Expected performance: 1080p @ 0.8-1.2 fps; 720p @ 2-3 fps. A 10-minute video encodes in ~5 hours. The output is indistinguishable from x264 at half the bitrate.
ffmpeg -i input.mkv -c:v libvpx-vp9 -b:v 5M -speed 1 -pass 2
-auto-alt-ref 1 -lag-in-frames 25 -tile-columns 4 -row-mt 1
-c:a libopus -b:a 128k output.webm
Vera S05 Note: Two-pass yields better bitrate distribution but is ~2x slower. Use for constrained bandwidth scenarios. vera s05 libvpx best
ffmpeg -i input.mkv -c:v libvpx-vp9 \
-b:v 0 -crf 18 -row-mt 1 \
-tile-columns 4 -frame-parallel 1 \
-auto-alt-ref 1 -lag-in-frames 25 \
-speed 2 -quality good \
-c:a libopus -b:a 128k \
output.webm
ffmpeg -i input -c:v libvpx-vp9 -b:v 1500k -threads 2 -cpu-used 5 -deadline realtime \
--quality good -g 60 -row-mt 1 -aq-mode 2 -f webm output.webm
(Replace --quality/flags with ffmpeg equivalents as needed.)
| Speed | Visual Quality | Use Case | |-------|----------------|-----------| | 0 | Near-lossless | Archival master | | 1 | Excellent | High-end VOD (Vera S05) | | 2 | Very good | Daily VOD encodes | | 3-4 | Good | Fast-turnaround content | | 5+ | Fair | Live or preview encodes |
For “best,” never exceed
-speed 2in VP9. Expected performance: 1080p @ 0
For older Vera S05 kernels, libvpx (VP8) is significantly faster. If VP9 is too heavy, fall back to VP8 with:
-c:v libvpx -cpu-used 4 -crf 10 -b:v 1M
VP8 at crf 10 looks excellent and runs 3x faster than VP9 on the same chip.
The "best" Libvpx setting on the Vera S05 is not -best. It is a pragmatic mix of -cpu-used 2, -row-mt 1, and accepting VP9 over VP8 only when you have patience. Vera S05 Note: Two-pass yields better bitrate distribution
Verdict: For live encoding, stick to VP8 at -cpu-used 5. For background jobs, use VP9 at -cpu-used 2. You will get broadcast-ready WebM files without melting your little S05 board.
Have you found a better -cpu-used value for the S05? Let us know in the comments.
For VLC:
For Kodi:
For MX Player: