Stable Diffusion 3.5 Large and Turbo - int8-convrot (ComfyUI)
Int8-quantized Stable Diffusion 3.5 Large-Turbo, and the T5-XXL text encoder in ComfyUI's native int8-convrot format.
2× smaller, much faster, and still generates stunning images (output 0.9999 correlated with the fp16 originals).
| File | Link |
|---|---|
| sd3.5_large_int8_convrot.safetensors | Download |
| sd3.5_large_turbo_int8_convrot.safetensors | Download |
| text_encoders/t5xxl_fp16_int8_convrot.safetensors | Download |
| clip_g.safetensors | Download |
| clip_l.safetensors | Download |
| VAE (diffusion_pytorch_model.safetensors) | Download |
Settings:
- Sampler: dpmpp_2m
- Scheduler: sgm_uniform
- CFG: 4.5 / turbo: 1
- Steps: 40 / turbo: 8
- Downloads last month
- 55
Model tree for rzgar/stable-diffusion-3.5-large-int8-convrot
Base model
stabilityai/stable-diffusion-3.5-large