April 2018 - ffmpeg-user - lists.ffmpeg.org

-Wlto-type-mismatch
by Reindl Harald 21 Apr '18

21 Apr '18

https://ffmpeg.org/releases/ffmpeg-4.0.tar.xz well, build with lto would show such things ./configure --enable-lto --extra-cflags "-flto=8 -ffat-lto-objects -fwhole-program" --extra-ldflags="--enable-lto --extra-cflags "-flto=8 -ffat-lto-objects -fwhole-program" you sadly need "-flto=8" (in fact -flto=%(nproc) within a rpm-spec) additionally to --enable-lto because otherwise the build takes ages because it uses only one CPU because "make -j8" is not inherited src/libavformat/rtpdec_formats.h:41:5: warning: type of 'ff_h264_handle_aggregated_packet' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/video.h:27:10: warning: type of 'ff_null_get_video_buffer' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:427:5: warning: type of 'ff_filter_init_hw_frames' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:160:5: warning: type of 'ff_outlink_get_status' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:147:6: warning: type of 'ff_inlink_set_status' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:141:6: warning: type of 'ff_inlink_request_frame' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:88:5: warning: type of 'ff_inlink_consume_frame' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:102:5: warning: type of 'ff_inlink_consume_samples' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:110:5: warning: type of 'ff_inlink_make_frame_writable' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:74:5: warning: type of 'ff_inlink_check_available_samples' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:67:5: warning: type of 'ff_inlink_check_available_frame' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/filters.h:133:5: warning: type of 'ff_inlink_acknowledge_status' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:244:6: warning: type of 'ff_avfilter_link_set_out_status' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:256:6: warning: type of 'ff_tlog_link' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:238:6: warning: type of 'ff_avfilter_link_set_in_status' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:332:5: warning: type of 'ff_request_frame' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:332:5: warning: type of 'ff_request_frame' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:272:5: warning: type of 'ff_insert_pad' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/video.h:39:10: warning: type of 'ff_get_video_buffer' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/audio.h:50:10: warning: type of 'ff_null_get_audio_buffer' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/audio.h:61:10: warning: type of 'ff_get_audio_buffer' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:362:5: warning: type of 'ff_filter_frame' does not match original declaration [-Wlto-type-mismatch] src/libavfilter/internal.h:362:5: warning: type of 'ff_filter_frame' does not match original declaration [-Wlto-type-mismatch]

1 0

Hardware acceleration
by pilot constatinus 20 Apr '18

20 Apr '18

Hi, i was recently trying out vaapi acceleration in ffmpeg. It works well with h264 hevc and vp8, but not with vp9. The output i get is: ffmpeg -vaapi_device /dev/dri/renderD128v -i in.mkv -vf 'format=nv12,hwupload' -c:v vp9_vaapi -b:v 5M out.webm [ vp9_vaapi ] Encoding profile not found (19). Also, how can i encode rgb ( bgrp, bgr0 ) while recording with hardware acceleration ( h264), not losing quality in yuv?

