August 2013 - ffmpeg-devel - lists.ffmpeg.org

[PATCH] lavfi/hue: allow changing brightness
by Paul B Mahol 22 Aug '13

22 Aug '13

Signed-off-by: Paul B Mahol <onemda(a)gmail.com> --- doc/filters.texi | 4 +++ libavfilter/vf_hue.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++------ 2 files changed, 66 insertions(+), 7 deletions(-) diff --git a/doc/filters.texi b/doc/filters.texi index 19d9784..a9808a1 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -4715,6 +4715,10 @@ defaults to "1". @item H Specify the hue angle as a number of radians. It accepts an expression, and defaults to "0". + +@item b +Specify the brightness in the [-10,10] range. It accepts an expression and +defaults to "0". @end table @option{h} and @option{H} are mutually exclusive, and can't be diff --git a/libavfilter/vf_hue.c b/libavfilter/vf_hue.c index d066c40..e5017fb 100644 --- a/libavfilter/vf_hue.c +++ b/libavfilter/vf_hue.c @@ -68,11 +68,15 @@ typedef struct { float saturation; char *saturation_expr; AVExpr *saturation_pexpr; + float brightness; + char *brightness_expr; + AVExpr *brightness_pexpr; int hsub; int vsub; int32_t hue_sin; int32_t hue_cos; double var_values[VAR_NB]; + uint8_t lut_l[256]; uint8_t lut_u[256][256]; uint8_t lut_v[256][256]; } HueContext; @@ -86,6 +90,8 @@ static const AVOption hue_options[] = { { .str = "1" }, .flags = FLAGS }, { "H", "set the hue angle radians expression", OFFSET(hue_expr), AV_OPT_TYPE_STRING, { .str = NULL }, .flags = FLAGS }, + { "b", "set the brightness expression", OFFSET(brightness_expr), AV_OPT_TYPE_STRING, + { .str = "0" }, .flags = FLAGS }, { NULL } }; @@ -102,6 +108,16 @@ static inline void compute_sin_and_cos(HueContext *hue) hue->hue_cos = rint(cos(hue->hue) * (1 << 16) * hue->saturation); } +static inline void create_luma_lut(HueContext *h) +{ + const float b = h->brightness; + int i; + + for (i = 0; i < 256; i++) { + h->lut_l[i] = av_clip_uint8(i + b * 25.5); + } +} + static inline void create_chrominance_lut(HueContext *h, const int32_t c, const int32_t s) { @@ -181,14 +197,15 @@ static av_cold int init(AVFilterContext *ctx) if (ret < 0) \ return ret; \ } while (0) + SET_EXPR(brightness, "b"); SET_EXPR(saturation, "s"); SET_EXPR(hue_deg, "h"); SET_EXPR(hue, "H"); #undef SET_EXPR av_log(ctx, AV_LOG_VERBOSE, - "H_expr:%s h_deg_expr:%s s_expr:%s\n", - hue->hue_expr, hue->hue_deg_expr, hue->saturation_expr); + "H_expr:%s h_deg_expr:%s s_expr:%s b_expr:%s\n", + hue->hue_expr, hue->hue_deg_expr, hue->saturation_expr, hue->brightness_expr); compute_sin_and_cos(hue); return 0; @@ -198,6 +215,7 @@ static av_cold void uninit(AVFilterContext *ctx) { HueContext *hue = ctx->priv; + av_expr_free(hue->brightness_pexpr); av_expr_free(hue->hue_deg_pexpr); av_expr_free(hue->hue_pexpr); av_expr_free(hue->saturation_pexpr); @@ -235,6 +253,22 @@ static int config_props(AVFilterLink *inlink) return 0; } +static void apply_luma_lut(HueContext *s, + uint8_t *ldst, const int dst_linesize, + uint8_t *lsrc, const int src_linesize, + int w, int h) +{ + int i; + + while (h--) { + for (i = 0; i < w; i++) + ldst[i] = s->lut_l[lsrc[i]]; + + lsrc += src_linesize; + ldst += dst_linesize; + } +} + static void apply_lut(HueContext *s, uint8_t *udst, uint8_t *vdst, const int dst_linesize, uint8_t *usrc, uint8_t *vsrc, const int src_linesize, @@ -267,6 +301,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *inpic) AVFilterLink *outlink = inlink->dst->outputs[0]; AVFrame *outpic; const int32_t old_hue_sin = hue->hue_sin, old_hue_cos = hue->hue_cos; + const float old_brightness = hue->brightness; int direct = 0; if (av_frame_is_writable(inpic)) { @@ -296,6 +331,17 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *inpic) } } + if (hue->brightness_expr) { + hue->brightness = av_expr_eval(hue->brightness_pexpr, hue->var_values, NULL); + + if (hue->brightness < -10 || hue->brightness > 10) { + hue->brightness = av_clipf(hue->brightness, -10, 10); + av_log(inlink->dst, AV_LOG_WARNING, + "Brightness value not in range [%d,%d]: clipping value to %0.1f\n", + -10, 10, hue->brightness); + } + } + if (hue->hue_deg_expr) { hue->hue_deg = av_expr_eval(hue->hue_deg_pexpr, hue->var_values, NULL); hue->hue = hue->hue_deg * M_PI / 180; @@ -305,18 +351,22 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *inpic) } av_log(inlink->dst, AV_LOG_DEBUG, - "H:%0.1f*PI h:%0.1f s:%0.f t:%0.1f n:%d\n", - hue->hue/M_PI, hue->hue_deg, hue->saturation, + "H:%0.1f*PI h:%0.1f s:%0.f b:%0.f t:%0.1f n:%d\n", + hue->hue/M_PI, hue->hue_deg, hue->saturation, hue->brightness, hue->var_values[VAR_T], (int)hue->var_values[VAR_N]); compute_sin_and_cos(hue); if (old_hue_sin != hue->hue_sin || old_hue_cos != hue->hue_cos) create_chrominance_lut(hue, hue->hue_cos, hue->hue_sin); + if (old_brightness != hue->brightness && hue->brightness) + create_luma_lut(hue); + if (!direct) { - av_image_copy_plane(outpic->data[0], outpic->linesize[0], - inpic->data[0], inpic->linesize[0], - inlink->w, inlink->h); + if (!hue->brightness) + av_image_copy_plane(outpic->data[0], outpic->linesize[0], + inpic->data[0], inpic->linesize[0], + inlink->w, inlink->h); if (inpic->data[3]) av_image_copy_plane(outpic->data[3], outpic->linesize[3], inpic->data[3], inpic->linesize[3], @@ -327,6 +377,9 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *inpic) inpic->data[1], inpic->data[2], inpic->linesize[1], FF_CEIL_RSHIFT(inlink->w, hue->hsub), FF_CEIL_RSHIFT(inlink->h, hue->vsub)); + if (hue->brightness) + apply_luma_lut(hue, outpic->data[0], outpic->linesize[0], + inpic->data[0], inpic->linesize[0], inlink->w, inlink->h); if (!direct) av_frame_free(&inpic); @@ -355,6 +408,8 @@ static int process_command(AVFilterContext *ctx, const char *cmd, const char *ar av_freep(&hue->hue_deg_expr); } else if (!strcmp(cmd, "s")) { SET_EXPR(saturation, "s"); + } else if (!strcmp(cmd, "b")) { + SET_EXPR(brightness, "b"); } else return AVERROR(ENOSYS); -- 1.7.11.2

3 2

HLS improvements
by Andrew Sinclair 22 Aug '13

22 Aug '13

Hi Guys, Is anyone interested in working with me to make some improvements to ffmpeg on how it handles multi bitrate / multiple variant HLS as well as adding support for features such as alternate audio and subtitles? Issue is at the moment it doesn't optimally parse a master playlist file and doesn't include sub title or alternate audio support. I know what needs to be done and have worked through this with a dev before in a closed source environment but would like to see this in ffmpeg but don't have time to do it myself at present. Would be happy to pay for someones time too. Let me know if anyone is interested. Best regards, Andrew

2 2

Piece of code to parallelize
by LuizGuilherme CM 22 Aug '13

22 Aug '13

Hi guys, I'm doing a term paper about parallel programming and my teacher asked me to parallelize a piece of code at production. So, I found FFmpeg, which work with a big array and has good chance to be parallelized, but I'm having a hard time looking into the code and find a good and the right piece to work out. If someone has some document which I didn't find before and could help me with this I appreciate. Sorry if this isn't the right place to send this. att, -- Luiz Guilherme Castilho Martins cel: 9920 5877 msn: luizgui(a)gmail.com

