Are there still inpaint model plans?
I remember that there was a mention in the old Chroma repo discussion about the plan for Chroma's inpaint version.
Will there be a Chroma inpaint as well?
This model can be used to inpaint with LANPaint.
my current plan is to train pixel space model atm, then swap the text encoder with qwenvl2.5 before jumping to inpainting model
https://www.reddit.com/r/StableDiffusion/comments/1mxwr4e/update_chroma_project_training_is_finished_the/
you can try the radiance checkpoint here, but you need a custom comfy ui fork to run it atm
https://huggingface.co/lodestones/chroma-debug-development-only/tree/main/radiance
https://github.com/blepping/ComfyUI/tree/feat_support_chroma_radiance
my current plan is to train pixel space model atm, then swap the text encoder with qwenvl2.5 before jumping to inpainting model
https://www.reddit.com/r/StableDiffusion/comments/1mxwr4e/update_chroma_project_training_is_finished_the/you can try the radiance checkpoint here, but you need a custom comfy ui fork to run it atm
https://huggingface.co/lodestones/chroma-debug-development-only/tree/main/radiance
https://github.com/blepping/ComfyUI/tree/feat_support_chroma_radiance
Would you be full fine-tuning Qwen2.5VL to support general uncensored content?