Not seeing any improvement to quality

#1
by nitefallen - opened

Are you sure this Lora is properly trained? Don't see any improvements whatsoever to Flux dev using the same fixed seed.
FYI result from running:
lora key not loaded: diffusion_model.double_blocks.16.img_attn.qkv.lora_down.weight
etc etc

Like I said in the readme this is a converted version of: https://huggingface.co/XLabs-AI/flux-RealismLora

I didn't train it.

How to fix this problem lora key not loaded??

Update ComfyUI

Update ComfyUI

Oh I did not see the announcements I just updated 20 mins ago and it did work. Thaanks.

Hwo do we run the convert .py on other loras? is there are tutorial anywhere?

Without and with comparison https://imgur.com/a/4d1t1ck

This comment has been hidden

Without and with comparison https://imgur.com/a/4d1t1ck

I also noticed that issue in FLux, many output look like cartoon, even if I said realistic in prompts!

how do u use it?

Sign up or log in to comment