ttj

/

flex-diffusion-2-1

like 26

Hello Jonathan, I was greatly appreciated of all the info you shared on your model and how you created it. I followed some of the suggestions you had and continued training 2.1-v as a low-noise model with high quality, high resolution outputs.

My base resolution is 1024x1024, and it supports up to 1536x1024 (and 1024x1536) landscape and portrait modes.

https://huggingface.co/ptx0/pseudo-flex-base

Again, thanks for the great starting point.