6 9

[PATCH] fate: Add EXIF test.
by Thilo Borgmann 21 Aug '13

21 Aug '13

Hi, this adds tests for EXIF in TIFF, JPEG and JPEG embedded in MP3. The three new samples are here: http://tinyurl.com/mwgrthp http://tinyurl.com/k3hd8wb http://tinyurl.com/l83y2ar -Thilo The following changes since commit 76d8d2388120910b46225299b3e10fb4dd18287f: Revert "doc/RELEASE_NOTES: add a note about AVISynth" (2013-08-19 12:39:28 +0200) are available in the git repository at: git@github.com:thiloborgmann/FFmpeg.git for you to fetch changes up to ea862e149e5429de1305c04491e64f4ef71fc18c: fate: Add EXIF test. (2013-08-19 14:45:19 +0200) ---------------------------------------------------------------- Thilo Borgmann (1): fate: Add EXIF test. tests/Makefile | 1 + tests/fate-run.sh | 4 + tests/fate/exif.mak | 14 ++ tests/ref/fate/exif-image-embedded | 360 +++++++++++++++++++++++++++++++++++++ tests/ref/fate/exif-image-jpg | 66 +++++++ tests/ref/fate/exif-image-tiff | 25 +++ 6 files changed, 470 insertions(+) create mode 100644 tests/fate/exif.mak create mode 100644 tests/ref/fate/exif-image-embedded create mode 100644 tests/ref/fate/exif-image-jpg create mode 100644 tests/ref/fate/exif-image-tiff diff --git a/tests/Makefile b/tests/Makefile index 914448e..61423f4 100644 --- a/tests/Makefile +++ b/tests/Makefile @@ -93,6 +93,7 @@ include $(SRC_PATH)/tests/fate/demux.mak include $(SRC_PATH)/tests/fate/dfa.mak include $(SRC_PATH)/tests/fate/dpcm.mak include $(SRC_PATH)/tests/fate/ea.mak +include $(SRC_PATH)/tests/fate/exif.mak include $(SRC_PATH)/tests/fate/ffmpeg.mak include $(SRC_PATH)/tests/fate/ffprobe.mak include $(SRC_PATH)/tests/fate/filter-audio.mak diff --git a/tests/fate-run.sh b/tests/fate-run.sh index 9f5690d..52ddc22 100755 --- a/tests/fate-run.sh +++ b/tests/fate-run.sh @@ -75,6 +75,10 @@ probefmt(){ run ffprobe -show_entries format=format_name -print_format default=nw=1:nk=1 -v 0 "$@" } +probeframes(){ + run ffprobe -show_frames -v 0 "$@" +} + ffmpeg(){ dec_opts="-threads $threads -thread_type $thread_type" ffmpeg_args="-nostats -cpuflags $cpuflags" diff --git a/tests/fate/exif.mak b/tests/fate/exif.mak new file mode 100644 index 0000000..31b29a0 --- /dev/null +++ b/tests/fate/exif.mak @@ -0,0 +1,14 @@ +# test exif metadata in TIFF images +FATE_SAMPLES_EXIF-$(call DEMDEC, IMAGE2, TIFF) += fate-exif-image-tiff +fate-exif-image-tiff: CMD = probeframes $(TARGET_SAMPLES)/exif/image_small.tiff + +# test exif metadata in JPG images +FATE_SAMPLES_EXIF-$(call DEMDEC, IMAGE2, MJPEG) += fate-exif-image-jpg +fate-exif-image-jpg: CMD = probeframes $(TARGET_SAMPLES)/exif/image_small.jpg + +# test exif metadata in MP3 with embedded JPEG images +FATE_SAMPLES_EXIF-$(call DEMDEC, IMAGE2, MJPEG) += fate-exif-image-embedded +fate-exif-image-embedded: CMD = probeframes $(TARGET_SAMPLES)/exif/embedded_small.mp3 + +# add all -yes targets to the tested targets +FATE_SAMPLES_FFMPEG += $(FATE_SAMPLES_EXIF-yes) diff --git a/tests/ref/fate/exif-image-embedded b/tests/ref/fate/exif-image-embedded new file mode 100644 index 0000000..4109c15 --- /dev/null +++ b/tests/ref/fate/exif-image-embedded @@ -0,0 +1,360 @@ +[FRAME] +media_type=video +key_frame=1 +pkt_pts=N/A +pkt_pts_time=N/A +pkt_dts=N/A +pkt_dts_time=N/A +pkt_duration=N/A +pkt_duration_time=N/A +pkt_pos=N/A +pkt_size=15760 +width=263 +height=263 +pix_fmt=yuvj420p +sample_aspect_ratio=1:1 +pict_type=I +coded_picture_number=0 +display_picture_number=0 +interlaced_frame=0 +top_field_first=0 +repeat_pict=0 +TAG:UserComment=AppleMark + +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=353600 +pkt_pts_time=0.025057 +pkt_dts=353600 +pkt_dts_time=0.025057 +pkt_duration=15040 +pkt_duration_time=0.001066 +pkt_pos=16292 +pkt_size=417 +sample_fmt=s16p +nb_samples=47 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=368640 +pkt_pts_time=0.026122 +pkt_dts=368640 +pkt_dts_time=0.026122 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=16709 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=737280 +pkt_pts_time=0.052245 +pkt_dts=737280 +pkt_dts_time=0.052245 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=17127 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=1105920 +pkt_pts_time=0.078367 +pkt_dts=1105920 +pkt_dts_time=0.078367 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=17545 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=1474560 +pkt_pts_time=0.104490 +pkt_dts=1474560 +pkt_dts_time=0.104490 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=17963 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=1843200 +pkt_pts_time=0.130612 +pkt_dts=1843200 +pkt_dts_time=0.130612 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=18381 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=2211840 +pkt_pts_time=0.156735 +pkt_dts=2211840 +pkt_dts_time=0.156735 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=18799 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=2580480 +pkt_pts_time=0.182857 +pkt_dts=2580480 +pkt_dts_time=0.182857 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=19217 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=2949120 +pkt_pts_time=0.208980 +pkt_dts=2949120 +pkt_dts_time=0.208980 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=19635 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=3317760 +pkt_pts_time=0.235102 +pkt_dts=3317760 +pkt_dts_time=0.235102 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=20053 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=3686400 +pkt_pts_time=0.261224 +pkt_dts=3686400 +pkt_dts_time=0.261224 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=20471 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=4055040 +pkt_pts_time=0.287347 +pkt_dts=4055040 +pkt_dts_time=0.287347 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=20889 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=4423680 +pkt_pts_time=0.313469 +pkt_dts=4423680 +pkt_dts_time=0.313469 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=21307 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=4792320 +pkt_pts_time=0.339592 +pkt_dts=4792320 +pkt_dts_time=0.339592 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=21725 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=5160960 +pkt_pts_time=0.365714 +pkt_dts=5160960 +pkt_dts_time=0.365714 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=22143 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=5529600 +pkt_pts_time=0.391837 +pkt_dts=5529600 +pkt_dts_time=0.391837 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=22561 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=5898240 +pkt_pts_time=0.417959 +pkt_dts=5898240 +pkt_dts_time=0.417959 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=22979 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=6266880 +pkt_pts_time=0.444082 +pkt_dts=6266880 +pkt_dts_time=0.444082 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=23397 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=6635520 +pkt_pts_time=0.470204 +pkt_dts=6635520 +pkt_dts_time=0.470204 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=23815 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=7004160 +pkt_pts_time=0.496327 +pkt_dts=7004160 +pkt_dts_time=0.496327 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=24233 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] +[FRAME] +media_type=audio +key_frame=1 +pkt_pts=7372800 +pkt_pts_time=0.522449 +pkt_dts=7372800 +pkt_dts_time=0.522449 +pkt_duration=368640 +pkt_duration_time=0.026122 +pkt_pos=24651 +pkt_size=418 +sample_fmt=s16p +nb_samples=1152 +channels=2 +channel_layout=stereo +[/FRAME] diff --git a/tests/ref/fate/exif-image-jpg b/tests/ref/fate/exif-image-jpg new file mode 100644 index 0000000..5695bb6 --- /dev/null +++ b/tests/ref/fate/exif-image-jpg @@ -0,0 +1,66 @@ +[FRAME] +media_type=video +key_frame=1 +pkt_pts=0 +pkt_pts_time=0.000000 +pkt_dts=0 +pkt_dts_time=0.000000 +pkt_duration=1 +pkt_duration_time=0.040000 +pkt_pos=N/A +pkt_size=46095 +width=400 +height=225 +pix_fmt=yuvj422p +sample_aspect_ratio=1:1 +pict_type=I +coded_picture_number=0 +display_picture_number=0 +interlaced_frame=0 +top_field_first=0 +repeat_pict=0 +TAG:ImageDescription= +TAG:Make=Canon +TAG:Model=Canon PowerShot SX200 IS +TAG:Orientation=1 +TAG:XResolution=180:1 +TAG:YResolution=180:1 +TAG:ResolutionUnit=2 +TAG:DateTime=2013:07:18 13:12:03 +TAG:YCbCrPositioning=2 +TAG:ExposureTime=1:1250 +TAG:FNumber=40:10 +TAG:ISOSpeedRatings=160 +TAG:ExifVersion=0221 +TAG:DateTimeOriginal=2013:07:18 13:12:03 +TAG:DateTimeDigitized=2013:07:18 13:12:03 +TAG:ComponentsConfiguration= +TAG:CompressedBitsPerPixel=3:1 +TAG:ShutterSpeedValue=329:32 +TAG:ApertureValue=128:32 +TAG:ExposureBiasValue=0:3 +TAG:MaxApertureValue=113:32 +TAG:MeteringMode=5 +TAG:Flash=16 +TAG:FocalLength=5000:1000 +TAG:MakerNote= +TAG:UserComment= +TAG:FlashpixVersion=0100 +TAG:ColorSpace=1 +TAG:PixelXDimension=4000 +TAG:PixelYDimension=2248 +TAG:GPSLatitudeRef=R98 +TAG:GPSLatitude=0100 +TAG:0x1001=4000 +TAG:0x1002=2248 +TAG:FocalPlaneXResolution=4000000:244 +TAG:FocalPlaneYResolution=2248000:183 +TAG:FocalPlaneResolutionUnit=2 +TAG:SensingMethod=2 +TAG:FileSource= +TAG:CustomRendered=0 +TAG:ExposureMode=0 +TAG:WhiteBalance=0 +TAG:DigitalZoomRatio=4000:4000 +TAG:SceneCaptureType=0 +[/FRAME] diff --git a/tests/ref/fate/exif-image-tiff b/tests/ref/fate/exif-image-tiff new file mode 100644 index 0000000..ddd85e5 --- /dev/null +++ b/tests/ref/fate/exif-image-tiff @@ -0,0 +1,25 @@ +[FRAME] +media_type=video +key_frame=1 +pkt_pts=0 +pkt_pts_time=0.000000 +pkt_dts=0 +pkt_dts_time=0.000000 +pkt_duration=1 +pkt_duration_time=0.040000 +pkt_pos=N/A +pkt_size=67604 +width=200 +height=112 +pix_fmt=rgb24 +sample_aspect_ratio=N/A +pict_type=? +coded_picture_number=0 +display_picture_number=0 +interlaced_frame=0 +top_field_first=0 +repeat_pict=0 +TAG:document_name=image_small.tiff +TAG:page_number=0 / 1 +TAG:software=ImageMagick 6.5.8-0 2010-02-09 Q16 http://www.imagemagick.org +[/FRAME]

