IP Adapter for composition #7419
asomoza
started this conversation in
Show and tell
Replies: 1 comment 14 replies
-
Beta Was this translation helpful? Give feedback.
14 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi, there's a new IP Adapter that was trained by @jaretburkett to just grab the composition of the image.
Here's the release tweet for SD 1.5 and for SDXL
I think it works good when the model you're using understand the concepts of the source image. Also the scale and the CFG play an important role in the quality of the generation.
Here's my tests with it:
Here's the code I used if someone wants to test it too, even better if you can share your results and findings here too.
Pictures are either generated by me or grabbed from https://www.publicdomainpictures.net/
Beta Was this translation helpful? Give feedback.
All reactions