AdaptDiff: Adaptive Guidance in Diffusion Models for Diverse and Identity-Consistent Face Synthesis (Student Abstract)

TL;DR

This paper introduces AdaptDiff, a method that dynamically adjusts guidance during diffusion-based face synthesis to enhance diversity and identity consistency in generated images.

Contribution

It proposes a novel adaptive guidance scheme that modulates negative conditions throughout sampling, improving face synthesis quality.

Findings

Enhanced diversity in generated face images.

Improved identity preservation across samples.

Effective suppression of undesired attributes.

Abstract

Diffusion models conditioned on identity embeddings enable the generation of synthetic face images that consistently preserve identity across multiple samples. Recent work has shown that introducing an additional negative condition through classifier-free guidance during sampling provides a mechanism to suppress undesired attributes, thus improving inter-class separability. Building on this insight, we propose a dynamic weighting scheme for the negative condition that adapts throughout the sampling trajectory. This strategy leverages the complementary strengths of positive and negative conditions at different stages of generation, leading to more diverse yet identity-consistent synthetic data.