2 1

AAC encoding seems to shorten the file slightly
by Will Munn 20 Apr '18

20 Apr '18

Hi, We have a script that creates AAC versions of WAV samples to be used in a musical context (looped stretched etc). Currently we are using apples encoder (afconvert) to do this on a Mac. We’d like to be able to move to ffmpeg so that we can run this without requiring OSX. I have written a test which does the conversion using ffmpeg instead of afconvert. However I am finding that ffmpeg is truncating the audio slightly. Using the same source wav file for example I ran: ffmpeg -y -I source.wav -b:a 128k ffmpeg.m4a And with afconvert: afconvert source.wav -o afconvert.m4a -q 127 -b 128000 -f m4af -d aac When you compare the 2 audio files with ffprobe for example you can see that the duration is smaller on the ffmpeg encoded file. Are you aware of this and is there a way around it? Thanks for any help/advice you may have! Will ffprobe -i Bip\ late\ bip.wav-afconvert.m4a -hide_banner -show_format Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'Bip late bip.wav-afconvert.m4a': Metadata: major_brand : M4A minor_version : 0 compatible_brands: M4A mp42isom creation_time : 2018-04-19T14:59:57.000000Z iTunSMPB : 00000000 00000840 000002B0 000000000002B110 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 Duration: 00:00:04.06, start: 0.047891, bitrate: 14 kb/s Stream #0:0(eng): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 6 kb/s (default) Metadata: creation_time : 2018-04-19T14:59:57.000000Z [FORMAT] filename=Bip late bip.wav-afconvert.m4a nb_streams=1 nb_programs=0 format_name=mov,mp4,m4a,3gp,3g2,mj2 format_long_name=QuickTime / MOV start_time=0.047891 duration=4.063492 size=7523 bit_rate=14810 probe_score=100 TAG:major_brand=M4A TAG:minor_version=0 TAG:compatible_brands=M4A mp42isom TAG:creation_time=2018-04-19T14:59:57.000000Z TAG:iTunSMPB= 00000000 00000840 000002B0 000000000002B110 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 [/FORMAT] ffprobe -i Bip\ late\ bip.wav.ffmpeg.m4a -hide_banner -show_format Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'Bip late bip.wav.ffmpeg.m4a': Metadata: major_brand : M4A minor_version : 512 compatible_brands: isomiso2 encoder : Lavf57.83.100 Duration: 00:00:04.02, start: 0.000000, bitrate: 12 kb/s Stream #0:0(und): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 9 kb/s (default) Metadata: handler_name : SoundHandler [FORMAT] filename=Bip late bip.wav.ffmpeg.m4a nb_streams=1 nb_programs=0 format_name=mov,mp4,m4a,3gp,3g2,mj2 format_long_name=QuickTime / MOV start_time=0.000000 duration=4.024000 size=6230 bit_rate=12385 probe_score=100 TAG:major_brand=M4A TAG:minor_version=512 TAG:compatible_brands=isomiso2 TAG:encoder=Lavf57.83.100 [/FORMAT]

2 1

What ffmpeg filter can I use to adjust colour temperature?
by Dmitri 20 Apr '18

20 Apr '18

I am using ffmpeg with a camera and need to adjust colour temperature to the lighting, adjust to warm/cold/bright white. I've found an algorithm that calculates temperature in Kelvin to RGB values (http://www.zombieprototypes.com/?p=210) and tried to use colorbalance filter by converting result values to -1 - 1 and adjusting midtones. But it doesn't seem right. Is there any default filter for that or any other that can take values like above? Must be a frequent request for video, but I can't find any examples. / Best regards

