pycvvdp: Sort the output for wildcharchars #9
Conversation
This allows us to have a CSV file (can be extended to JSON/XML) in
future for quality metrics computed by the program.
By default it saves to $output/${basename}-metrics.csv format.
This is enabled by default because why not
This is problematic if we do not sort when we feed two folders. Python do not care frame orders, but we care, so in some cases it was comparing Frame 2 with Frame 90. So to be correct, we sort it before returing the list.
mantiuk
left a comment
mantiuk
left a comment
There was a problem hiding this comment.
Thank you for contributing this pull request.
The problem is clear but the solution is not ideal. It introduced a dependency on pandas, which we would rather avoid. It also adds duplicate functionality with the --save argument.
The new code now sorts the file names after the wildcard expansions. If this does not solve the problem, please use the interactive mode:
https://github.com/gfxdisp/ColorVideoVDP?tab=readme-ov-file#interactive-mode
|
What do you mean duplicate functionality of save? I think interactive mode is orthogonal to this PR. We can always kill pandas if we need, it will be useful if you prefer to match like how modern tools work which supports CSV/XML/JSON and others. So it can be easy adjusted, while if we use classic CSV/filebased IO, it will be harder to support more output formats in future, |
|
Yes, sorry, I missed f081427 We will look into how to unify the interface with VMAF at some point. But this will be a separate PR. Right now, you can use |
|
b854cfc addresses this |
2 participants
This is problematic if we do not sort when we feed two folders.
Python does not care about frame orders, but we care, so in some cases,
it was comparing Frame 2 with Frame 90. So to be correct, we
sort it before returning the list.
PS: Will rebase once #7 is merged