How to record a webcam to a file outside of X11?

Question

I'm working with teachers to automatically record their classes, so we can review them and improve the quality of teaching. We have computers running Ubuntu 17.10 with multiple webcams in a couple of classrooms - but I could run other software if it makes this task easier.

I can successfully record a stream from the webcam to an h264 encoded file using gstreamer. The following should work for most people with gstreamer installed, but I've got fancier pipelines using vaapi that can simultaneously encode multiple 4k streams on a NUC with room to spare! My point is that Gstreamer works great when I'm typing at a terminal in the GUI. The example:

.\gst-launch-1.0.exe -e autovideosrc ! videoconvert ! \
  openh264enc max-bitrate=256000 ! h264parse ! \
  mp4mux ! filesink location=somefile.mp4

I imagine I could also do this with ffmpeg, or OpenCV, or maybe even VLC (I can record a webcam via the GUI, so I guess I could use that to generate a command line?).

But when I tried any of the above, for example, via SSH, I get errors from GStreamer and OpenCV, and blank videos from ffmpeg (I haven't tried VLC because I don't currently have access to these machines). I need to automate - but I could potentially leave a user logged in. I just need to have some way to capture webcam to disk with some amount of reasonable compression.

I naively thought I could throw something like the above into a cron job and I'd be good to go (intending to send a SIGINT to end recording). But anything that can be automatically scheduled somehow would be great.

EDIT: Below is an approach I'm trying using ffmpeg. You can see from the output that I can't figure out how to specify pixel_format in a way that ffmpeg pays attention to! First, the command (using mkv because that seems to be a "low-stress" format, but have also tried mov and mp4):

ffmpeg -hwaccel vaapi -vaapi_device /dev/dri/renderD128 \
  -f v4l2 -framerate 30 -video_size hd720 -pixel_format yuv420p -i /dev/video1 output.mkv

Like I said, I'm trying to get hardware acceleration, and you can see below that VAAPI is working (but I think just for decoding). You can easily remove the options from the first line, and I get similar results either way. I didn't include the header with compile options and library versions, as it's standard Ubuntu 17.10.

libva info: VA-API version 0.40.0
libva info: va_getDriverName() returns 0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so
libva info: Found init function __vaDriverInit_0_40
libva info: va_openDriver() returns 0
Input #0, video4linux2,v4l2, from '/dev/video1':
  Duration: N/A, start: 42437.238243, bitrate: 442368 kb/s
    Stream #0:0: Video: rawvideo (YUY2 / 0x32595559), yuyv422, 1280x720, 442368 kb/s, 30 fps, 30 tbr, 1000k tbn, 1000k tbc
File 'output.mkv' already exists. Overwrite ? [y/N] y
Stream mapping:
  Stream #0:0 -> #0:0 (rawvideo (native) -> h264 (libx264))
Press [q] to stop, [?] for help
No pixel format specified, yuv422p for H.264 encoding chosen.
Use -pix_fmt yuv420p for compatibility with outdated media players.
[libx264 @ 0x55d1a26a71a0] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 AVX2 LZCNT BMI2
[libx264 @ 0x55d1a26a71a0] profile High 4:2:2, level 3.1, 4:2:2 8-bit
[libx264 @ 0x55d1a26a71a0] 264 - core 148 r2795 aaa9aa8 - H.264/MPEG-4 AVC codec - Copyleft 2003-2017 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=6 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=25 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=23.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00
Output #0, matroska, to 'output.mkv':
  Metadata:
    encoder         : Lavf57.71.100
    Stream #0:0: Video: h264 (libx264) (H264 / 0x34363248), yuv422p, 1280x720, q=-1--1, 30 fps, 1k tbn, 30 tbc
    Metadata:
      encoder         : Lavc57.89.100 libx264
    Side data:
      cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: -1
Past duration 0.717049 too large
Past duration 0.879128 too large
frame=  567 fps= 16 q=27.0 size=    2156kB time=00:00:34.16 bitrate= 516.9kbits/s speed=0.938x

I exit with ctrl-C. Which results in what appears to be an orderly exit:

[libx264 @ 0x55d1a26a71a0] frame I:11    Avg QP:15.75  size: 18573
[libx264 @ 0x55d1a26a71a0] frame P:2176  Avg QP:19.91  size:  4435
[libx264 @ 0x55d1a26a71a0] frame B:173   Avg QP:20.00  size:  3232
[libx264 @ 0x55d1a26a71a0] consecutive B-frames: 90.1%  0.1%  0.6%  9.2%
[libx264 @ 0x55d1a26a71a0] mb I  I16..4: 34.0% 56.1%  9.8%
[libx264 @ 0x55d1a26a71a0] mb P  I16..4:  0.1%  1.2%  0.0%  P16..4: 32.7%  3.1%  6.1%  0.0%  0.0%    skip:56.8%
[libx264 @ 0x55d1a26a71a0] mb B  I16..4:  0.0%  0.3%  0.0%  B16..8: 31.9%  0.7%  0.1%  direct: 1.4%  skip:65.6%  L0:41.8% L1:57.9% BI: 0.3%
[libx264 @ 0x55d1a26a71a0] 8x8 transform intra:81.2% inter:92.4%
[libx264 @ 0x55d1a26a71a0] coded y,uvDC,uvAC intra: 25.4% 20.1% 2.1% inter: 10.2% 7.4% 0.0%
[libx264 @ 0x55d1a26a71a0] i16 v,h,dc,p: 78% 10%  7%  5%
[libx264 @ 0x55d1a26a71a0] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu:  7%  6% 72%  2%  3%  3%  2%  2%  3%
[libx264 @ 0x55d1a26a71a0] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 34% 21% 25%  4%  5%  3%  4%  1%  3%
[libx264 @ 0x55d1a26a71a0] i8c dc,h,v,p: 69% 14% 15%  2%
[libx264 @ 0x55d1a26a71a0] Weighted P-Frames: Y:1.7% UV:0.1%
[libx264 @ 0x55d1a26a71a0] ref P L0: 49.3%  2.7% 29.6% 18.1%  0.3%
[libx264 @ 0x55d1a26a71a0] ref B L0: 69.3% 24.1%  6.6%
[libx264 @ 0x55d1a26a71a0] ref B L1: 86.6% 13.4%
[libx264 @ 0x55d1a26a71a0] kb/s:529.95
Exiting normally, received signal 2.

Answer 1

Use

ffmpeg -hwaccel vaapi -vaapi_device /dev/dri/renderD128 \
  -f v4l2 -framerate 30 -video_size hd720 -pixel_format yuv420p -i /dev/video1 \
  -c:v libx264 -pix_fmt yuv420p output.mkv

The pixel format for the output should be specified after the inputs and before the output, and it's specified using pix_fmt.