Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Divide-and-Denoise: A Game-Theoretic Method for Fairly Composing Diffusion Models

About

The abundance of pre-trained diffusion models provides an opportunity for composition. Combining several models, however, runs the risk of one model dominating or models disagreeing with each other. Here, we propose Divide-and-Denoise, a method for coordinating multiple pre-trained diffusion models during sampling. Much like managing a specialized workforce, our method creates a fair but efficient division of labor across models. Central to our method is the notion of an allocation which defines the responsibility of each model to every region of the noisy sample. At every timestep, we then denoise by (i) updating the allocation by solving a fair division game, where we divide the sample into regions that maximize total utility under fairness constraints, and (ii) aligning the models with this allocation, where we guide each model to denoise within its assigned region. This leads to a new composite denoising process that evolves in tandem with a division process. We evaluate Divide-and-Denoise on conditional image generation. Across several quality metrics, including the GenEval benchmark, our method outperforms baselines and resolves common failures including missing objects and mismatched attributes. Experiments show that Divide-and-Denoise utilizes each model's expertise without neglecting any other model.

Abhi Gupta, Polina Barabanshchikova, Vikas Garg, Samuel Kaski, Tommi Jaakkola• 2026

Related benchmarks

TaskDatasetResultRank
Multi-Concept Image GenerationGenEval
Image Coverage88.5
12
Text-to-Image GenerationGenEval Descriptive Concepts Attribute Binding Stable Diffusion setup (100 unique pairs)
GenEval Success Rate (Images)55.75
5
Text-to-Image Generation40 hand-designed conflicting concept sets (test)
CLIP Joint Score31.13
5
Multi-object image generationImageNet-1K
CLIP Joint Score29.03
4
Showing 4 of 4 rows

Other info

Follow for update