[FFmpeg-user] HW Acceleration 101? 2-Up Streaming from RTSP-->ffmpeg-->YouTube

Steven Kan steven at kan.org
Tue Nov 9 01:07:03 EET 2021


Hmm. I hadn’t considered that. What I’m actually doing is combining two 1920  x 1080 streams into a 3840 x 1080 stream. Would an NVidia 1030 be able to do that, twice?

Back on my Mac mini, I’m still learning how to manage services in macOS vs. Raspbian, but I have the stream running right now:

https://www.youtube.com/channel/UCcIZVSZfzrxS6ynL8DQeC-Q/live

My largest display is only 1920 x 1200. Can anyone with a 4K display tell me if YT is presenting this as a 3840-wide stream?

Thanks!

p.s. re: my previous misgivings about a reduction to “only 30% CPU’ was totally misguided. That’s 30% of one core, and the Mini has 8 cores, so the system is very lightly loaded. GUI-level Activity Monitor shows Idle at ~92%. I should be able to run many instances of ffmpeg on here, if I can get them all running properly. 

> On Nov 8, 2021, at 9:43 AM, andrei ka <andrei.k.gml at gmail.com> wrote:
> 
> you could simply plug a recent low profile nvidia (e.g. 1030) into pcie
> slot of your hpe micro and nvenc would do 2 fhd h264 encodes like a charm
> &rei
> 
> On Sun, Nov 7, 2021 at 11:40 PM Steven Kan <steven at kan.org> wrote:
> 
>>> On Jan 18, 2021, at 10:42 PM, Carl Eugen Hoyos <ceffmpeg at gmail.com>
>> wrote:
>>> 
>>> Am Mo., 18. Jan. 2021 um 23:34 Uhr schrieb Steven Kan <steven at kan.org>:
>>>> 
>>>>> On Jan 18, 2021, at 12:50 PM, Michael Koch <
>> astroelectronic at t-online.de> wrote:
>>> 
>>>>>> C:\Program Files\ffmpeg\bin> .\ffmpeg.exe -re -thread_queue_size 1024
>> -i rtsp://anonymous:password@192.168.1.47:554 -i rtsp://
>> anonymous:password at 192.168.1.50:554 -vcodec h264_amf -acodec copy -t
>> 01:47:02 -filter_complex hstack=inputs=2 -f flv out.flv
>>>>>> 
>>>>>> [snip
>>>> 
>>>>>> Input #0, rtsp, from 'rtsp://anonymous:password@192.168.1.47:554':
>>>>>> Metadata:
>>>>>>  title           : Media Server
>>>>>> Duration: N/A, start: 0.080000, bitrate: N/A
>>>>>>  Stream #0:0: Video: h264 (High), yuvj420p(pc, bt709, progressive),
>> 1920x1080, 25 fps, 25 tbr, 90k tbn, 180k tbc
>>>>>> Input #1, rtsp, from 'rtsp://anonymous:password@192.168.1.50:554':
>>>>>> Metadata:
>>>>>>  title           : Media Server
>>>>>> Duration: N/A, start: 0.100000, bitrate: N/A
>>>>>>  Stream #1:0: Video: h264 (Main), yuv420p(progressive), 1920x1080,
>> 100 tbr, 90k tbn, 180k tbc
>>>>> 
>>>>> I see that the two streams have different pixel formats yuvj420p and
>> yuv420p. You could try to bring them to the same pixel format before using
>> hstack.
>>>>> [0]format=yuv420p[a];[a][1]hstack
>>>>> 
>>>>> It's only a wild guess, I'm not sure.
>>>> 
>>>> Do I put this into the filter_complex argument, e.g. -filter_complex
>> "[0]format=yuv420p[a];[a][1] hstack=inputs=2”
>>>> 
>>>> That still results in the "Conversion failed!” error.
>>> 
>>> There is a "scale" missing behind format iirc but for performance
>> reasons you
>>> want to overwrite the pix_fmt instead, not sure if this is possible.
>> 
>> Some progress, here. I’m now attempting this on an M1-powered Mac mini:
>> 
>> ./ffmpeg -thread_queue_size 1024 -hwaccel videotoolbox -i rtsp://
>> anonymous:password at 192.168.1.47:554 -hwaccel videotoolbox -i rtsp://
>> anonymous:password at 192.168.1.50:554 -vcodec h264_videotoolbox -acodec
>> copy -t 01:00:00 -filter_complex hstack=inputs=2 -f flv "rtmp://
>> a.rtmp.youtube.com/live2/<my-youtube-streaming-key”
>> 
>> which results in:
>> 
>> ffmpeg version 4.4 Copyright (c) 2000-2021 the FFmpeg developers
>>  built with Apple clang version 12.0.0 (clang-1200.0.32.27)
>>  configuration: --prefix=/Volumes/tempdisk/sw
>> --extra-cflags=-fno-stack-check --arch=arm64 --cc=/usr/bin/clang
>> --enable-gpl --enable-videotoolbox --enable-libopenjpeg --enable-libopus
>> --enable-libmp3lame --enable-libx264 --enable-libx265 --enable-libvpx
>> --enable-libwebp --enable-libass --enable-libfreetype --enable-libtheora
>> --enable-libvorbis --enable-libsnappy --enable-libaom --enable-libvidstab
>> --enable-libzimg --enable-libsvtav1 --enable-version3
>> --pkg-config-flags=--static --disable-ffplay --enable-postproc
>> --enable-nonfree --enable-neon --enable-runtime-cpudetect
>> --disable-indev=qtkit --disable-indev=x11grab_xcb
>>  libavutil      56. 70.100 / 56. 70.100
>>  libavcodec     58.134.100 / 58.134.100
>>  libavformat    58. 76.100 / 58. 76.100
>>  libavdevice    58. 13.100 / 58. 13.100
>>  libavfilter     7.110.100 /  7.110.100