2 1

hwaccel cuvid with filters
by Guido Holz 19 Apr '18

19 Apr '18

Hi all, have a question of how to use in a correct way the hwaccel of nvidia. I get the error "cuda is not supported as input pixel format". I didn't find help how to use hwaccel with filters. Maybe someone can explain how to use in filters? Many thanks! the following works: ffmpeg -y -hwaccel cuvid -c:v h264_cuvid -i Input.ts -t 3 -c:v h264_nvenc -profile:v main -level 3.1 -crf 22 output.mp4 doesn't work: ffmpeg -y -hwaccel cuvid -c:v h264_cuvid -i Input.ts -t 3 -filter_complex_script filter.txt c:v h264_nvenc -profile:v main -level 3.1 -crf 22 output.mp4 filter.txt: color=black:960x540[c]; [c]drawtext=fontfile=CUTE.ttf: fontsize=50: text='NAME': fontcolor=white: x=80: y=400, drawtext=fontfile=CUTE.ttf: fontsize=50: text='GO': fontcolor=white: x=120: y=330, split[text][alpha]; [text][alpha]alphamerge[txta]; [txta]perspective= x0=256: y0=306: x1=575: y1=306: x2=0: y2=H: x3=W: y3=H: sense=1[persp]; [persp][0]blend= all_mode='softlight' And the error is on bottom: ffmpeg version 3.2.4-tessus Copyright (c) 2000-2017 the FFmpeg developers built with gcc 5.4.0 (Ubuntu 5.4.0-6ubuntu1~16.04.4) 20160609 configuration: --prefix=/usr/local/ --extra-version=tessus --enable-avisynth --enable-fontconfig --enable-gpl --enable-libass --enable-libfreetype --enable-libgsm --enable-libmodplug --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopus --enable-libschroedinger --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libx264 --enable-libx265 --enable-libxvid --enable-version3 --disable-ffplay --disable-indev=qtkit --disable-indev=x11grab_xcb --enable-cuda --enable-cuvid --enable-nvenc --enable-nonfree --enable-libnpp --extra-cflags=-I/usr/local/cuda/include --extra-ldflags=-L/usr/local/cuda/lib64 libavutil 55. 34.101 / 55. 34.101 libavcodec 57. 64.101 / 57. 64.101 libavformat 57. 56.101 / 57. 56.101 libavdevice 57. 1.100 / 57. 1.100 libavfilter 6. 65.100 / 6. 65.100 libswscale 4. 2.100 / 4. 2.100 libswresample 2. 3.100 / 2. 3.100 libpostproc 54. 1.100 / 54. 1.100 Input #0, mpegts, from 'DEFAULT/PREROLL.ts': Duration: 00:00:03.00, start: 1.480000, bitrate: 1728 kb/s Program 1 Metadata: service_name : Service01 service_provider: FFmpeg Stream #0:0[0x100]: Video: h264 (Main) ([27][0][0][0] / 0x001B), yuv420p(progressive), 960x540 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 90k tbn, 50 tbc Codec AVOption crf (Select the quality for constant quality mode) specified for output file #0 (PREROLL_TMP/GUIDO.ts) has not been used for any stream. The most likely reason is either wrong type (e.g. a video option with no video streams) or that it is a private option of some encoder which was not actually used for any stream. Output #0, mpegts, to 'PREROLL_TMP/GUIDO.ts': Metadata: encoder : Lavf57.56.101 Stream #0:0: Video: h264 (h264_nvenc) (Main), yuv420p, 960x540 [SAR 1:1 DAR 16:9], q=-1--1, 2000 kb/s, 25 fps, 90k tbn, 25 tbc Metadata: encoder : Lavc57.64.101 h264_nvenc Side data: cpb: bitrate max/min/avg: 0/0/2000000 buffer size: 4000000 vbv_delay: -1 Stream mapping: Stream #0:0 (h264_cuvid) -> blend:bottom blend -> Stream #0:0 (h264_nvenc) Press [q] to stop, [?] for help [graph 0 input from stream 0:0 @ 0x31e7b20] Changing frame properties on the fly is not supported by all filters. [swscaler @ 0x3220cc0] cuda is not supported as input pixel format Failed to inject frame into filter network: Invalid argument Conversion failed!

3 2

weird seek behaviour in hls streams
by Daniel Oberhoff 19 Apr '18

19 Apr '18

Hi all, We run a service that extensively relies on hls streams recorded from rtsp capable cameras. We rotuinely generate clips from recorded streams and seem to run into weird behavior of -ss and -t. also it seems the hls stream always has a start-time of 1.4s. This is not the core problem though. when we generate short clips deep in to the stream i.e. -t 1000 -s 10 it often happens we get no output at all. any ideas? Best

2 8

Timestamp and missing frame when converting 4K h.264 to h.265
by Gralfus 19 Apr '18

