index 88c058c403..be731a014f 100644 --- a/libavcodec/libvpxenc.c +++ b/libavcodec/libvpxenc.c @@ -902,7 +902,17 @@ static void set_colorspace(AVCodecContext *avctx) vpx_cs = ctx->vpx_cs; } else { switch (avctx->colorspace) { - case AVCOL_SPC_RGB: vpx_cs = VPX_CS_SRGB; break; + case AVCOL_SPC_RGB: + // RGB pixel formats (GBRP, GBRP10, GBRP12) set ctx->vpx_cs + // to VPX_CS_SRGB in set_pix_fmt() and take the branch above, + // so reaching here means the pixel format is YUV with + // incorrect RGB colorspace metadata. + av_log(avctx, AV_LOG_WARNING, + "RGB colorspace is not compatible with pixel format %s, " + "using BT.709 instead.\n", + av_get_pix_fmt_name(avctx->pix_fmt)); + vpx_cs = VPX_CS_BT_709; + break; case AVCOL_SPC_BT709: vpx_cs = VPX_CS_BT_709; break; case AVCOL_SPC_UNSPECIFIED: vpx_cs = VPX_CS_UNKNOWN; break; case AVCOL_SPC_RESERVED: vpx_cs = VPX_CS_RESERVED; break; -- 2.52.0

3 9

[PATCH] avfilter/af_lv2: call lilv_instance_activate before lilv_instance_run
by Karl Mogensen 08 Mar '26

08 Mar '26

Why: the change is done to comply with lilv expectations of hosts. Added call lilv_instance_activate in the config_output function to abide by lilv documentation that states it must be called before lilv_instance_run: "This MUST be called before calling lilv_instance_run()" - documentation source (https://github.com/lv2/lilv/blob/main/include/lilv/lilv.h) Added call lilv_instance_deactivate in the uninit function to abide by lv2 documentation: "If a host calls activate(), it MUST call deactivate() at some point in the future" - documentation source (https://gitlab.com/lv2/lv2/-/blob/main/include/lv2/core/lv2.h) Added instance_activated integer to LV2Context struct to track if instance was activated and only do lilv_instance_deactivate if was activated to abide by lv2 documentation: "Hosts MUST NOT call deactivate() unless activate() was previously called." - documentation source (https://gitlab.com/lv2/lv2/-/blob/main/include/lv2/core/lv2.h) Regarding the patcheck warning (possibly constant :instance_activated): This is a false positive since the struct member is zero-initialized. Fixes: trac issue #11661 (https://trac.ffmpeg.org/ticket/11661) Reported-by: Dave Flater Signed-off-by: Karl Mogensen <karlmogensen0(a)gmail.com> --- libavfilter/af_lv2.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/libavfilter/af_lv2.c b/libavfilter/af_lv2.c index e7f3f9c588..df935601d4 100644 --- a/libavfilter/af_lv2.c +++ b/libavfilter/af_lv2.c @@ -74,6 +74,7 @@ typedef struct LV2Context { float *controls; LilvInstance *instance; + int instance_activated; LilvNode *atom_AtomPort; LilvNode *atom_Sequence; @@ -387,6 +388,9 @@ static int config_output(AVFilterLink *outlink) inlink->min_samples = inlink->max_samples = 4096; } + lilv_instance_activate(s->instance); + s->instance_activated = 1; + return 0; } @@ -574,6 +578,8 @@ static av_cold void uninit(AVFilterContext *ctx) lilv_node_free(s->lv2_OutputPort); lilv_node_free(s->lv2_InputPort); uri_table_destroy(&s->uri_table); + if (s->instance_activated) + lilv_instance_deactivate(s->instance); lilv_instance_free(s->instance); lilv_world_free(s->world); av_freep(&s->mins); -- 2.52.0

3 2