>>  libswscale      5.  9.100 /  5.  9.100
>>  libswresample   3.  9.100 /  3.  9.100
>>  libpostproc    55.  9.100 / 55.  9.100
>> Input #0, rtsp, from 'rtsp://anonymous:password@192.168.1.47:554':
>>  Metadata:
>>    title           : Media Server
>>  Duration: N/A, start: 0.080000, bitrate: N/A
>>  Stream #0:0: Video: h264 (High), yuvj420p(pc, bt709, progressive),
>> 1920x1080, 25 fps, 25 tbr, 90k tbn, 180k tbc
>> Input #1, rtsp, from 'rtsp://anonymous:password@192.168.1.50:554':
>>  Metadata:
>>    title           : Media Server
>>  Duration: N/A, start: 0.060000, bitrate: N/A
>>  Stream #1:0: Video: h264 (Main), yuv420p(progressive), 1920x1080, 100
>> tbr, 90k tbn, 180k tbc
>>  Stream #1:1: Audio: aac (LC), 8000 Hz, mono, fltp
>> Stream mapping:
>>  Stream #0:0 (h264) -> hstack:input0
>>  Stream #1:0 (h264) -> hstack:input1
>>  hstack -> Stream #0:0 (h264_videotoolbox)
>>  Stream #1:1 -> #0:1 (copy)
>> Press [q] to stop, [?] for help
>> [rtsp @ 0x14a00ca00] max delay reached. need to consume packet
>> [rtsp @ 0x14a00ca00] RTP: missed 146 packets
>> [rtsp @ 0x139013a00] Thread message queue blocking; consider raising the
>> thread_queue_size option (current value: 8)
>> [rtsp @ 0x14a00ca00] max delay reached. need to consume packet
>> [rtsp @ 0x14a00ca00] RTP: missed 80 packets
>> Output #0, flv, to 'rtmp://
>> a.rtmp.youtube.com/live2/q9rg-sqaq-f0mg-yj1c-e42f':
>>  Metadata:
>>    title           : Media Server
>>    encoder         : Lavf58.76.100
>>  Stream #0:0: Video: h264 ([7][0][0][0] / 0x0007), nv12(progressive),
>> 3840x1080, q=2-31, 200 kb/s, 1k tbn (default)
>>    Metadata:
>>      encoder         : Lavc58.134.100 h264_videotoolbox
>>  Stream #0:1: Audio: aac (LC) ([10][0][0][0] / 0x000A), 8000 Hz, mono,
>> fltp
>> [h264_videotoolbox @ 0x14b01b800] Color range not set for nv12. Using MPEG
>> range.
>> [rtsp @ 0x139013a00] max delay reached. need to consume
>> packettrate=6235.2kbits/s dup=0 drop=1 speed=0.794x
>> [rtsp @ 0x139013a00] RTP: missed 29 packets
>> [rtsp @ 0x139013a00] max delay reached. need to consume packet
>> [rtsp @ 0x139013a00] RTP: missed 78 packets
>> [rtsp @ 0x139013a00] max delay reached. need to consume packet
>> [rtsp @ 0x139013a00] RTP: missed 132 packets
>> [rtsp @ 0x139013a00] max delay reached. need to consume packet
>> [rtsp @ 0x139013a00] RTP: missed 49 packets
>> [rtsp @ 0x139013a00] max delay reached. need to consume packet
>> [rtsp @ 0x139013a00] RTP: missed 12 packets
>> [h264 @ 0x139074600] hardware accelerator failed to decode picture
>> [h264 @ 0x139074c00] hardware accelerator failed to decode picture
>> [h264 @ 0x139075200] hardware accelerator failed to decode picture
>> [h264 @ 0x139070a00] hardware accelerator failed to decode picture
>> [h264 @ 0x139072800] hardware accelerator failed to decode picture
>> [h264 @ 0x139072e00] hardware accelerator failed to decode picture
>> [h264 @ 0x139073400] hardware accelerator failed to decode picture
>> [h264 @ 0x139073a00] hardware accelerator failed to decode picture
>> Error while decoding stream #1:0: Unknown error occurred
>> [h264 @ 0x139074000] hardware accelerator failed to decode picture
>> Error while decoding stream #1:0: Unknown error occurred
>> 
>> After a few seconds those errors go away, and then the status line just
>> reads:
>> 
>> frame=20249 fps= 55 q=-0.0 size=   67873kB time=00:06:12.99
>> bitrate=1490.7kbits/s dup=0 drop=2 speed=   1x
>> 
>> I don’t know why the errors occur, nor why they resolve themselves, but
>> eventually it works. ¯\_(ツ)_/¯
>> 
>> top shows ffmpeg consuming about ~30% of the CPU, which is more than I had
>> hoped for, but far better than I’d seen on my previous i5- and AMD-powered
>> boxes.
>> 
>> So I should be able to stream 4 cameras to 2 YT channels without
>> completely cooking my mini.
>> 
>> I tried using hevc_videotoolbox, but I get the error, "Video codec hevc
>> not compatible with flv"
>> 
>> I am hopeful that further optimization of ffmpeg for AS is coming!
>> _______________________________________________
>> ffmpeg-user mailing list
>> ffmpeg-user at ffmpeg.org
>> https://ffmpeg.org/mailman/listinfo/ffmpeg-user
>> 
>> To unsubscribe, visit link above, or email
>> ffmpeg-user-request at ffmpeg.org with subject "unsubscribe".
>> 
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-user
> 
> To unsubscribe, visit link above, or email
> ffmpeg-user-request at ffmpeg.org with subject "unsubscribe".



More information about the ffmpeg-user mailing list