Instructions to use shuttleai/shuttle-3-diffusion with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use shuttleai/shuttle-3-diffusion with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("shuttleai/shuttle-3-diffusion", dtype=torch.bfloat16, device_map="cuda") prompt = "Venus floating market at dawn, fantasy digital art, highly detailed, atmospheric lighting with film-like light leaks, impressive background, studio photo style, cinematic, intricate details." image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
My thoughts and feedback on Shuttle 3 Diffusion
I haven't yet tried Shuttle 3.1. I'm keen to see how strong the prompt following is and how well it handles styles.
My thoughts and feedback on Shuttle Diffusion 3: overall, not a bad model—better than most people think. I'll start with the bad points: images can appear over-contrasted and lack colour range and realism with humans can be worse than Schnell sometimes.
Good points: it's fast, handles styles better than Flux Schnell, and LORAs do work (not as strong as Flux Dev, but they work on all models I've tested and my trained ones). Great tip by lowering the steps to 2 and doubling the resolution, it's possible to produce realistic images close to Dev. My hope is that this can become DreamShaper style model good with all styles.