fix noisy images at high step counts #1086
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
At step counts greater than ~75, the ksamplers start producing noisy images when using the Karras noise schedule. This PR reverts to using the model's own noise schedule, which eliminates the problem at the cost of slowing convergence at lower step counts. This PR also introduces a new CLI `--save_intermediates <n>' argument, which will save every nth intermediate image into a subdirectory named `intermediates/<image_prefix>'. Addresses issue #1083.
tildebyte
approved these changes
Oct 13, 2022
|
Good catch lstein. This patch seems worthy of a new release. |
Collaborator
Author
Contributor
|
Thanks for the quick fix! |
|
The actual fix for the noisy image is in: |
Closed
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
At step counts greater than ~75, the ksamplers start producing noisy images when using the Karras noise schedule. This PR reverts to using the model's own noise schedule, which eliminates the problem at the cost of slowing convergence at lower step counts.
This PR also introduces a new CLI
--save_intermediates <n>argument, which will save every nth intermediate image into a subdirectory namedintermediates/<image_prefix>.Addresses issue #1083.