Skip to content

Conversation

ZFTurbo
Copy link

@ZFTurbo ZFTurbo commented Jul 15, 2022

First downsample then upsample back to original shape.

@iver56
Copy link
Owner

iver56 commented Jul 15, 2022

What is the motivation for adding this transform? The effect would be pretty much the same as applying a low-pass filter, right?

@ZFTurbo
Copy link
Author

ZFTurbo commented Jul 15, 2022

I don't know if effect is the same. I was trying to emulate users behaviour during video/audio editing.

@iver56
Copy link
Owner

iver56 commented Jul 15, 2022

I personally just apply a LPF to emulate such behavior. Doing a downsampling (including the usual LPF that is involved in that process) and then an upsampling requires more compute, but gives a similar result

@ZFTurbo
Copy link
Author

ZFTurbo commented Jul 16, 2022

May be we can add it in repo but add your comment that it's possible to replace with LPF? It's for the persons who will seek for the same augmentation. I don't see a problem to have it in repo.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants