You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Img2img Alternative script is still an important workflow item for many who are performing tasks such as temporal consistent video via img2img. Although newer techniques are in development that perform editing functions in better and more sophisticated ways, I think there is always a benefit to being able to perform accurate image inversion when necessary. Currently img2imgalt script performs poorly with 2.x models and the code is outdated and unmaintained, but still in use by many today.
I would like to consider and open up discussion regarding EDICT (https://arxiv.org/abs/2211.12446) which proposes a method of "determining a mathematically exact inversion of real and model-generated images by maintaining two coupled noise vectors which are used to invert each other in an alternating fashion". I believe maybe this could be used in place of or in conjunction with the current alternative img2img script.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi everyone,
Img2img Alternative script is still an important workflow item for many who are performing tasks such as temporal consistent video via img2img. Although newer techniques are in development that perform editing functions in better and more sophisticated ways, I think there is always a benefit to being able to perform accurate image inversion when necessary. Currently img2imgalt script performs poorly with 2.x models and the code is outdated and unmaintained, but still in use by many today.
I would like to consider and open up discussion regarding EDICT (https://arxiv.org/abs/2211.12446) which proposes a method of "determining a mathematically exact inversion of real and model-generated images by maintaining two coupled noise vectors which are used to invert each other in an alternating fashion". I believe maybe this could be used in place of or in conjunction with the current alternative img2img script.
Thanks!
TwoPerCent
Beta Was this translation helpful? Give feedback.
All reactions