[FFmpeg-devel] evaluating the experimental status of ffv1 version 3

[David Rice]

Hi all,

On Oct 4, 2012, at 11:04 AM, Peter B. wrote:

> I've now converted the .yuv10 images sequences to video files and verified their checksum for yuv422p10le in order to be sure that that conversion process was lossless.
> 
> Now I'm running the FFv1 testsuite on them again, so we'll get the results for YUV422 10bit sources.
> 
> Unfortunately, I won't be able to post the results before next week, as I'm away from my setup for almost one week :(

I have some more time available to allocate to testing and review of experimental ffv1 version 3 and wanted to check in on the current state of the evaluations. I see that Peter's test scripts detected some hard to find errors and that the git commit log has been active for ffv1. I wanted to check in to determine where the priorities still exist for testing and review.

Peter's test scripts provide a lot of stats on file size and encoding speed for different combinations of coder, context, gop size, slicecrc use, and other options. I could send this data through gnuplot to provide clearer summaries of the hundreds of lines of test results, especially gop size, slice count, and compression rates.

Dave Rice