To address the limitations of current evaluation methods in multi-modal image generation—which often assess textual and visual conditions in isolation—the researchers introduce UFO, a unified framework designed for simultaneous omni-condition alignment evaluation. Unlike existing embedding-based or Multi-modal Large Language Model (MLLM)-based approaches, UFO utilizes an Atomized Chain-of-Evaluation paradigm. This method decomposes the complex task of omni-condition alignment into a sequential chain of fine-grained, disentangled Atomic Evaluation Units (AEUs). These units are organized into distinct modality-relevance classes, allowing the framework to utilize either general or dedicated functional calls to verify specific AEU types accurately. To validate this approach, the authors developed UFO-Bench, a specialized benchmark intended to holistically assess how customization models handle the mutual interactions between diverse textual and visual conditions. Empirical results indicate that UFO achieves superior performance in mimicking human preference, demonstrating an average correlation improvement of 15.25% compared to existing evaluation methods. By focusing on the simultaneous alignment of multiple modalities rather than isolated checks, UFO provides a more consistent metric for subject-driven customization models.