cs.CVFeb 14, 2026

Diff-Aid: Inference-time Adaptive Interaction Denoising for Rectified Text-to-Image Generation

Authors: Binglei Li, Mengping Yang, Zhiyu Tan, Qiang Xiang, Junping Zhang, Hao Li

Organizations: Fudan University · Shanghai Innovation Institute · Shanghai Academy of AI for Science

Abstract

Recent text-to-image (T2I) diffusion models have achieved remarkable advancement, yet faithfully following complex textual descriptions remains challenging due to insufficient interactions between textual and visual features. Prior approaches enhance such interactions via architectural design or handcrafted textual condition weighting, but lack flexibility and overlook the dynamic interactions across different blocks and denoising stages. To provide a more flexible and efficient solution to this problem, we propose Diff-Aid, a lightweight inference-time method that adaptively adjusts per-token text and image interactions across transformer blocks and denoising timesteps. Beyond improving generation quality, Diff-Aid yields interpretable modulation patterns that reveal how different blocks, timesteps, and textual tokens contribute to semantic alignment during denoising. As a plug-and-play module, Diff-Aid can be seamlessly integrated into downstream applications for further improvement, including style LoRAs, controllable generation, and zero-shot editing. Experiments on strong baselines (SD 3.5 and FLUX) demonstrate consistent improvements in prompt adherence, visual quality, and human preference across various metrics. Our code and models will be released.

Figures & tables

Appendix figures & tables23 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. RAISE: Requirement-Adaptive Evolutionary Refinement for Training-Free Text-to-Image Alignment

    Feb 28, 2026Liyao Jiang, Ruichen Chen, Chao Gao +1T2I GenerationText-Image Alignment

  2. TexTailor: Inference-Time Textual Guidance Tailoring for Multimodal Diffusion Transformers

    Jan 5, 2026Binglei Li, Mengping Yang, Zhiyu Tan +2Diffusion TransformerCompositional T2I Generation

  3. DICE: Distilling Classifier-Free Guidance into Text Embeddings

    Feb 6, 2025Zhenyu Zhou, Defang Chen, Can Wang +2Diffusion Model AlignmentDiffusion Model Guidance