Please see our Quickstart Guide to Stable Diffusion 3.5 for all the latest info!
This is a smaller checkpoint for SD3.5 Large that contains the text encoders/CLIP built in.
Read our blog for instructions on how to use SD3.5 with ComfyUI.
https://blog.comfy.org/sd3-5-comfyui/
Originally posted https://huggingface.co/Comfy-Org/stable-diffusion-3.5-fp8
Description
FAQ
Comments (7)
It's incredible how fast the community has started to work on SD 3.5, having a quantized version already is awesome!
Gad Dammm. Yessss. Thank a lot for your Model
Okay, this checkpoint says it has clip. I am taking it that means that we can skip the installing the clip on the ComfyUI Blog instructions.
Would be awesome to have some generations showing the difference between this and the origjnal
Confusing model names just like Flux.
What is this ? which one is this ?
It crashes on 8GB vram after ~4minutes trying to load the checkpoint.
There is another fp8 on hugging face that has "Scaled" in it's name, didn't try that one but it's very similar in size and name to this one.
if only Flux could have the stuff this model has... love the styles, with some effort it will become SD1.5 of today.
It is censored AF, not even doing dress slits.
Probably was trained on Mormon fashion catalogue.
Hands look like mutated monster paws.
14Gb of data - and hands even worse than SD1.5 What a pathetic comedy.
https://prnt.sc/3dLM6KUKs_Bb
https://prnt.sc/pYwWyc0OXrMF
Just rename it to monster generator or something.
Details
Files
comfyorgStableDiffusion_v10.safetensors
Mirrors
comfyorgStableDiffusion_v10.safetensors
comfyorgStableDiffusion_v10.safetensors
sd3.5_large_fp8_scaled.safetensors
sd3.5_large_fp8_scaled.safetensors
sd3.5_large_fp8_scaled.safetensors
sd3.5_large_fp8_scaled.safetensors
Imagine-Force_v3.safetensors
sd3.5_large_fp8_scaled.safetensors
sd3.5_large_fp8_scaled.safetensors