0

DMM: Building a Versatile Image Generation Model via Distillation-Based Model Merging

The paper introduces score distillation based model merging for text-to-image generation, enabling a single versatile model to produce images in different styles based on style vectors.

Year
2025
Venue
arXiv 2025
Authors
7
Hosting
Abstract onlyARXIV-DEFAULT

Cite

Notes

Only stored in your browser.

Attribution

Abstract & full text
arxiv.org/abs/2504.12364ARXIV-DEFAULT
TL;DR
Semantic Scholar
Attribution policy →

Abstract

The success of text-to-image (T2I) generation models has spurred a proliferation of numerous model checkpoints fine-tuned from the same base model on various specialized datasets. This overwhelming specialized model production introduces new challenges for high parameter redundancy and huge storage cost, thereby necessitating the development of effective methods to consolidate and unify the capabilities of diverse powerful models into a single one. A common practice in model merging adopts static linear interpolation in the parameter space to achieve the goal of style mixing. However, it neglects the features of T2I generation task that numerous distinct models cover sundry styles which may lead to incompatibility and confusion in the merged model. To address this issue, we introduce a style-promptable image generation pipeline which can accurately generate arbitrary-style images under the control of style vectors. Based on this design, we propose the score distillation based model merging paradigm (DMM), compressing multiple models into a single versatile T2I model. Moreover, we rethink and reformulate the model merging task in the context of T2I generation, by presenting new merging goals and evaluation protocols. Our experiments demonstrate that DMM can compactly reorganize the knowledge from multiple teacher models and achieve controllable arbitrary-style generation.

Authors

7