3 11

[PATCH] lavfi/hue: use lookup tables
by Paul B Mahol 21 Aug '13

21 Aug '13

Signed-off-by: Paul B Mahol <onemda(a)gmail.com> --- libavfilter/vf_hue.c | 76 +++++++++++++++++++++++++++++++--------------------- 1 file changed, 45 insertions(+), 31 deletions(-) diff --git a/libavfilter/vf_hue.c b/libavfilter/vf_hue.c index cc8ac78..a687991 100644 --- a/libavfilter/vf_hue.c +++ b/libavfilter/vf_hue.c @@ -73,6 +73,8 @@ typedef struct { int32_t hue_sin; int32_t hue_cos; double var_values[VAR_NB]; + uint8_t lut_u[256]; + uint8_t lut_v[256]; } HueContext; #define OFFSET(x) offsetof(HueContext, x) @@ -94,12 +96,41 @@ static inline void compute_sin_and_cos(HueContext *hue) /* * Scale the value to the norm of the resulting (U,V) vector, that is * the saturation. - * This will be useful in the process_chrominance function. + * This will be useful in the apply_lut function. */ hue->hue_sin = rint(sin(hue->hue) * (1 << 16) * hue->saturation); hue->hue_cos = rint(cos(hue->hue) * (1 << 16) * hue->saturation); } +static inline void create_chrominance_lut(HueContext *s, const int32_t cos, + const int32_t sin) +{ + int32_t i, u, v, new_u, new_v; + + /* + * If we consider U and V as the components of a 2D vector then its angle + * is the hue and the norm is the saturation + */ + for (i = 0; i < 256; i++) { + /* Normalize the components from range [16;140] to [-112;112] */ + u = i - 128; + v = i - 128; + /* + * Apply the rotation of the vector : (cos * u) - (sin * v) + * (sin * u) + (cos * v) + * De-normalize the components (without forgetting to scale 128 + * by << 16) + * Finally scale back the result by >> 16 + */ + new_u = ((cos * u) - (sin * v) + (1 << 15) + (128 << 16)) >> 16; + new_v = ((sin * u) + (cos * v) + (1 << 15) + (128 << 16)) >> 16; + + /* Prevent a potential overflow */ + s->lut_u[i] = av_clip_uint8_c(new_u); + s->lut_v[i] = av_clip_uint8_c(new_v); + } +} + static int set_expr(AVExpr **pexpr_ptr, char **expr_ptr, const char *expr, const char *option, void *log_ctx) { @@ -202,36 +233,17 @@ static int config_props(AVFilterLink *inlink) return 0; } -static void process_chrominance(uint8_t *udst, uint8_t *vdst, const int dst_linesize, - uint8_t *usrc, uint8_t *vsrc, const int src_linesize, - int w, int h, - const int32_t c, const int32_t s) +static void apply_lut(HueContext *s, + uint8_t *udst, uint8_t *vdst, const int dst_linesize, + uint8_t *usrc, uint8_t *vsrc, const int src_linesize, + int w, int h) { - int32_t u, v, new_u, new_v; int i; - /* - * If we consider U and V as the components of a 2D vector then its angle - * is the hue and the norm is the saturation - */ while (h--) { for (i = 0; i < w; i++) { - /* Normalize the components from range [16;140] to [-112;112] */ - u = usrc[i] - 128; - v = vsrc[i] - 128; - /* - * Apply the rotation of the vector : (c * u) - (s * v) - * (s * u) + (c * v) - * De-normalize the components (without forgetting to scale 128 - * by << 16) - * Finally scale back the result by >> 16 - */ - new_u = ((c * u) - (s * v) + (1 << 15) + (128 << 16)) >> 16; - new_v = ((s * u) + (c * v) + (1 << 15) + (128 << 16)) >> 16; - - /* Prevent a potential overflow */ - udst[i] = av_clip_uint8_c(new_u); - vdst[i] = av_clip_uint8_c(new_v); + udst[i] = s->lut_u[usrc[i]]; + vdst[i] = s->lut_v[vsrc[i]]; } usrc += src_linesize; @@ -249,6 +261,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *inpic) HueContext *hue = inlink->dst->priv; AVFilterLink *outlink = inlink->dst->outputs[0]; AVFrame *outpic; + const int32_t old_hue_sin = hue->hue_sin, old_hue_cos = hue->hue_cos; int direct = 0; if (av_frame_is_writable(inpic)) { @@ -292,6 +305,8 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *inpic) hue->var_values[VAR_T], (int)hue->var_values[VAR_N]); compute_sin_and_cos(hue); + if (old_hue_sin != hue->hue_sin || old_hue_cos != hue->hue_cos) + create_chrominance_lut(hue, hue->hue_cos, hue->hue_sin); if (!direct) { av_image_copy_plane(outpic->data[0], outpic->linesize[0], @@ -303,11 +318,10 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *inpic) inlink->w, inlink->h); } - process_chrominance(outpic->data[1], outpic->data[2], outpic->linesize[1], - inpic->data[1], inpic->data[2], inpic->linesize[1], - FF_CEIL_RSHIFT(inlink->w, hue->hsub), - FF_CEIL_RSHIFT(inlink->h, hue->vsub), - hue->hue_cos, hue->hue_sin); + apply_lut(hue, outpic->data[1], outpic->data[2], outpic->linesize[1], + inpic->data[1], inpic->data[2], inpic->linesize[1], + FF_CEIL_RSHIFT(inlink->w, hue->hsub), + FF_CEIL_RSHIFT(inlink->h, hue->vsub)); if (!direct) av_frame_free(&inpic); -- 1.7.11.2

3 10

[RFC] lavf/tee per-output stream selection
by Stefano Sabatini 21 Aug '13

21 Aug '13

Hi, I need to be be able to send different streams to different tee outputs- In other words I need to select which streams should be routed to each tee output, while right now it is possible to specify the streams just once, and they will be the same for every output. This allows to create outputs with different streams, while avoiding duplicated encoding and the need to run different processes communicating through pipes. A possible design would involve a syntax like this: ffmpeg -i INPUT -c:v libx264 -c:a libvo-aacenc -tee_select "v|a" -f tee "out-v.ts|out-a.ts|out-av.ts" This would select the streams for each output, the non-matching streams won't be added to the corresponding output, and the corresponding packets will be ignored. This requires some somehow complex changes to the tee logic, but won't affect the rest of the library. At some point it would be nice to make possible to select streams on the base of their metadata (e.g. to select only streams in a certain language) using stream selection, but I can devise a simple syntax, not involving demonic escaping issues. Comments are welcome, especially from the maintainer of the tee virtual muxer. -- FFmpeg = Fanciful and Fiendish Minimalistic Patchable Exciting Game

4 21

[PATCH] pngdec: frame multithreading support
by Paul B Mahol 21 Aug '13

21 Aug '13

Signed-off-by: Paul B Mahol <onemda(a)gmail.com> --- libavcodec/pngdec.c | 71 +++++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 53 insertions(+), 18 deletions(-) diff --git a/libavcodec/pngdec.c b/libavcodec/pngdec.c index 1358ac5..dcdbb2f 100644 --- a/libavcodec/pngdec.c +++ b/libavcodec/pngdec.c @@ -28,6 +28,7 @@ #include "internal.h" #include "png.h" #include "pngdsp.h" +#include "thread.h" /* TODO: * - add 16 bit depth support @@ -40,7 +41,8 @@ typedef struct PNGDecContext { AVCodecContext *avctx; GetByteContext gb; - AVFrame *prev; + ThreadFrame last_picture; + ThreadFrame picture; int state; int width, height; @@ -505,13 +507,18 @@ static int decode_frame(AVCodecContext *avctx, PNGDecContext * const s = avctx->priv_data; const uint8_t *buf = avpkt->data; int buf_size = avpkt->size; - AVFrame *p = data; + AVFrame *p; AVDictionary *metadata = NULL; uint8_t *crow_buf_base = NULL; uint32_t tag, length; int64_t sig; int ret; + if (s->last_picture.f) + ff_thread_release_buffer(avctx, &s->last_picture); + FFSWAP(ThreadFrame, s->picture, s->last_picture); + p = s->picture.f; + bytestream2_init(&s->gb, buf, buf_size); /* check signature */ @@ -635,8 +642,10 @@ static int decode_frame(AVCodecContext *avctx, goto fail; } - if (ff_get_buffer(avctx, p, AV_GET_BUFFER_FLAG_REF) < 0) + if (ff_thread_get_buffer(avctx, &s->picture, AV_GET_BUFFER_FLAG_REF) < 0) goto fail; + ff_thread_finish_setup(avctx); + p->pict_type = AV_PICTURE_TYPE_I; p->key_frame = 1; p->interlaced_frame = !!s->interlace_type; @@ -820,16 +829,17 @@ static int decode_frame(AVCodecContext *avctx, } /* handle p-frames only if a predecessor frame is available */ - if (s->prev->data[0]) { + if (s->last_picture.f->data[0]) { if ( !(avpkt->flags & AV_PKT_FLAG_KEY) && avctx->codec_tag != AV_RL32("MPNG") - && s->prev->width == p->width - && s->prev->height== p->height - && s->prev->format== p->format + && s->last_picture.f->width == p->width + && s->last_picture.f->height== p->height + && s->last_picture.f->format== p->format ) { int i, j; uint8_t *pd = p->data[0]; - uint8_t *pd_last = s->prev->data[0]; + uint8_t *pd_last = s->last_picture.f->data[0]; + ff_thread_await_progress(&s->last_picture, INT_MAX, 0); for (j = 0; j < s->height; j++) { for (i = 0; i < s->width * s->bpp; i++) { pd[i] += pd_last[i]; @@ -839,13 +849,13 @@ static int decode_frame(AVCodecContext *avctx, } } } + ff_thread_report_progress(&s->picture, INT_MAX, 0); av_frame_set_metadata(p, metadata); metadata = NULL; - av_frame_unref(s->prev); - if ((ret = av_frame_ref(s->prev, p)) < 0) - goto fail; + if ((ret = av_frame_ref(data, s->picture.f)) < 0) + return ret; *got_frame = 1; @@ -863,17 +873,35 @@ static int decode_frame(AVCodecContext *avctx, goto the_end; } +static int update_thread_context(AVCodecContext *dst, const AVCodecContext *src) +{ + PNGDecContext *psrc = src->priv_data; + PNGDecContext *pdst = dst->priv_data; + + if (dst == src) + return 0; + + ff_thread_release_buffer(dst, &pdst->picture); + if (psrc->picture.f->data[0]) + return ff_thread_ref_frame(&pdst->picture, &psrc->picture); + + return 0; +} + static av_cold int png_dec_init(AVCodecContext *avctx) { PNGDecContext *s = avctx->priv_data; - s->prev = av_frame_alloc(); - if (!s->prev) + s->avctx = avctx; + s->last_picture.f = av_frame_alloc(); + s->picture.f = av_frame_alloc(); + if (!s->last_picture.f || !s->picture.f) return AVERROR(ENOMEM); - ff_pngdsp_init(&s->dsp); - - s->avctx = avctx; + if (!avctx->internal->is_copy) { + avctx->internal->allocate_progress = 1; + ff_pngdsp_init(&s->dsp); + } return 0; } @@ -882,7 +910,12 @@ static av_cold int png_dec_end(AVCodecContext *avctx) { PNGDecContext *s = avctx->priv_data; - av_frame_free(&s->prev); + if (s->last_picture.f) + ff_thread_release_buffer(avctx, &s->last_picture); + av_frame_free(&s->last_picture.f); + if (s->picture.f) + ff_thread_release_buffer(avctx, &s->picture); + av_frame_free(&s->picture.f); return 0; } @@ -895,6 +928,8 @@ AVCodec ff_png_decoder = { .init = png_dec_init, .close = png_dec_end, .decode = decode_frame, - .capabilities = CODEC_CAP_DR1 /*| CODEC_CAP_DRAW_HORIZ_BAND*/, + .init_thread_copy = ONLY_IF_THREADS_ENABLED(png_dec_init), + .update_thread_context = ONLY_IF_THREADS_ENABLED(update_thread_context), + .capabilities = CODEC_CAP_DR1 | CODEC_CAP_FRAME_THREADS /*| CODEC_CAP_DRAW_HORIZ_BAND*/, .long_name = NULL_IF_CONFIG_SMALL("PNG (Portable Network Graphics) image"), }; -- 1.7.11.2

2 2

[PATCH] Add paragraph to website about posting attitude on project mailing lists
by Robert Krüger 20 Aug '13

20 Aug '13

Hi, since it regularly happens that it feels necessary to remind complaining people about the nature of an open source project without being impolite or rude, I tried to formulate this in a paragraph that can be directly linked to in ML responses. I don't know if that is wanted or if you agree with the tone of it but I thought I'd try. Robert

