UFO: Chain-of-Evaluation for Omni-Condition Alignment in Multi-Modal Image Generation
Researchers propose UFO, a unified evaluation framework for multi-modal image generation that improves correlation with human judgments by 15.25%.
The paper introduces UFO, the first unified framework for evaluating omni-condition alignment in multi-modal image generation, addressing the weakness of existing embedding-based and MLLM-based methods that assess each modal condition in isolation. Its Atomized Chain-of-Evaluation paradigm decomposes alignment into fine-grained, disentangled Atomic Evaluation Units verified via general or dedicated functional calls. UFO achieves the highest correlation with human evaluation preferences, with an average improvement of 15.25%. The authors also release UFO-Bench, a benchmark for evaluating subject-driven customization models under diverse textual and visual condition interactions.
- First unified framework evaluating simultaneous alignment with textual and visual conditions
- Atomized Chain-of-Evaluation decomposes alignment into Atomic Evaluation Units
- Average 15.25% improvement in correlation with human evaluation preferences
- UFO-Bench evaluates customization models under diverse condition interactions
Full article169 words · extracted from huggingface.co · click to collapse
Multi-modal image generation, particularly subject-driven customization, has garnered growing attention in recent years. Despite the rapid advancement of generative models, their evaluation remains largely lagging. Existing methods, whether embedding-based or Multi-modal Large Language Model (MLLM)-based, evaluate alignment with each modal condition in isolation, which contradicts the simultaneous condition alignment objective of multi-modal image generation, leading to poor consistency with human judgments. To address this challenge, we propose UFO, the first unified framework for omni-condition alignment simultaneous evaluation. Specifically, UFO introduces a novel Atomized Chain-of-Evaluation paradigm, i.e., it first decomposes omni-condition alignment into a sequential chain of fine-grained, disentangled Atomic Evaluation Units (AEUs), categorizes them into distinct modality-relevance classes, and then employs general or dedicated functional calls for accurate verification of different AEU types. Experimental results demonstrate that UFO achieves the highest correlation with human evaluation preferences, delivering an average improvement of 15.25%. Furthermore, we present UFO-Bench, a dedicated benchmark designed to holistically evaluate the performance of existing customization models under the diverse mutual interactions of textual and visual conditions.
Text extracted automatically; images, tables and formatting may be missing. Original: https://huggingface.co/papers/2609.12397