Dummy codec for benchmarking purposes?

[pbristow]

Has anyone produced a dummy or "null" codec, for benchmarking video tools such as AVIsynth and Virtualdub?

I often find myself wanting to compare the performance of several AVIsynth scripts, and wishing I could eliminate the encoding time of the codec from the process. My usual timing technique is to tell VirtualDub to encode either raw video or "Quality 1" DivX into an AVI file, and then use the "created" and "modified" timestamps of the file to tell me how long it took. However, sometimes the encoding time for DivX is significant compared to the execution time of the AVIsynth script (e.g. fairly simple scripts outputing Full HD video), and sometimes the disk i/o and space requirements of raw video are the problem. (Also DivX has annoying constraints on frame rate and file size - No good when you're trying to benchmark a script that puts out 1080p60!)

[EDIT - I've tried a variety of other codecs that seemed like they would be quicker: Most turn out to have annoying constraints of one kind or another.]

What would be really useful is a "null codec", that basically just discards all the frames that are fed to it except either the first or the last one, writes that one frame into the AVI file exactly as received, and then closes the file. Result: Encoding makes virtually zero contribution to the test time. [EDIT: And theoretically, no constraints at all on what framerates/sizes, colouspaces etc. can be fed to it.]

Has anyone done anything like this, or does any one fancy cobbling one together?

[J_Darnley]

avsutil, ffmpeg's null output, mplayer's benchmark

[Groucho2004]

I wrote a small cli tool for benchmarking purposes. You might want to try it: AVSInfo