7 12

[PATCH] lavfi/gradfun: add GBRP format
by Paul B Mahol 20 Aug '13

20 Aug '13

Signed-off-by: Paul B Mahol <onemda(a)gmail.com> --- libavfilter/vf_gradfun.c | 1 + 1 file changed, 1 insertion(+) diff --git a/libavfilter/vf_gradfun.c b/libavfilter/vf_gradfun.c index b2e7007..c3661e1 100644 --- a/libavfilter/vf_gradfun.c +++ b/libavfilter/vf_gradfun.c @@ -152,6 +152,7 @@ static int query_formats(AVFilterContext *ctx) AV_PIX_FMT_GRAY8, AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUV411P, AV_PIX_FMT_YUV440P, + AV_PIX_FMT_GBRP, AV_PIX_FMT_NONE }; -- 1.7.11.2

2 1

Any way to delay input frame in avcodec_encode_video2
by Smirnov eugene 20 Aug '13

20 Aug '13

Currently AVFrame passed to encoder has const modifier, this logically prevents me from using any manipulation with buffer references, like av_frame_ref, and av_frame_unref. However writing an asynchronous encoder I’m thinking it doesn’t make sense to copy entire AVFrame data planes for every frame from performance point of view. Looking at libx264 integration there is nothing to do with av_frame_ref, so internally x264 does copy of all uncompressed in help of x264_frame_copy_picture(). Why const modifier is so necessary? Will there be any problems if I start using c-style cast in form of (AVFrame*)frame and then play with av_frame_ref / unref functions

2 1

Invalid and inefficient vfw-avi packed B frames detected
by Sheyong Shim 20 Aug '13

20 Aug '13

I got message(Invalid and inefficient vfw-avi packed B frames detected) while video was converted. After converting the video, I checked pixel values in video. Values are different from original video. Why values change after converting? Please tell me why. Please send me an answer*.* ffmpeg version 1.1.1 Copyright (c) 2000-2013 the FFmpeg developers built on Jan 20 2013 23:05:28 with gcc 4.7.2 (GCC) configuration: --enable-gpl --enable-version3 --disable-w32threads --enable-avisynth --enable-bzlib --enable-fontconfig --enable-frei0r --enable-gnutls --enable-libass --enable-libbluray --enable-libcaca --enable-libfreetype --enable-libgsm --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-librtmp --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libtwolame --enable-libvo-aacenc --enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libxavs --enable-libxvid --enable-zlib libavutil 52. 13.100 / 52. 13.100 libavcodec 54. 86.100 / 54. 86.100 libavformat 54. 59.106 / 54. 59.106 libavdevice 54. 3.102 / 54. 3.102 libavfilter 3. 32.100 / 3. 32.100 libswscale 2. 1.103 / 2. 1.103 libswresample 0. 17.102 / 0. 17.102 libpostproc 52. 2.100 / 52. 2.100 [mpeg4 @ 02e218c0] Invalid and inefficient vfw-avi packed B frames detected Input #0, avi, from 'D:\video_output\avi\1.avi': Metadata: encoder : Lavf54.59.106 Duration: 00:04:25.52, start: 0.000000, bitrate: 2744 kb/s Stream #0:0: Video: mpeg4 (Advanced Simple Profile) (DX50 / 0x30355844), yuv420p, 720x544 [SAR 1:1 DAR 45:34], 25 tbr, 25 tbn, 25 tbc Stream #0:1: Audio: ac3 ([0] [0][0] / 0x2000), 48000 Hz, stereo, fltp, 192 kb/s Output #0, matroska, to 'D:\video_output\avi\1_2.mkv': Metadata: encoder : Lavf54.59.106 Stream #0:0: Video: ffv1 (FFV1 / 0x31564646), yuv420p, 720x544 [SAR 1:1 DAR 45:34], q=2-31, 200 kb/s, 1k tbn, 25 tbc Stream #0:1: Audio: pcm_s16le, 48000 Hz, stereo, s16, 1536 kb/s Stream mapping: Stream #0:0 -> #0:0 (mpeg4 -> ffv1) Stream #0:1 -> #0:1 (ac3 -> pcm_s16le) Press [q] to stop, [?] for help [mpeg4 @ 02e27920] Invalid and inefficient vfw-avi packed B frames detected frame= 36 fps=0.0 q=0.0 size= 1916kB time=00:00:01.56 bitrate=10008.5kbits/s ... frame= 6565 fps= 46 q=0.0 size= 862611kB time=00:04:22.72 bitrate=26897.5kbits/s frame= 6594 fps= 46 q=0.0 size= 864799kB time=00:04:23.87 bitrate=26848.0kbits/s frame CRC mismatch frame= 6629 fps= 46 q=0.0 size= 866407kB time=00:04:25.20 bitrate=26763.2kbits/s frame= 6638 fps= 46 q=0.0 Lsize= 866521kB time=00:04:25.56 bitrate=26730.5kbits/s video:816725kB audio:49668kB subtitle:0 global headers:0kB muxing overhead 0.014713%

2 1

[PATCH] Matroska demuxer adds WebVTT support
by Matthew Heaney 19 Aug '13

19 Aug '13

WebM files now support inband text tracks, as described in the following specification: http://wiki.webmproject.org/webm-metadata/temporal-metadata/webvtt-in-webm The Matroska demuxer now detects the presence of WebVTT tracks, synthesizing WebVTT packets (having codec id AV_CODEC_ID_WEBVTT) and pushing them downstream in the normal way. --- libavformat/matroska.c | 5 ++ libavformat/matroskadec.c | 130 +++++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 134 insertions(+), 1 deletion(-) diff --git a/libavformat/matroska.c b/libavformat/matroska.c index ee57c18..953c572 100644 --- a/libavformat/matroska.c +++ b/libavformat/matroska.c @@ -57,6 +57,11 @@ const CodecTags ff_mkv_codec_tags[]={ {"A_VORBIS" , AV_CODEC_ID_VORBIS}, {"A_WAVPACK4" , AV_CODEC_ID_WAVPACK}, + {"D_WEBVTT/SUBTITLES" , AV_CODEC_ID_WEBVTT}, + {"D_WEBVTT/CAPTIONS" , AV_CODEC_ID_WEBVTT}, + {"D_WEBVTT/DESCRIPTIONS", AV_CODEC_ID_WEBVTT}, + {"D_WEBVTT/METADATA" , AV_CODEC_ID_WEBVTT}, + {"S_TEXT/UTF8" , AV_CODEC_ID_SUBRIP}, {"S_TEXT/UTF8" , AV_CODEC_ID_TEXT}, {"S_TEXT/UTF8" , AV_CODEC_ID_SRT}, diff --git a/libavformat/matroskadec.c b/libavformat/matroskadec.c index a1aa0eb..d08194d 100644 --- a/libavformat/matroskadec.c +++ b/libavformat/matroskadec.c @@ -1566,7 +1566,8 @@ static int matroska_read_header(AVFormatContext *s) /* Apply some sanity checks. */ if (track->type != MATROSKA_TRACK_TYPE_VIDEO && track->type != MATROSKA_TRACK_TYPE_AUDIO && - track->type != MATROSKA_TRACK_TYPE_SUBTITLE) { + track->type != MATROSKA_TRACK_TYPE_SUBTITLE && + track->type != MATROSKA_TRACK_TYPE_METADATA) { av_log(matroska->ctx, AV_LOG_INFO, "Unknown or unsupported track type %"PRIu64"\n", track->type); @@ -1860,6 +1861,16 @@ static int matroska_read_header(AVFormatContext *s) st->codec->bits_per_coded_sample = track->audio.bitdepth; if (st->codec->codec_id != AV_CODEC_ID_AAC) st->need_parsing = AVSTREAM_PARSE_HEADERS; + } else if (codec_id == AV_CODEC_ID_WEBVTT) { + st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE; + + if (!strcmp(track->codec_id, "D_WEBVTT/CAPTIONS")) { + st->disposition |= AV_DISPOSITION_CAPTIONS; + } else if (!strcmp(track->codec_id, "D_WEBVTT/DESCRIPTIONS")) { + st->disposition |= AV_DISPOSITION_DESCRIPTIONS; + } else if (!strcmp(track->codec_id, "D_WEBVTT/METADATA")) { + st->disposition |= AV_DISPOSITION_METADATA; + } } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) { st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE; #if FF_API_ASS_SSA @@ -2226,6 +2237,115 @@ fail: return ret; } +static int matroska_parse_webvtt(MatroskaDemuxContext *matroska, + MatroskaTrack *track, + AVStream *st, + uint8_t *data, int data_len, + uint64_t timecode, + uint64_t duration, + int64_t pos) +{ + AVPacket *pkt; + uint8_t *id, *settings, *text, *buf; + int id_len, settings_len, text_len; + + uint8_t *p = data; + uint8_t *q = data + data_len; + + if (p >= q) + return AVERROR_INVALIDDATA; + + id = p; + id_len = strcspn(p, "\r\n"); + + p += id_len; + if (p >= q) + return AVERROR_INVALIDDATA; + + if (*p == '\r') + p++; + if (p >= q || *p != '\n') + return AVERROR_INVALIDDATA; + p++; + + if (p >= q) + return AVERROR_INVALIDDATA; + + settings = p; + settings_len = strcspn(p, "\r\n"); + + p += settings_len; + if (p >= q) + return AVERROR_INVALIDDATA; + + if (*p == '\r') + p++; + if (p >= q || *p != '\n') + return AVERROR_INVALIDDATA; + p++; + + if (p >= q) + return AVERROR_INVALIDDATA; + + text = p; + text_len = q - p; + + while (text_len > 0) { + const int len = text_len - 1; + const uint8_t c = p[len]; + if (c != '\r' && c != '\n') + break; + text_len = len; + } + + pkt = av_mallocz(sizeof(AVPacket)); + if (av_new_packet(pkt, text_len) < 0) { + av_free(pkt); + return AVERROR(ENOMEM); + } + + memcpy(pkt->data, text, text_len); + + if (id_len > 0) { + buf = av_packet_new_side_data(pkt, + AV_PKT_DATA_WEBVTT_IDENTIFIER, + id_len); + if (buf == NULL) { + av_free(pkt); + return AVERROR(ENOMEM); + } + memcpy(buf, id, id_len); + } + + if (settings_len > 0) { + buf = av_packet_new_side_data(pkt, + AV_PKT_DATA_WEBVTT_SETTINGS, + settings_len); + if (buf == NULL) { + av_free(pkt); + return AVERROR(ENOMEM); + } + memcpy(buf, settings, settings_len); + } + + // Do we need this for subtitles? + // pkt->flags = AV_PKT_FLAG_KEY; + + pkt->stream_index = st->index; + pkt->pts = timecode; + + // Do we need this for subtitles? + // pkt->dts = timecode; + + pkt->duration = duration; + pkt->pos = pos; + + dynarray_add(&matroska->packets, &matroska->num_packets, pkt); + matroska->prev_pkt = pkt; + + return 0; +} + static int matroska_parse_frame(MatroskaDemuxContext *matroska, MatroskaTrack *track, AVStream *st, @@ -2445,6 +2565,14 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, if (res) goto end; + } else if (st->codec->codec_id == AV_CODEC_ID_WEBVTT) { + res = matroska_parse_webvtt(matroska, track, st, + data, lace_size[n], + timecode, lace_duration, + pos); + if (res) + goto end; + } else { res = matroska_parse_frame(matroska, track, st, data, lace_size[n], timecode, lace_duration, -- 1.8.3