19 Apr '18

Background of issue: I have a new Panasonic 4K video camera that produces its 4K content in h.264 MP4 format (3840×2160). When I use ffmpeg to convert this to the h.265 codec to shrink the file size, I see a couple of errors about a missing key frame and timestamp, but then the file is generated but with serious visual artifacts. Video from the same camera that is shot natively in just HD (1920x1080 h.264 MP4) quality has no issue when converting via ffmpeg to h.265. Here is the code I used: ffmpeg -i S1000001.MP4 -c:v libx265 -c:a libfdk_aac -crf 28 265-4k.mp4 Here are the errors that I see when I issue the command: [mov,mp4,m4a,3gp,3g2,mj2 @ 0x559875557a40] st: 0 edit list: 1 Missing key frame while searching for timestamp: 3003 [mov,mp4,m4a,3gp,3g2,mj2 @ 0x559875557a40] st: 0 edit list 1 Cannot find an index entry before timestamp: 3003. I've listed the details of the ffmpeg build on my blog: https://wjwoodrow.wordpress.com/2018/04/14/new-camcorder-panasonic-hc-wxf1-… I can list them here if that will help. Ffmpeg works fine for converting other HD video, but not the new 4K frame size. -- Sent from: http://www.ffmpeg-archive.org/

4 7

ffmpeg ignores video filter (-vf:v:0/1), possible bug?
by Jakub Vojáček 19 Apr '18

19 Apr '18

Hello, we are receiving multicast that has 2 audios and video. We'd like to create a new multicast from this one and add one video in different bitrate and resolution. The command we're trying is ffmpeg -i udp://@239.10.10.4:5004?fifo_size=100000000 \ -map i:0x7f9 -c:v:0 h264_nvenc -r:v:0 25 *-vf:v:0 "fps=25,scale_npp=640:360:format=same:interp_algo=lanczos"* -aspect:v:0 16:9 -g:v:0 80 -b:v:0 400K -minrate:v:0 400k -maxrate:v:0 400k \ -map i:0x7f9 -c:v:1 libx264 -r:v:1 25 *-vf:v:1 "fps=25,scale=1280:720"* -aspect:v:1 16:9 -g:v:1 80 -b:v:1 4000K -minrate:v:1 4000k -maxrate:v:1 4000k \ -map i:0x7fc -c:a:0 aac -b:a:0 128k -ar 48k \ -map i:0x7fd -c:a:1 aac -b:a:1 64k -ar 48k \ -f mpegts udp://@239.0.0.123:1234?overrun_nonfatal_option=1 The ffmpeg starts but ignored the video filter options, both outputted videos are in same resolution and bitrate, please see below: Input #0, mpegts, from 'udp://@239.10.10.4:5004?fifo_size=100000000': Duration: N/A, start: 83042.271433, bitrate: N/A Program 10104 Metadata: service_name : CT 24 service_provider: Ceska televize Stream #0:0[0x7f9]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv, top first), 720x576 [SAR 64:45 DAR 16:9], 25 fps, 25 tbr, 90k tbn, 50 tbc Stream #0:1[0x7fa](cze): Subtitle: dvb_teletext ([6][0][0][0] / 0x0006), 492x250 Stream #0:2[0x7fc](cze): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, stereo, s16p, 192 kb/s Stream #0:3[0x7fd](cze): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, mono, s16p, 64 kb/s (visual impaired) Stream #0:4[0x7ff]: Unknown: none ([5][0][0][0] / 0x0005) Stream mapping: Stream #0:0 -> #0:0 (mpeg2video (native) -> h264 (h264_nvenc)) Stream #0:0 -> #0:1 (mpeg2video (native) -> h264 (libx264)) Stream #0:2 -> #0:2 (mp2 (native) -> aac (native)) Stream #0:3 -> #0:3 (mp2 (native) -> aac (native)) Press [q] to stop, [?] for help [libx264 @ 0x55f373354a40] VBV maxrate specified, but no bufsize, ignored [libx264 @ 0x55f373354a40] using SAR=1/1 [libx264 @ 0x55f373354a40] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 AVX2 LZCNT BMI2 [libx264 @ 0x55f373354a40] profile High, level 3.1 Output #0, mpegts, to 'udp://@239.0.0.123:1235?overrun_nonfatal_option=1': Metadata: encoder : Lavf58.9.100 Stream #0:0:* Video: h264 (h264_nvenc) (Main), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], q=-1--1, 400 kb/s, 25 fps, 90k tbn, 25 tbc* Metadata: encoder : Lavc58.10.100 h264_nvenc Side data: cpb: bitrate max/min/avg: 400000/0/400000 buffer size: 800000 vbv_delay: -1 Stream #0:1: *Video: h264 (libx264), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], q=-1--1, 4000 kb/s, 25 fps, 90k tbn, 25 tbc* Metadata: encoder : Lavc58.10.100 libx264 Side data: cpb: bitrate max/min/avg: 4000000/0/4000000 buffer size: 0 vbv_delay: -1 Stream #0:2(cze): Audio: aac (LC), 48000 Hz, stereo, fltp, 128 kb/s Metadata: encoder : Lavc58.10.100 aac Stream #0:3(cze): Audio: aac (LC), 48000 Hz, mono, fltp, 64 kb/s (visual impaired) Metadata: encoder : Lavc58.10.100 aac It seems like the *-vf:v:0 *and *-vf:v:1* are being ignored while all other modifiers (-g:v:0, -b:v:1 -minrate:v:0, ... are working). Is it a bug in ffmpeg or am I doing something wrong? Thank you Jakub

