Exploiting the Signal-Leak Bias in Diffusion Models
This addresses the problem of limited control in image generation for users of diffusion models, though it is incremental as it builds on existing models.
The paper identifies a signal-leak bias in diffusion models that causes sub-optimal style matching, and exploits it to generate images with more varied brightness and better style or color matching without additional training.
There is a bias in the inference pipeline of most diffusion models. This bias arises from a signal leak whose distribution deviates from the noise distribution, creating a discrepancy between training and inference processes. We demonstrate that this signal-leak bias is particularly significant when models are tuned to a specific style, causing sub-optimal style matching. Recent research tries to avoid the signal leakage during training. We instead show how we can exploit this signal-leak bias in existing diffusion models to allow more control over the generated images. This enables us to generate images with more varied brightness, and images that better match a desired style or color. By modeling the distribution of the signal leak in the spatial frequency and pixel domains, and including a signal leak in the initial latent, we generate images that better match expected results without any additional training.