Mitigating Diffusion Model Hallucinations with Dynamic Guidance
This addresses the problem of unreliable image generation in diffusion models for AI practitioners, though it appears incremental as it builds on existing guidance techniques.
The paper tackles the problem of diffusion models producing hallucinatory samples with structural inconsistencies by introducing Dynamic Guidance, which selectively sharpens the score function along artifact-causing directions while preserving semantic variations. The method substantially reduces hallucinations on controlled and natural image datasets, significantly outperforming baselines.
Diffusion models, despite their impressive demos, often produce hallucinatory samples with structural inconsistencies that lie outside of the support of the true data distribution. Such hallucinations can be attributed to excessive smoothing between modes of the data distribution. However, semantic interpolations are often desirable and can lead to generation diversity, thus we believe a more nuanced solution is required. In this work, we introduce Dynamic Guidance, which tackles this issue. Dynamic Guidance mitigates hallucinations by selectively sharpening the score function only along the pre-determined directions known to cause artifacts, while preserving valid semantic variations. To our knowledge, this is the first approach that addresses hallucinations at generation time rather than through post-hoc filtering. Dynamic Guidance substantially reduces hallucinations on both controlled and natural image datasets, significantly outperforming baselines.