Pipeline | Renderstats
Yesterday I was asked by a friend if I had a way to collect a bunch of statistics about a folder of rendered images and warn about missing frames. I didn’t have anything at hand and so I quickly threw something together. As these things go, it quickly went through a couple of revisions until we were both happy with the output it provided.
I think some of you might find value in it as well. So without further ado I present to you the renderstats script.
Functionality is simple. You give it a folder or a file path and a bunch of options and it spits out some stats about the folder contents like so:
renderstats.py /Volumes/ProjectsRaid/renders/
Listing render stats for "/Volumes/ProjectsRaid/renders/":
Overall rendertime: 27 hours 58 minutes 38.07 seconds for 405 files
That's an mean/average of 4 minutes 8.69 seconds per frame.
Maximum per frame time was: 5 minutes 31.00 seconds
Minimum per frame time was: 4 minutes 42.00 seconds
Found the following continuous frame ranges:
1405-1750
2409
Expected a continuous range from 1405 to 2409. Missing 658 frames:
1751-2408
Some files (1) are smaller then the threshold and are likely broken or incomplete:
2409Options are:
-hshow the help message-rGo through all subfolders recursively.-fOutput statistics to file instead of stdout.-nNAME Name to be used for the renderstats file. (default: renderstats.txt)-sSIZE File size below which the file is considered broken. (default: 128 bytes)-tTHRESHOLD Size deviation from the mean before file is considered broken. (default: 0.5)
And that’s pretty much it. It pretty simple under the hood and does not care much where you point it to. So you could point it to a folder full of text file or LXO files instead and get some stats back as well although the missing frames very likely won’t work as that looks for a contiguous frame counter (it will recognize almost any naming scheme that ends with a frame counter at the end followed by the extension).
You can grab the script here: → renderstats.py
