| title | TILT: Improving Compositional Generation in Diffusion Models with a Model-Intrinsic Reward |
|---|---|
| authors | Debottam Dutta, Jaehoon Hahm, Jianchong Chen, Romit Roy Choudhury |
| date | 2026-05-16 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2607.21606 |
| upvotes | 2 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Recent advances in powerful text-to-image generation models have made it increasingly important to develop test-time methods that modify the sampling trajectory to produce images more faithful to complex compositional prompts. We present TILT, a training-free framework for compositional text-to-image generation via test-time reward alignment. We interpret compositional failures as overlap modes between joint and single-concept distributions, and define a reward that favors samples where all concepts are jointly present. This reward is intrinsic to the base model and does not require any external supervision or reward models. This yields a KL-constrained objective with a closed-form tilted target distribution and principled guiding steps for diffusion sampling. The interaction of concept distributions together with the above reward naturally leads to two different guidance strategies while a hybrid approach that balances their respective benefits produces stronger performance. Experiments on prompts from T2ICompBench show that our method improves compositional alignment while preserving image quality compared to previous baselines. |
cite
@misc{dutta2026tilt,
title={TILT: Improving Compositional Generation in Diffusion Models with a Model-Intrinsic Reward},
author={Debottam Dutta and Jaehoon Hahm and Jianchong Chen and Romit Roy Choudhury},
year={2026},
eprint={2607.21606},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2607.21606}
}