If the process fails in step 3, subsequent executions need to re-do steps 1 and 2, which can be slow.
I propose the following:
add a step 0, which creates a folder within clips with the name of the input file
if the folder already exists, skip to step 3*
write all of the chunk folders (clip_NN) into that source-file-specific folder
...this will make it faster to re-process a source video (maybe because step 3 didn't work, or maybe because they want to try a different model, or maybe who knows why)
*let's assume for now that if step 1 worked, so did step 2; but we should revisit that
The main autologger process involves three steps:
If the process fails in step 3, subsequent executions need to re-do steps 1 and 2, which can be slow.
I propose the following:
clips
with the name of the input fileclip_NN
) into that source-file-specific folder...this will make it faster to re-process a source video (maybe because step 3 didn't work, or maybe because they want to try a different model, or maybe who knows why)
*let's assume for now that if step 1 worked, so did step 2; but we should revisit that