Stable Diffusion can do that for years, ChatGPT already shows impressive results in one free try. Please pretty please. And a lot more consistency with reference-based image generation couldn't hurt.
#T-pose in image generation
tacit pelican
olive elbow
at the moment Mistral is using Flux, so its a seperate call to create an image and that is then sent to the ui
iirc
@cursive crane
cursive crane
at the moment Mistral is using Flux, so its a seperate call to create an image a...
Yes, the image generation model is not the same one powering the chat
tacit pelican
i honestly don't know what that means, how come the image generation is not the same one powering the chat ? is there a way for me to improve image generation then ? by getting directly to mistral's flux ?
i'm using chat pro btw
cursive crane
i honestly don't know what that means, how come the image generation is not the ...
Oh sorry if it wasnt clear, basically there is a vision llm for vision and text tasks, and a different flux model that powers image generation
tacit pelican
Don't be sorry, i'm pretty new to all this. Is there a way to feed the flux model with a controlnet like openpose to guide character posing generation ?