Lightning-Fast Image Inversion and Editing for Text-to-Image Diffusion Models
This addresses the bottleneck of slow image inversion for users of diffusion models, enabling real-time editing applications, though it is an incremental improvement over existing numerical methods.
The paper tackles the problem of slow and poor-quality deterministic inversion for text-to-image diffusion models by formulating it as finding roots of an implicit equation and developing an efficient guided Newton-Raphson method. It achieves inversion in 0.4 seconds on an A100 GPU for models like SDXL-Turbo and Flux, enabling interactive image editing with high-quality reconstructions.
Diffusion inversion is the problem of taking an image and a text prompt that describes it and finding a noise latent that would generate the exact same image. Most current deterministic inversion techniques operate by approximately solving an implicit equation and may converge slowly or yield poor reconstructed images. We formulate the problem by finding the roots of an implicit equation and devlop a method to solve it efficiently. Our solution is based on Newton-Raphson (NR), a well-known technique in numerical analysis. We show that a vanilla application of NR is computationally infeasible while naively transforming it to a computationally tractable alternative tends to converge to out-of-distribution solutions, resulting in poor reconstruction and editing. We therefore derive an efficient guided formulation that fastly converges and provides high-quality reconstructions and editing. We showcase our method on real image editing with three popular open-sourced diffusion models: Stable Diffusion, SDXL-Turbo, and Flux with different deterministic schedulers. Our solution, Guided Newton-Raphson Inversion, inverts an image within 0.4 sec (on an A100 GPU) for few-step models (SDXL-Turbo and Flux.1), opening the door for interactive image editing. We further show improved results in image interpolation and generation of rare objects.