6 42

Expert for video converting and streaming wanted
by Alexander Eberharter 19 Aug '13

19 Aug '13

We require guidance for developing a server application based on FFmpeg which is able to convert and stream files directly to a client application. The target machine is a windows server. M-BOX (www.mbox.at) is a database tool written in Java and SQL Server for archiving multimedia data such as documents, images and video files. It is developed by M-BOX GmbH located in Schwaz, Austria. Our clients demand functionality to convert files at the server and maybe also stream them to the client application (without buffering the whole converted file on the server!). Your job is to guide our development of an application which is capable of doing these things mentioned above. Input to the application are multimedia files of any common format the output should cover as many as possible of common multimedia formats. Please reply to info(a)m-box.at <mailto:info@m-box.at> if you are interested Alexander Eberharter M-BOX GmbH Innsbrucker Strasse 16 6130 Schwaz Austria

1 0

[PATCH] matroskadec: Fix TTA duration calculation
by James Almer 18 Aug '13

18 Aug '13

Still not perfect due to rounding errors, but several samples should demux correctly now instead of none. Signed-off-by: James Almer <jamrial(a)gmail.com> --- libavformat/matroskadec.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavformat/matroskadec.c b/libavformat/matroskadec.c index d52877c..e4b11a2 100644 --- a/libavformat/matroskadec.c +++ b/libavformat/matroskadec.c @@ -1738,7 +1738,7 @@ static int matroska_read_header(AVFormatContext *s) avio_wl16(&b, track->audio.channels); avio_wl16(&b, track->audio.bitdepth); avio_wl32(&b, track->audio.out_samplerate); - avio_wl32(&b, matroska->ctx->duration * track->audio.out_samplerate); + avio_wl32(&b, av_rescale(matroska->ctx->duration, track->audio.out_samplerate, AV_TIME_BASE)); } else if (codec_id == AV_CODEC_ID_RV10 || codec_id == AV_CODEC_ID_RV20 || codec_id == AV_CODEC_ID_RV30 || codec_id == AV_CODEC_ID_RV40) { extradata_offset = 26; -- 1.8.1.5

3 20

[PATCH] lavf/matroska: Adding the new SeekPreRoll element
by Vignesh Venkatasubramanian 17 Aug '13

17 Aug '13

In order to encapsuate Opus in Matroska, there is a new element that has been added to the Matroska Spec, SeekPreRoll. It has the duration in nanoseconds that has to be decoded before every seek. Spec: http://matroska.org/technical/specs/index.html#SeekPreRoll Proposal for encapsulateing Opus in Matroska: http://wiki.xiph.org/MatroskaOpus Signed-off-by: Vignesh Venkatasubramanian <vigneshv(a)google.com> --- libavformat/matroska.h | 1 + libavformat/matroskaenc.c | 8 ++++++++ 2 files changed, 9 insertions(+) diff --git a/libavformat/matroska.h b/libavformat/matroska.h index 2eb2fe6..9c8071a 100644 --- a/libavformat/matroska.h +++ b/libavformat/matroska.h @@ -91,6 +91,7 @@ #define MATROSKA_ID_CODECINFOURL 0x3B4040 #define MATROSKA_ID_CODECDOWNLOADURL 0x26B240 #define MATROSKA_ID_CODECDECODEALL 0xAA +#define MATROSKA_ID_SEEKPREROLL 0x56BB #define MATROSKA_ID_TRACKNAME 0x536E #define MATROSKA_ID_TRACKLANGUAGE 0x22B59C #define MATROSKA_ID_TRACKFLAGENABLED 0xB9 diff --git a/libavformat/matroskaenc.c b/libavformat/matroskaenc.c index c201d5e..56109f9 100644 --- a/libavformat/matroskaenc.c +++ b/libavformat/matroskaenc.c @@ -121,6 +121,9 @@ typedef struct MatroskaMuxContext { /** per-cuepoint - 2 1-byte EBML IDs, 2 1-byte EBML sizes, 8-byte uint max */ #define MAX_CUEPOINT_SIZE(num_tracks) 12 + MAX_CUETRACKPOS_SIZE*num_tracks +/** Seek preroll value for opus */ +#define OPUS_SEEK_PREROLL 80000000 + static int ebml_id_size(unsigned int id) { @@ -642,6 +645,10 @@ static int mkv_write_tracks(AVFormatContext *s) } } + if (codec->codec_id == AV_CODEC_ID_OPUS) { + put_ebml_uint(pb, MATROSKA_ID_SEEKPREROLL, OPUS_SEEK_PREROLL); + } + if (mkv->mode == MODE_WEBM && !(codec->codec_id == AV_CODEC_ID_VP8 || codec->codec_id == AV_CODEC_ID_VP9 || ((codec->codec_id == AV_CODEC_ID_OPUS)&&(codec->strict_std_compliance <= FF_COMPLIANCE_EXPERIMENTAL)) || @@ -1721,6 +1728,7 @@ AVOutputFormat ff_matroska_audio_muxer = { .priv_data_size = sizeof(MatroskaMuxContext), .audio_codec = CONFIG_LIBVORBIS_ENCODER ? AV_CODEC_ID_VORBIS : AV_CODEC_ID_AC3, + AV_CODEC_ID_OPUS, .video_codec = AV_CODEC_ID_NONE, .write_header = mkv_write_header, .write_packet = mkv_write_packet, -- 1.8.3

4 6

[RFC] fate: Add exif test.
by Thilo Borgmann 17 Aug '13

17 Aug '13

Hi, to add a test for exif metadata, I need to know if we can somehow check these better than by console output? Do we have tests for MP3 tags or similar? I need a little guidance through the Makefile-magic... What I have done until now (see attached) is including a new .mak file into tests/Makefile, and how get the corresponding target called. What could the target execute to test for metatdata? (CMD = ...?) How generate a corresponding reference file? Thanks, Thilo p.s. is IFF the correct demuxer for TIFF files? --- tests/Makefile | 1 + tests/fate/exif.mak | 5 +++++ 2 files changed, 6 insertions(+) create mode 100644 tests/fate/exif.mak diff --git a/tests/Makefile b/tests/Makefile index 914448e..61423f4 100644 --- a/tests/Makefile +++ b/tests/Makefile @@ -93,6 +93,7 @@ include $(SRC_PATH)/tests/fate/demux.mak include $(SRC_PATH)/tests/fate/dfa.mak include $(SRC_PATH)/tests/fate/dpcm.mak include $(SRC_PATH)/tests/fate/ea.mak +include $(SRC_PATH)/tests/fate/exif.mak include $(SRC_PATH)/tests/fate/ffmpeg.mak include $(SRC_PATH)/tests/fate/ffprobe.mak include $(SRC_PATH)/tests/fate/filter-audio.mak diff --git a/tests/fate/exif.mak b/tests/fate/exif.mak new file mode 100644 index 0000000..8295b18 --- /dev/null +++ b/tests/fate/exif.mak @@ -0,0 +1,5 @@ +FATE_SAMPLES_EXIF-$(call DEMDEC, IFF, TIFF) += fate-exif-image +fate-exif-iamge: CMD = framecrc -i $(TARGET_SAMPLES)/exif/image_small.tiff + +FATE_SAMPLES_FFMPEG += $(FATE_SAMPLES_EXIF-yes) +fate-exif: $(FATE_SAMPLES_EXIF-yes) -- 1.8.3.2

2 2

[PATCH] avisynth: Support video input from AviSynth 2.5 properly.
by Stephen Hutchinson 17 Aug '13

17 Aug '13

Uses the 2.5 compatibility header included with the variant of FFMS2 that uses AviSynth's C-interface. A copy of this header is now provided in compat/avisynth. avs_get_row_size_p and avs_get_height_p changed between versions 2.5 and 2.6. Since the avisynth_c.h header that avformat uses assumes AviSynth 2.6, it would cause 2.5 to crash if given any kind of real video (the Version() function was known to work, though). AvxSynth was unaffected by this issue because, despite being based on AviSynth 2.5.8 and using 2.5.8's interface version number of 3, it actually uses 2.6's versions of these functions. --- compat/avisynth/avisynth_c_25.h | 68 +++++++++++++++++++++++++++++++++++++++++ libavformat/avisynth.c | 33 ++++++++------------ 2 files changed, 81 insertions(+), 20 deletions(-) create mode 100644 compat/avisynth/avisynth_c_25.h diff --git a/compat/avisynth/avisynth_c_25.h b/compat/avisynth/avisynth_c_25.h new file mode 100644 index 0000000..9288761 --- /dev/null +++ b/compat/avisynth/avisynth_c_25.h @@ -0,0 +1,68 @@ +// Copyright (c) 2011 FFmpegSource Project +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +/* these are defines/functions that are used and were changed in the switch to 2.6 + * and are needed to maintain full compatility with 2.5 */ + +enum { + AVS_CS_YV12_25 = 1<<3 | AVS_CS_YUV | AVS_CS_PLANAR, // y-v-u, planar + AVS_CS_I420_25 = 1<<4 | AVS_CS_YUV | AVS_CS_PLANAR, // y-u-v, planar +}; + +AVSC_INLINE int avs_get_height_p_25(const AVS_VideoFrame * p, int plane) { + switch (plane) + { + case AVS_PLANAR_U: case AVS_PLANAR_V: + if (p->pitchUV) + return p->height>>1; + return 0; + } + return p->height;} + +AVSC_INLINE int avs_get_row_size_p_25(const AVS_VideoFrame * p, int plane) { + int r; + switch (plane) + { + case AVS_PLANAR_U: case AVS_PLANAR_V: + if (p->pitchUV) + return p->row_size>>1; + else + return 0; + case AVS_PLANAR_U_ALIGNED: case AVS_PLANAR_V_ALIGNED: + if (p->pitchUV) + { + r = ((p->row_size+AVS_FRAME_ALIGN-1)&(~(AVS_FRAME_ALIGN-1)) )>>1; // Aligned rowsize + if (r < p->pitchUV) + return r; + return p->row_size>>1; + } + else + return 0; + case AVS_PLANAR_Y_ALIGNED: + r = (p->row_size+AVS_FRAME_ALIGN-1)&(~(AVS_FRAME_ALIGN-1)); // Aligned rowsize + if (r <= p->pitch) + return r; + return p->row_size; + } + return p->row_size; +} + +AVSC_INLINE int avs_is_yv12_25(const AVS_VideoInfo * p) + { return ((p->pixel_type & AVS_CS_YV12_25) == AVS_CS_YV12_25)||((p->pixel_type & AVS_CS_I420_25) == AVS_CS_I420_25); } diff --git a/libavformat/avisynth.c b/libavformat/avisynth.c index 8a480ed..afacf04 100644 --- a/libavformat/avisynth.c +++ b/libavformat/avisynth.c @@ -36,6 +36,7 @@ #include <windows.h> #undef EXTERN_C #include "compat/avisynth/avisynth_c.h" + #include "compat/avisynth/avisynth_c_25.h" #define AVISYNTH_LIB "avisynth" #else #include <dlfcn.h> @@ -471,9 +472,20 @@ static int avisynth_read_packet_video(AVFormatContext *s, AVPacket *pkt, int dis for (i = 0; i < avs->n_planes; i++) { plane = avs->planes[i]; src_p = avs_get_read_ptr_p(frame, plane); + pitch = avs_get_pitch_p(frame, plane); + +#ifdef _WIN32 + if (avs_library->avs_get_version(avs->clip) == 3) { + rowsize = avs_get_row_size_p_25(frame, plane); + planeheight = avs_get_height_p_25(frame, plane); + } else { + rowsize = avs_get_row_size_p(frame, plane); + planeheight = avs_get_height_p(frame, plane); + } +#else rowsize = avs_get_row_size_p(frame, plane); planeheight = avs_get_height_p(frame, plane); - pitch = avs_get_pitch_p(frame, plane); +#endif // Flip RGB video. if (avs_is_rgb24(avs->vi) || avs_is_rgb(avs->vi)) { @@ -481,26 +493,7 @@ static int avisynth_read_packet_video(AVFormatContext *s, AVPacket *pkt, int dis pitch = -pitch; } - // An issue with avs_bit_blt on 2.5.8 prevents video from working correctly. - // This problem doesn't exist for 2.6 and AvxSynth, so enable the workaround - // for 2.5.8 only. This only displays the warning and exits if the script has - // video. 2.5.8's internal interface version is 3, so avs_get_version allows - // it to work only in the circumstance that the interface is 5 or higher (4 is - // unused). There's a strong chance that AvxSynth, having been based on 2.5.8, - // would also be identified as interface version 3, but since AvxSynth doesn't - // suffer from this problem, special-case it. -#ifdef _WIN32 - if (avs_library->avs_get_version(avs->clip) > 3) { avs_library->avs_bit_blt(avs->env, dst_p, rowsize, src_p, pitch, rowsize, planeheight); - } else { - av_log(s, AV_LOG_ERROR, "Video input from AviSynth 2.5.8 is not supported. Please upgrade to 2.6.\n"); - avs->error = 1; - av_freep(&pkt->data); - return AVERROR_UNKNOWN; - } -#else - avs_library->avs_bit_blt(avs->env, dst_p, rowsize, src_p, pitch, rowsize, planeheight); -#endif dst_p += rowsize * planeheight; } -- 1.8.1.2

