I've always liked Flux even though it is heavily and unnecessarily censored. Chroma is basically more what Flux should have been from the beginning. It is based on Flux Schnell and has all of the benefits of Flux without most of the Flux censorship.
GonzaLomo Chroma is a blend of Chroma 1 HD with a number of LoRAs, including some of my own. The primary goal was to try to retain Chroma's creativity while making it more consistent and stable.
The images I post with this model are created using my workflow which generates the image using GonzaLomo Chroma and then does a light refinement using GonzaLomo v6.0 Photo XL. The amount of refinement is configurable by changing the denoise value on the refiner's sampler.
Description
This is a Q4 quantized version of the V2.0 UNET.
FAQ
Comments (14)
Hello—first of all, thank you so much for all the effort. Your checkpoints are fabulous.
I have a question:
I’m using the workflow and adding the LoRAs for a character, but when I start the refiner, the facial features are lost. I’ve played around with the denoise settings, but I haven’t found a solution that fixes this. So I suspect that general image enhancement without affecting the face might not be compatible (maybe inpainting?). I can’t think of any other solutions…
Best regards, and Happy New Year!
i still like use chroma, shcnell to create image with my old trained loras, then use other model to refine. keep made a great work sir cheer!
It's got a bit of same face. Really like it with the lenovo lora and
"sampler_name": "dpmpp_2m_alt",
"scheduler": "bong_tangent",
I just want to thank you for this post, I've been messing with Chroma for a week trying all kinds of sampler and scheduler combos and other things trying to get consistently good images with little to no luck. and this combo of dpmpp_2m_alt and bong_tangent has finally done it for me.
This checkpoint has a great aesthetic, mostly realistic with a painterly flair. Also it has very minimal hallucinations. A great model!
Thanks so much!
havent tried but looking at examples it feels that you turned fluxface back xD
Not sure what you're saying
@GBRX bro look at this iconic flux chin
ye.. this checkpoint has flux chin which is that cleft chin common to flux models. also stuck with fp8 and have to upcast to try to make int8 quant :( Good model though.
hhm, seems like the 2.0 version covers the pictures with small "blocky artifacts" on close examination - even here on the posted examples vissible, if you do a close look, 1.2 examples seem to be fine, will try this, after I spent an hour on troubleshoot my own worlflow/setup until I noticed.
Thanks for the feedback. I'll take a look.
Hello! May be you can make nvfp4 version?
Maybe, I'd have to look into how to do that first ;)