Identity-preserving Distillation Sampling by Fixed-Point Iterator

TL;DR

This paper introduces Identity-preserving Distillation Sampling (IDS) with a fixed-point iterative regularization (FPR) technique to enhance text-conditioned image and 3D object generation, maintaining identity and structure during editing.

Contribution

The paper proposes a novel regularization method, FPR, that modifies scores to preserve identity and structure in score distillation sampling, improving image and 3D editing quality.

Findings

FPR effectively maintains identity in image editing.

IDS with FPR produces clearer, more consistent results.

Structural integrity is better preserved compared to existing methods.

Abstract

Score distillation sampling (SDS) demonstrates a powerful capability for text-conditioned 2D image and 3D object generation by distilling the knowledge from learned score functions. However, SDS often suffers from blurriness caused by noisy gradients. When SDS meets the image editing, such degradations can be reduced by adjusting bias shifts using reference pairs, but the de-biasing techniques are still corrupted by erroneous gradients. To this end, we introduce Identity-preserving Distillation Sampling (IDS), which compensates for the gradient leading to undesired changes in the results. Based on the analysis that these errors come from the text-conditioned scores, a new regularization technique, called fixed-point iterative regularization (FPR), is proposed to modify the score itself, driving the preservation of the identity even including poses and structures. Thanks to a self-correction by FPR, the proposed method provides clear and unambiguous representations corresponding to the given prompts in image-to-image editing and editable neural radiance field (NeRF). The structural consistency between the source and the edited data is obviously maintained compared to other state-of-the-art methods.