Thanks!
In the meantime, I pushed a change to git that should fix the log path issue, so it should just work on Linux now.
Thanks!
In the meantime, I pushed a change to git that should fix the log path issue, so it should just work on Linux now.
Thank you, I will try this new version. Meanwhile I can prevent the broken pipes by reducing the Batch size to 1000 or 800 depending on video resolution. Keeping the %MEM for all four processes around 40%, and not running any other heavy programs does the trick.
After everyone has tested this program, whats the general consensus on what type of videos it works best with?
Were you able to test it out, how did it go? I tried a version a while back the results weren’t so great I’m curious on this new version. What was your opinion of it?
I think it has great potential. As mentioned by @FunscriptAI, the estimation of the center of motion is where the biggest improvement can be made. I modified the script to read the center from a file, which is generated from the video using machine learning.