scalable decoding outputs multiple pictures per temporal unit
aomdec/dav1d give different outputs.
bash-4.4$ ~/Projects/aom/x86-64/aomdec ~/Downloads/av1-1-b8-22-svc-L2T1.ivf -o /tmp/b8.aom.yuv --rawvideo --md5
f8305c4b796b620388fee614df3ebac6 /tmp/b8.aom.yuv
bash-4.4$ tools/dav1d -i ~/Downloads/av1-1-b8-22-svc-L2T1.ivf -o - --muxer=md5 --quiet
47df59c53e601d4addb9ac9c5fd6431e
Also note that the resolution of the first frame is not the same:
bash-4.4$ ~/Projects/aom/x86-64/aomdec ~/Downloads/av1-1-b8-22-svc-L2T1.ivf -o /tmp/b8.aom.y4m --limit=1
bash-4.4$ tools/dav1d -i ~/Downloads/av1-1-b8-22-svc-L2T1.ivf -o /tmp/b8.dav1d.y4m --limit=1 --quiet
bash-4.4$ head -n1 /tmp/b8.dav1d.y4m
YUV4MPEG2 W640 H360 C420jpeg Ip F240:8
bash-4.4$ head -n1 /tmp/b8.aom.y4m
YUV4MPEG2 W1280 H720 F30:1 Ip C420jpeg
https://people.xiph.org/~tterribe/av1/samples-all/av1-1-b8-22-svc-L2T1.ivf
https://people.xiph.org/~tterribe/av1/samples-all/av1-1-b8-22-svc-L2T2.ivf
$ tools/dav1d -i ~/Downloads/av1-1-b8-22-svc-L2T1.ivf -o - --muxer=md5 --quiet
47df59c53e601d4addb9ac9c5fd6431e
$ ~/Projects/aom/x86-64/aomdec ~/Downloads/av1-1-b8-22-svc-L2T1.ivf -o - --rawvideo --md5 --all-layers
47df59c53e601d4addb9ac9c5fd6431e -
So it seems that we need to allow selecting a specific operating point (aomdec --oppoint=X
, the first sample here has one 360p and one 720p OP) and then this can be considered done. It's a tool problem, not a decoder problem.