2 1

How to keep/inject proper ambisonic audio metadata using ffmpeg ?
by yael Laouar 19 Apr '18

19 Apr '18

I'm trying to encode a file using ffmpeg but I would like this file to have 360 metadata (audio and video) with this command : ffmpeg -i videoAmb2k_injected.mp4 -c copy -map_metadata 0 -map_metadata:s:v 0:s:v -map_metadata:s:a 0:s:a -b:a 128k testAddMetadata2.mp4 The input file "videoAmb2k_injected.mp4" got 360 metadata thanks to the spatial media metadata injector. But using this command, VLC does not play the file as a 360 video, and the audio isn't spatialize neither. By adding "-strict unofficial" option, I was allowed to keep 360 video metadata between input and output, and the video is played as spherical. However, audio isn't spatialized. How can I inject or keep the right audio metadata so that any video player such as VLC3 interprets the audio as ambisonic ? Thanks :)

1 0

HLS Questions
by Ronak 18 Apr '18

18 Apr '18

Hi there, I'm using ffmpeg to generate MPEG DASH & HLS content using a backing fragmented MP4 asset, and I've noticed some inconsistencies with ffmpeg that I wasn't sure about. For reference, I'm running the following commands: Generate a fMP4 backed HLS stream: ffmpeg -i "${FILE}.mp4" -codec copy -hls_time 0.975238095238095 -hls_segment_type fmp4 -hls_flags single_file+append_list -hls_playlist_type vod "${FILE}_1sec_v7.m3u8" Generate a fMP4 backed MPEG DASH/HLS stream: ffmpeg -i "${FILE}.mp4" -codec copy -f dash -single_file_name "${FILE}_1sec.m4s" -min_seg_duration 975238.095238095 -hls_playlist 1 "${FILE}_1sec.mpd" My questions are as follows: 1. The first command above generates an HLS manifest with version 7. The second command generates an HLS manifest with version 6. Why are they different, when they are using the same exact set of features from the HLS protocol definition? Shouldn't both be HLSv7 (the first version where Apple supports fMP4 backing assets)? 2. Is it possible to use a fMP4 and a SegmentTemplate/SegmentTimeline with byte ranges? A fMP4 has inherent benefits over numerous separate .m4s files, but then we have to pay the price of a really long manifest file. Why can't ffmpeg templatize the byte ranges themselves as well? 3. Is ffmpeg planning on supporting HLSv8 soon? That will allow us to use variable substitution in HLS, just like we can with MPEG DASH. Is there an ETA on when this would be available in ffmpeg's trunk? Thanks, Ronak

1 0