[New Pipeline] InstaFlow! One-Step Stable Diffusion with Rectified Flow

[ayushtues]

Model/Pipeline/Scheduler description

InstaFlow is an ultra-fast, one-step image generator that achieves image quality close to Stable Diffusion, significantly reducing the demand of computational resources. This efficiency is made possible through a recent Rectified Flow technique, which trains probability flows with straight trajectories, hence inherently requiring only a single step for fast inference.

InstaFlow has several advantages:

Ultra-Fast Inference: InstaFlow models are one-step generators, which directly map noises to images and avoid multi-step sampling of diffusion models. On our machine with A100 GPU, the inference time is around 0.1 second, saving ~90% of the inference time compared to the original Stable Diffusion.
High-Quality: InstaFlow generates images with intricate details like Stable Diffusion, and have similar FID on MS COCO 2014 as state-of-the-art text-to-image GANs, like StyleGAN-T.
Simple and Efficient Training: The training process of InstaFlow merely involves supervised training. Leveraging pre-trained Stable Diffusion, it only takes 199 A100 GPU days to get InstaFlow-0.9B.

Open source status

• The model implementation is available.
• The model weights are available (Only relevant if addition is not a scheduler).

Provide useful links for the implementation

Code & Weights - https://github.com/gnobitab/InstaFlow
HF Space - https://huggingface.co/spaces/XCLiu/InstaFlow

Might be interesting to add it as a community pipeline atleast

[ayushtues]

Was waiting for this to get open-sourced for a while now, happy to help in adding it if it seems interesting, @sayakpaul @patrickvonplaten

[sayakpaul]

How does the quality and speed fair to SDXL Turbo or LCM LoRA?

[ayushtues]

[ayushtues]

It claims to take 0.1 sec to generate a 512x512 image

[sayakpaul]

That's hard to believe. With a couple of optimizations, SDXL Turbo could likely achieve this feat. I would add it as a community pipeline to start off.

[ayushtues]

SDXL turbo website says 0.234 seconds

[ayushtues]

Okay, will try adding it as a community pipeline for now then, thanks!

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.