Diagnosing and Correcting Concept Omission in Multimodal Diffusion Transformers
For practitioners of text-to-image generation, this work addresses the common failure mode of concept omission with a practical correction method.
Multimodal Diffusion Transformers often omit specified objects or attributes in generated images. The authors identify an 'omission signal' in text embeddings and propose Omission Signal Intervention (OSI) to amplify it, significantly reducing omission in FLUX.1-Dev and SD3.5-Medium.
Multimodal Diffusion Transformers (MM-DiTs) have achieved remarkable progress in text-to-image generation, yet they frequently suffer from concept omission, where specified objects or attributes fail to emerge in the generated image. By performing linear probing on text tokens, we demonstrate that text embeddings can distinguish a characteristic `omission signal' representing the absence of target concepts. Leveraging this insight, we propose Omission Signal Intervention (OSI), which amplifies the omission signal to actively catalyze the generation of missing concepts. Comprehensive experiments on FLUX.1-Dev and SD3.5-Medium demonstrate that OSI significantly alleviates concept omission even in extreme scenarios.