Instructions to use Tongyi-MAI/Z-Image-Turbo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Tongyi-MAI/Z-Image-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("Tongyi-MAI/Z-Image-Turbo", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
Will ControlNet be supported in the future?
I must admit that your model is incredible in every way. I really liked the level of realism; it's really magnificent. It would be a significant loss for me if there were no ControlNet support, such as instantX and Diffsynth. Please end this joy with controlNet support.
good luck with the rest, which would be really heavy
I know it was a stupid thing to do about duplicating the post, but I was really excited about ControlNet. Honestly, all kinds of editing models fail when it comes to the type of image I'm working with. But to be fair, the strongest ControlNet module I'm currently using is instantX for qwen models
After trying Z Image, I was shocked by the image quality and the result—more than I expected. I swear to you, if a dedicated ControlNet module were created for Z Image, it would be a true revolution in my work.
I really hope so, z-image has potential to be a proper sdxl replacement, loras seem simple enough to train once the base model gets released, I imagine full finetunes will follow. No controlnet would be a massive disappointment.