Instructions to use tensorart/stable-diffusion-3.5-medium-turbo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use tensorart/stable-diffusion-3.5-medium-turbo with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("tensorart/stable-diffusion-3.5-medium-turbo", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Possible 1-step version
Hello! I use a NVIDIA GeForce GTX 1080 as my GPU, and while I do think that the 4-step and 8-step distills here are quite fast (especially if you set ComfyUI to preview them during generation; it makes the waiting time MUCHHH more bearable), they still take 7s and 15s (using the Turbo LoRA on base 3.5 Medium quantized at Q4_K_S) to generate respectively, and I'm interested in seeing faster results.
Do you know if it's possible to create a 1-step version of 3.5 Medium Turbo? Even a 2-step version would work fine too. Preferably, you could do both! I'm asking this because I've used Stable Diffusion XL Turbo before, and I was quite impressed by its generation quality when I first used it (even to this day, I still am), so I'd be happy to see a model that produces even better-quality images, while still being fast.
Thank you for your time.