2 1

[PATCH] lavf/tee: add support for bitstream filtering
by Stefano Sabatini 16 Aug '13

16 Aug '13

TODO: add documentation, bump minor --- libavformat/tee.c | 231 +++++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 221 insertions(+), 10 deletions(-) diff --git a/libavformat/tee.c b/libavformat/tee.c index 7b8b371..96ced3e 100644 --- a/libavformat/tee.c +++ b/libavformat/tee.c @@ -27,10 +27,16 @@ #define MAX_SLAVES 16 +typedef struct { + AVFormatContext *fmt_ctx; + AVBitStreamFilterContext **bsf_ctxs; ///< bitstream filters per stream +} TeeSlave; + typedef struct TeeContext { const AVClass *class; unsigned nb_slaves; - AVFormatContext *slaves[MAX_SLAVES]; + TeeSlave slaves[MAX_SLAVES]; + char *bsfs; } TeeContext; static const char *const slave_delim = "|"; @@ -38,10 +44,18 @@ static const char *const slave_opt_open = "["; static const char *const slave_opt_close = "]"; static const char *const slave_opt_delim = ":]"; /* must have the close too */ +#define OFFSET(x) offsetof(TeeContext, x) +#define E AV_OPT_FLAG_ENCODING_PARAM +static const AVOption tee_options[] = { + { "tee_bsfs", "set bitstream filters for each slave", OFFSET(bsfs), AV_OPT_TYPE_STRING, {.str = NULL}, .flags=E }, + { NULL }, +}; + static const AVClass tee_muxer_class = { .class_name = "Tee muxer", .item_name = av_default_item_name, .version = LIBAVUTIL_VERSION_INT, + .option = tee_options, }; static int parse_slave_options(void *log, char *slave, @@ -82,7 +96,93 @@ fail: return ret; } -static int open_slave(AVFormatContext *avf, char *slave, AVFormatContext **ravf) +/** + * Parse bitstream filter, in the form: + * BSF ::= BSF_NAME[:STREAM_SPECIFIER] + */ +static int parse_bsf(void *log_ctx, char *bsf, + AVFormatContext *fmt_ctx, + AVBitStreamFilterContext **bsf_ctxs) +{ + int i, ret = 0; + char *bsf1 = av_strdup(bsf); + char *bsf_name, *spec; + + if (!bsf1) + return AVERROR(ENOMEM); + bsf_name = av_strtok(bsf1, ":", &spec); + + /* select the streams matched by the specifier */ + for (i = 0; i < fmt_ctx->nb_streams; i++) { + AVBitStreamFilterContext *bsf_ctx; + + if (spec && spec[0]) { + ret = avformat_match_stream_specifier(fmt_ctx, fmt_ctx->streams[i], spec); + if (ret < 0) { + av_log(log_ctx, AV_LOG_ERROR, + "Invalid stream specifier for bitstream filter '%s'\n", bsf); + goto end; + } + + if (ret == 0) /* no match */ + continue; + } + + bsf_ctx = av_bitstream_filter_init(bsf_name); + if (!bsf_ctx) { + av_log(log_ctx, AV_LOG_ERROR, + "Cannot create bitstream filter with name '%s' for stream %d, " + "unknown filter or internal error happened\n", + bsf_name, i); + ret = AVERROR_UNKNOWN; + goto end; + } + + /* append bsf context to the list of per-stream bsf contexts */ + if (!bsf_ctxs[i]) { + bsf_ctxs[i] = bsf_ctx; + } else { + AVBitStreamFilterContext *bsf_ctx1 = bsf_ctxs[i]; + while (bsf_ctx1->next) + bsf_ctx1 = bsf_ctx1->next; + bsf_ctx1->next = bsf_ctx; + } + } + +end: + av_free(bsf1); + return ret; +} + +/** + * Parse list of bitstream filters in the form: + * BSFS ::= BSF[,BSFS] + * BSF ::= BSF_NAME[:STREAM_SPECIFIER] + */ +static int parse_slave_bsfs(void *log_ctx, char *slave_bsfs, + AVFormatContext *fmt_ctx, + AVBitStreamFilterContext **bsf_ctx) +{ + char *bsf, *slave_bsfs1 = av_strdup(slave_bsfs); + char *saveptr, *buf = slave_bsfs1; + int ret; + + if (!slave_bsfs1) + return AVERROR(ENOMEM); + + while (bsf = av_strtok(buf, ",", &saveptr)) { + buf = NULL; + ret = parse_bsf(log_ctx, bsf, fmt_ctx, bsf_ctx); + if (ret < 0) + goto end; + } + +end: + av_free(slave_bsfs1); + return ret; +} + +static int open_slave(AVFormatContext *avf, char *slave, char *slave_bsfs, TeeSlave *tee_slave) { int i, ret; AVDictionary *options = NULL; @@ -138,6 +238,7 @@ static int open_slave(AVFormatContext *avf, char *slave, AVFormatContext **ravf) slave, av_err2str(ret)); goto fail; } + if (options) { entry = NULL; while ((entry = av_dict_get(options, "", entry, AV_DICT_IGNORE_SUFFIX))) @@ -146,7 +247,16 @@ static int open_slave(AVFormatContext *avf, char *slave, AVFormatContext **ravf) goto fail; } - *ravf = avf2; + tee_slave->fmt_ctx = avf2; + tee_slave->bsf_ctxs = av_mallocz(avf2->nb_streams * sizeof(*(tee_slave->bsf_ctxs))); + if (!tee_slave->bsf_ctxs) + return AVERROR(ENOMEM); + + /* generate the array of bitstream filters */ + if (slave_bsfs) { + if ((ret = parse_slave_bsfs(avf, slave_bsfs, avf2, tee_slave->bsf_ctxs)) < 0) + return ret; + } return 0; fail: @@ -158,14 +268,50 @@ static void close_slaves(AVFormatContext *avf) { TeeContext *tee = avf->priv_data; AVFormatContext *avf2; - unsigned i; + unsigned i, j; for (i = 0; i < tee->nb_slaves; i++) { - avf2 = tee->slaves[i]; + avf2 = tee->slaves[i].fmt_ctx; + + for (j = 0; j < avf2->nb_streams; j++) { + AVBitStreamFilterContext *bsf_ctx = tee->slaves[i].bsf_ctxs[j]; + while (bsf_ctx) { + bsf_ctx = bsf_ctx->next; + av_bitstream_filter_close(bsf_ctx); + } + } + avio_close(avf2->pb); avf2->pb = NULL; avformat_free_context(avf2); - tee->slaves[i] = NULL; + tee->slaves[i].fmt_ctx = NULL; + } +} + +static void log_slave(TeeSlave *slave, void *log_ctx, int log_level) +{ + int i; + av_log(log_ctx, log_level, "filename:'%s' format:%s\n", + slave->fmt_ctx->filename, slave->fmt_ctx->oformat->name); + for (i = 0; i < slave->fmt_ctx->nb_streams; i++) { + AVStream *st = slave->fmt_ctx->streams[i]; + AVBitStreamFilterContext *bsf_ctx = slave->bsf_ctxs[i]; + + av_log(log_ctx, log_level, " stream:%d codec:%s type:%s", + i, avcodec_get_name(st->codec->codec_id), + st->codec->codec ? + av_get_media_type_string(st->codec->codec->type) : "copy"); + if (bsf_ctx) { + int first = 1; + av_log(log_ctx, log_level, " bsfs:"); + while (bsf_ctx) { + av_log(log_ctx, log_level, "%s%s", first ? "" : ",", + bsf_ctx->filter->name); + bsf_ctx = bsf_ctx->next; + first = 0; + } + } + av_log(log_ctx, log_level, "\n"); } } @@ -174,7 +320,9 @@ static int tee_write_header(AVFormatContext *avf) TeeContext *tee = avf->priv_data; unsigned nb_slaves = 0, i; const char *filename = avf->filename; - char *slaves[MAX_SLAVES]; + const char *bsfs = tee->bsfs; + char *slaves [MAX_SLAVES]; + char *slave_bsfs[MAX_SLAVES] = { NULL }; int ret; while (*filename) { @@ -192,10 +340,29 @@ static int tee_write_header(AVFormatContext *avf) filename++; } + i = 0; + while (bsfs && *bsfs) { + if (i >= nb_slaves) { + av_log(avf, AV_LOG_ERROR, + "%d bitstream filter sequences are specified, but only %d slaves " + "were specified, aborting\n", i+1, nb_slaves); + ret = AVERROR(EINVAL); + goto fail; + } + if (!(slave_bsfs[i++] = av_get_token(&bsfs, slave_delim))) { + ret = AVERROR(ENOMEM); + goto fail; + } + if (strspn(bsfs, slave_delim)) + bsfs++; + } + for (i = 0; i < nb_slaves; i++) { - if ((ret = open_slave(avf, slaves[i], &tee->slaves[i])) < 0) + if ((ret = open_slave(avf, slaves[i], slave_bsfs[i], &tee->slaves[i])) < 0) goto fail; + log_slave(&tee->slaves[i], avf, AV_LOG_VERBOSE); av_freep(&slaves[i]); + av_freep(&slave_bsfs[i]); } tee->nb_slaves = nb_slaves; @@ -208,6 +375,48 @@ fail: return ret; } +static int filter_packet(void *log_ctx, AVPacket *pkt, + AVCodecContext *enc_ctx, AVBitStreamFilterContext *bsf_ctx) +{ + while (bsf_ctx) { + AVPacket new_pkt = *pkt; + int ret = av_bitstream_filter_filter(bsf_ctx, enc_ctx, NULL, + &new_pkt.data, &new_pkt.size, + pkt->data, pkt->size, + pkt->flags & AV_PKT_FLAG_KEY); + if (ret == 0 && new_pkt.data != pkt->data && new_pkt.destruct) { + uint8_t *buf = av_malloc(new_pkt.size + FF_INPUT_BUFFER_PADDING_SIZE); + if (buf) { + memcpy(buf, new_pkt.data, new_pkt.size); + memset(buf + new_pkt.size, 0, FF_INPUT_BUFFER_PADDING_SIZE); + new_pkt.data = buf; + new_pkt.buf = NULL; + ret = 1; + } else + ret = AVERROR(ENOMEM); + } + + if (ret > 0) { + av_free_packet(pkt); + new_pkt.buf = av_buffer_create(new_pkt.data, new_pkt.size, + av_buffer_default_free, NULL, 0); + if (!new_pkt.buf) + return AVERROR(ENOMEM); + } else if (ret < 0) { + av_log(log_ctx, AV_LOG_ERROR, + "Failed to filter bitstream %s for stream %d with codec %s", + bsf_ctx->filter->name, pkt->stream_index, + enc_ctx->codec ? enc_ctx->codec->name : "copy"); + return ret; + } + *pkt = new_pkt; + + bsf_ctx = bsf_ctx->next; + } + + return 0; +} + static int tee_write_trailer(AVFormatContext *avf) { TeeContext *tee = avf->priv_data; @@ -216,7 +425,7 @@ static int tee_write_trailer(AVFormatContext *avf) unsigned i; for (i = 0; i < tee->nb_slaves; i++) { - avf2 = tee->slaves[i]; + avf2 = tee->slaves[i].fmt_ctx; if ((ret = av_write_trailer(avf2)) < 0) if (!ret_all) ret_all = ret; @@ -241,7 +450,7 @@ static int tee_write_packet(AVFormatContext *avf, AVPacket *pkt) AVRational tb, tb2; for (i = 0; i < tee->nb_slaves; i++) { - avf2 = tee->slaves[i]; + avf2 = tee->slaves[i].fmt_ctx; s = pkt->stream_index; if (s >= avf2->nb_streams) { if (!ret_all) @@ -259,6 +468,8 @@ static int tee_write_packet(AVFormatContext *avf, AVPacket *pkt) pkt2.pts = av_rescale_q(pkt->pts, tb, tb2); pkt2.dts = av_rescale_q(pkt->dts, tb, tb2); pkt2.duration = av_rescale_q(pkt->duration, tb, tb2); + + filter_packet(avf2, &pkt2, avf2->streams[s]->codec, tee->slaves[i].bsf_ctxs[s]); if ((ret = av_interleaved_write_frame(avf2, &pkt2)) < 0) if (!ret_all) ret_all = ret; -- 1.8.1.2

