HybridStitch: Pixel and Timestep Level Model Stitching for Diffusion Acceleration

TL;DR

HybridStitch introduces a novel diffusion model acceleration method that dynamically combines large and small models at pixel and timestep levels, significantly speeding up text-to-image generation without quality loss.

Contribution

It presents a hybrid approach that intelligently switches between models based on image complexity, improving efficiency over existing methods.

Findings

Achieves 1.83× speedup on Stable Diffusion 3

Outperforms all existing mixture model methods

Maintains high generation quality

Abstract

Diffusion models have demonstrated a remarkable ability in Text-to-Image (T2I) generation applications. Despite the advanced generation output, they suffer from heavy computation overhead, especially for large models that contain tens of billions of parameters. Prior work has illustrated that replacing part of the denoising steps with a smaller model still maintains the generation quality. However, these methods only focus on saving computation for some timesteps, ignoring the difference in compute demand within one timestep. In this work, we propose HybridStitch, a new T2I generation paradigm that treats generation like editing. Specifically, we introduce a hybrid stage that jointly incorporates both the large model and the small model. HybridStitch separates the entire image into two regions: one that is relatively easy to render, enabling an early transition to the smaller model, and another that is more complex and therefore requires refinement by the large model. HybridStitch employs the small model to construct a coarse sketch while exploiting the large model to edit and refine the complex regions. According to our evaluation, HybridStitch achieves 1.83$\times$ speedup on Stable Diffusion 3, which is faster than all existing mixture of model methods.