I've made this model using the https://github.com/lum3on/ComfyUI-ModelQuantizer nodes and the full V48 version, so it could run better in my RX6800.
I know that it is also needed if you want to run TorchCompile in RTX 3000 series, so here you have it.
Description
Quantizer from the model provided by Phr00t in huggingface - https://huggingface.co/Phr00t/Chroma-Rapid-AIO
Use a CFG of 1, 8 steps and sa_solver sampler with the beta scheduler (V3). However, the "ipndm" sampler gives good results too, particularly more realistic ones.
FAQ
Comments (5)
guys, can somebody please help me: how to install it and get good results with Forge. I tried it previosly, but every setting that i did doesnt help me to get good images, whats wrong? And i have problem with Cuda's or something with my 5080, what should i do?
I do not use Forge, so I'm not able to help you.
Have you tried first any official forge guide or something? it could be something with chroma itself