2 21

Transcoding HLS segments post-hoc
by Ray Hilton 16 Aug '13

16 Aug '13

Hi, Sorry, this isn't directly to do with internal dev, but the consulting page recommended emailing this list if I wasn't sure who to contact. Rather than individually emailing everyone with more or less the same thing, i'll post here and hopefully someone will be able to help! We might be able to get money together to pay for consulting if the solution to this problem is some custom ffmpeg work. Bit of background: we are providing a streaming live/on-demand solution to the community radio sector in Australia. We're trying to find a way to provide HLS streaming from live, continuous audio in a variety of bitrates and codecs and ideally take advantage of adaptive streaming (i.e. segments are in sync). First, is this possible: I am segmenting audio from a live stream using AAC at 128k. I send those segments to our cluster and they transcode them into various other bitrates/formats using things like neroaac (better low bitrate quality) and lame. This works, but it introduces gaps between the segments. This is an ideal solution as the origin (radio stations) upload just the highest quality format and we distribute the workload to do the rest. We can also use the same implementation for "live" as well as "on demand". What is ffmpeg doing internally when segmenting to avoid the gaps? The second option is to use small machines in the studio to transcode from audio device to multiple nitrates/formats. Ideally, for adaptive streaming, we want these segments to be in sync - is that possible? -- Ray Hilton (maker of mobile apps) - 13/243 Collins St, Melbourne VIC 3000 | +61 (0) 430 484 708 | http://ray.sh

2 1

[PATCH 1/2] lavfi/transpose: support slice threading
by Paul B Mahol 16 Aug '13

16 Aug '13

Signed-off-by: Paul B Mahol <onemda(a)gmail.com> --- libavfilter/vf_transpose.c | 72 ++++++++++++++++++++++++++++++---------------- 1 file changed, 47 insertions(+), 25 deletions(-) diff --git a/libavfilter/vf_transpose.c b/libavfilter/vf_transpose.c index 3ee9c6d..82f68e5 100644 --- a/libavfilter/vf_transpose.c +++ b/libavfilter/vf_transpose.c @@ -133,31 +133,19 @@ static AVFrame *get_video_buffer(AVFilterLink *inlink, int w, int h) ff_default_get_video_buffer(inlink, w, h); } -static int filter_frame(AVFilterLink *inlink, AVFrame *in) +typedef struct ThreadData { + AVFrame *in, *out; +} ThreadData; + +static int filter_slice(AVFilterContext *ctx, void *arg, int jobnr, + int nb_jobs) { - TransContext *trans = inlink->dst->priv; - AVFilterLink *outlink = inlink->dst->outputs[0]; - AVFrame *out; + TransContext *trans = ctx->priv; + ThreadData *td = arg; + AVFrame *out = td->out; + AVFrame *in = td->in; int plane; - if (trans->passthrough) - return ff_filter_frame(outlink, in); - - out = ff_get_video_buffer(outlink, outlink->w, outlink->h); - if (!out) { - av_frame_free(&in); - return AVERROR(ENOMEM); - } - - out->pts = in->pts; - - if (in->sample_aspect_ratio.num == 0) { - out->sample_aspect_ratio = in->sample_aspect_ratio; - } else { - out->sample_aspect_ratio.num = in->sample_aspect_ratio.den; - out->sample_aspect_ratio.den = in->sample_aspect_ratio.num; - } - for (plane = 0; out->data[plane]; plane++) { int hsub = plane == 1 || plane == 2 ? trans->hsub : 0; int vsub = plane == 1 || plane == 2 ? trans->vsub : 0; @@ -165,12 +153,14 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in) int inh = in->height >> vsub; int outw = FF_CEIL_RSHIFT(out->width, hsub); int outh = FF_CEIL_RSHIFT(out->height, vsub); + int start = (outh * jobnr ) / nb_jobs; + int end = (outh * (jobnr+1)) / nb_jobs; uint8_t *dst, *src; int dstlinesize, srclinesize; int x, y; - dst = out->data[plane]; dstlinesize = out->linesize[plane]; + dst = out->data[plane] + start * dstlinesize; src = in->data[plane]; srclinesize = in->linesize[plane]; @@ -180,11 +170,11 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in) } if (trans->dir&2) { - dst += out->linesize[plane] * (outh-1); + dst = out->data[plane] + dstlinesize * (outh-start-1); dstlinesize *= -1; } - for (y = 0; y < outh; y++) { + for (y = start; y < end; y++) { switch (pixstep) { case 1: for (x = 0; x < outw; x++) @@ -219,6 +209,37 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in) } } + return 0; +} + +static int filter_frame(AVFilterLink *inlink, AVFrame *in) +{ + AVFilterContext *ctx = inlink->dst; + TransContext *trans = ctx->priv; + AVFilterLink *outlink = ctx->outputs[0]; + ThreadData td; + AVFrame *out; + + if (trans->passthrough) + return ff_filter_frame(outlink, in); + + out = ff_get_video_buffer(outlink, outlink->w, outlink->h); + if (!out) { + av_frame_free(&in); + return AVERROR(ENOMEM); + } + + out->pts = in->pts; + + if (in->sample_aspect_ratio.num == 0) { + out->sample_aspect_ratio = in->sample_aspect_ratio; + } else { + out->sample_aspect_ratio.num = in->sample_aspect_ratio.den; + out->sample_aspect_ratio.den = in->sample_aspect_ratio.num; + } + + td.in = in, td.out = out; + ctx->internal->execute(ctx, filter_slice, &td, NULL, FFMIN(outlink->h, ctx->graph->nb_threads)); av_frame_free(&in); return ff_filter_frame(outlink, out); } @@ -274,4 +295,5 @@ AVFilter avfilter_vf_transpose = { .inputs = avfilter_vf_transpose_inputs, .outputs = avfilter_vf_transpose_outputs, + .flags = AVFILTER_FLAG_SLICE_THREADS, }; -- 1.7.11.2

2 6

Fixed point implementations of AAC decoder, AC3 decoder and 32bit FFT
by Nedeljko Babic 16 Aug '13

16 Aug '13

Hello, Three patches with implementation of fixed point AAC decoder are sent. Since this implementation depends on implementation of 32 bit fixed point FFT and some functions implemented for fixed point AC3 decoder that were submitted earlier and since these implementations are not reviewed yet, I am resubmitting them rebased to the newest version of FFmpeg. Regards, -Nedeljko

5 21

[PATCH] sws: add dither enum
by Michael Niedermayer 15 Aug '13

15 Aug '13

This allows specifying more dither algorithms without using up flags and without ambiguities. Also initialize the new field based on the flags and use it Signed-off-by: Michael Niedermayer <michaelni(a)gmx.at> --- libswscale/options.c | 6 ++++++ libswscale/output.c | 6 +++--- libswscale/swscale.c | 2 +- libswscale/swscale_internal.h | 9 +++++++++ libswscale/swscale_unscaled.c | 2 +- libswscale/utils.c | 12 ++++++++---- libswscale/version.h | 2 +- 7 files changed, 29 insertions(+), 10 deletions(-) diff --git a/libswscale/options.c b/libswscale/options.c index 81d8c28..f7edd21 100644 --- a/libswscale/options.c +++ b/libswscale/options.c @@ -69,6 +69,12 @@ static const AVOption swscale_options[] = { { "dst_v_chr_pos", "destination vertical chroma position in luma grid/256" , OFFSET(dst_v_chr_pos), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 512, VE }, { "dst_h_chr_pos", "destination horizontal chroma position in luma grid/256", OFFSET(dst_h_chr_pos), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 512, VE }, + { "sws_dither", "Dither algorithm", OFFSET(dither), AV_OPT_TYPE_INT, { .i64 = SWS_DITHER_AUTO }, 0, 1, VE, "sws_dither" }, + { "auto", "leave choice to sws", 0, AV_OPT_TYPE_CONST, { .i64 = SWS_DITHER_AUTO }, INT_MIN, INT_MAX, VE, "sws_dither" }, + { "off", "no dither", 0, AV_OPT_TYPE_CONST, { .i64 = SWS_DITHER_NONE }, INT_MIN, INT_MAX, VE, "sws_dither" }, + { "bayer", "bayer dither", 0, AV_OPT_TYPE_CONST, { .i64 = SWS_DITHER_BAYER }, INT_MIN, INT_MAX, VE, "sws_dither" }, + { "ed", "error diffusion", 0, AV_OPT_TYPE_CONST, { .i64 = SWS_DITHER_ED }, INT_MIN, INT_MAX, VE, "sws_dither" }, + { NULL } }; diff --git a/libswscale/output.c b/libswscale/output.c index 54b6ebd..2d390c7 100644 --- a/libswscale/output.c +++ b/libswscale/output.c @@ -349,7 +349,7 @@ yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter, Y1 = av_clip_uint8(Y1); Y2 = av_clip_uint8(Y2); } - if (c->flags & SWS_ERROR_DIFFUSION) { + if (c->dither == SWS_DITHER_ED) { Y1 += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4; c->dither_error[0][i] = err; acc = 2*acc + (Y1 >= 128); @@ -386,7 +386,7 @@ yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2], int yalpha1 = 4096 - yalpha; int i; - if (c->flags & SWS_ERROR_DIFFUSION) { + if (c->dither == SWS_DITHER_ED) { int err = 0; int acc = 0; for (i = 0; i < dstW; i +=2) { @@ -443,7 +443,7 @@ yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0, const uint8_t * const d128 = dither_8x8_220[y & 7]; int i; - if (c->flags & SWS_ERROR_DIFFUSION) { + if (c->dither == SWS_DITHER_ED) { int err = 0; int acc = 0; for (i = 0; i < dstW; i +=2) { diff --git a/libswscale/swscale.c b/libswscale/swscale.c index 6721dc3..87b867d 100644 --- a/libswscale/swscale.c +++ b/libswscale/swscale.c @@ -1051,7 +1051,7 @@ int attribute_align_arg sws_scale(struct SwsContext *c, src2[0] = base; } - if (!srcSliceY && (c->flags & SWS_BITEXACT) && (c->flags & SWS_ERROR_DIFFUSION) && c->dither_error[0]) + if (!srcSliceY && (c->flags & SWS_BITEXACT) && c->dither == SWS_DITHER_ED && c->dither_error[0]) for (i = 0; i < 4; i++) memset(c->dither_error[i], 0, sizeof(c->dither_error[0][0]) * (c->dstW+2)); diff --git a/libswscale/swscale_internal.h b/libswscale/swscale_internal.h index e7cc26e..a14cc02 100644 --- a/libswscale/swscale_internal.h +++ b/libswscale/swscale_internal.h @@ -61,6 +61,13 @@ struct SwsContext; +typedef enum SwsDither { + SWS_DITHER_NONE = 0, + SWS_DITHER_AUTO, + SWS_DITHER_BAYER, + SWS_DITHER_ED, +} SwsDither; + typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[], int srcStride[], int srcSliceY, int srcSliceH, uint8_t *dst[], int dstStride[]); @@ -593,6 +600,8 @@ typedef struct SwsContext { void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width); int needs_hcscale; ///< Set if there are chroma planes to be converted. + + SwsDither dither; } SwsContext; //FIXME check init (where 0) diff --git a/libswscale/swscale_unscaled.c b/libswscale/swscale_unscaled.c index 3b07800..22196bd 100644 --- a/libswscale/swscale_unscaled.c +++ b/libswscale/swscale_unscaled.c @@ -1180,7 +1180,7 @@ void ff_get_unscaled_swscale(SwsContext *c) /* yuv2bgr */ if ((srcFormat == AV_PIX_FMT_YUV420P || srcFormat == AV_PIX_FMT_YUV422P || srcFormat == AV_PIX_FMT_YUVA420P) && isAnyRGB(dstFormat) && - !(flags & (SWS_ACCURATE_RND|SWS_ERROR_DIFFUSION)) && !(dstH & 1)) { + !(flags & SWS_ACCURATE_RND) && c->dither != SWS_DITHER_ED && !(dstH & 1)) { c->swScale = ff_yuv2rgb_get_func_ptr(c); } diff --git a/libswscale/utils.c b/libswscale/utils.c index 5aacd81..f5826c3 100644 --- a/libswscale/utils.c +++ b/libswscale/utils.c @@ -1196,23 +1196,27 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter, } } + if (c->dither == SWS_DITHER_AUTO) { + if (flags & SWS_ERROR_DIFFUSION) + c->dither = SWS_DITHER_ED; + } + if(dstFormat == AV_PIX_FMT_BGR4_BYTE || dstFormat == AV_PIX_FMT_RGB4_BYTE || dstFormat == AV_PIX_FMT_BGR8 || dstFormat == AV_PIX_FMT_RGB8) { - if (flags & SWS_ERROR_DIFFUSION && !(flags & SWS_FULL_CHR_H_INT)) { + if (c->dither == SWS_DITHER_ED && !(flags & SWS_FULL_CHR_H_INT)) { av_log(c, AV_LOG_DEBUG, "Error diffusion dither is only supported in full chroma interpolation for destination format '%s'\n", av_get_pix_fmt_name(dstFormat)); flags |= SWS_FULL_CHR_H_INT; c->flags = flags; } - if (!(flags & SWS_ERROR_DIFFUSION) && (flags & SWS_FULL_CHR_H_INT)) { + if (c->dither != SWS_DITHER_ED && (flags & SWS_FULL_CHR_H_INT)) { av_log(c, AV_LOG_DEBUG, "Ordered dither is not supported in full chroma interpolation for destination format '%s'\n", av_get_pix_fmt_name(dstFormat)); - flags |= SWS_ERROR_DIFFUSION; - c->flags = flags; + c->dither = SWS_DITHER_ED; } } if (isPlanarRGB(dstFormat)) { diff --git a/libswscale/version.h b/libswscale/version.h index 1c45209..06f119a 100644 --- a/libswscale/version.h +++ b/libswscale/version.h @@ -27,7 +27,7 @@ #include "libavutil/avutil.h" #define LIBSWSCALE_VERSION_MAJOR 2 -#define LIBSWSCALE_VERSION_MINOR 4 +#define LIBSWSCALE_VERSION_MINOR 5 #define LIBSWSCALE_VERSION_MICRO 100 #define LIBSWSCALE_VERSION_INT AV_VERSION_INT(LIBSWSCALE_VERSION_MAJOR, \ -- 1.7.9.5

4 7

[PATCH] avformat/riff: add DM4V FourCC
by Piotr Bandurski 15 Aug '13

15 Aug '13

2 1

[PATCH] lavfi/tile: make color of padding configurable
by Paul B Mahol 15 Aug '13

15 Aug '13

Signed-off-by: Paul B Mahol <onemda(a)gmail.com> --- doc/filters.texi | 4 ++++ libavfilter/vf_tile.c | 5 +++-- 2 files changed, 7 insertions(+), 2 deletions(-) diff --git a/doc/filters.texi b/doc/filters.texi index 9a871b9..d390ff4 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -7102,6 +7102,10 @@ Set the inner border thickness (i.e. the number of pixels between frames). For more advanced padding options (such as having different values for the edges), refer to the pad video filter. +@item color +Specify the color of the padded area, it can be the name of a color +(case insensitive match) or a 0xRRGGBB[AA] sequence. +The default value of @var{color} is "black". @end table @subsection Examples diff --git a/libavfilter/vf_tile.c b/libavfilter/vf_tile.c index f5eb763..9807e7e 100644 --- a/libavfilter/vf_tile.c +++ b/libavfilter/vf_tile.c @@ -41,6 +41,7 @@ typedef struct { FFDrawContext draw; FFDrawColor blank; AVFrame *out_ref; + uint8_t rgba_color[4]; } TileContext; #define REASONABLE_SIZE 1024 @@ -57,6 +58,7 @@ static const AVOption tile_options[] = { AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1024, FLAGS }, { "padding", "set inner border thickness in pixels", OFFSET(padding), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1024, FLAGS }, + { "color", "set the color of the padded area", OFFSET(rgba_color), AV_OPT_TYPE_COLOR, {.str = "black"}, .flags = FLAGS }, {NULL}, }; @@ -113,8 +115,7 @@ static int config_props(AVFilterLink *outlink) outlink->frame_rate = av_mul_q(inlink->frame_rate, (AVRational){ 1, tile->nb_frames }); ff_draw_init(&tile->draw, inlink->format, 0); - /* TODO make the color an option, or find an unified way of choosing it */ - ff_draw_color(&tile->draw, &tile->blank, (uint8_t[]){ 0, 0, 0, -1 }); + ff_draw_color(&tile->draw, &tile->blank, tile->rgba_color); outlink->flags |= FF_LINK_FLAG_REQUEST_LOOP; -- 1.7.11.2

3 4

Re: [FFmpeg-devel] [FFmpeg-user] question: Hi, I want to ask a question about capture audio from audio card
by Wei Gao 15 Aug '13

15 Aug '13

2013/8/3 Roger Pack <rogerdpack2(a)gmail.com> > I guess it is a bug in windows 8...will file it though, thank you. > > Hi, Roger Pack, sorry for the late reply because busy these days. Do you mean it is a bug in ffmpeg? Because we run perfect on win7 using dshow. do you want any information that I can submit to you such as any debug informantion? Thanks very much Best regards > > > >> >> > _______________________________________________ > >> >> > ffmpeg-devel mailing list > >> >> > ffmpeg-devel(a)ffmpeg.org > >> >> > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel > >> >> > > >> >> > >> > > >> > > _______________________________________________ > > ffmpeg-devel mailing list > > ffmpeg-devel(a)ffmpeg.org > > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel > > >

1 0