Extending One-Step Image Generation from Class Labels to Text via Discriminative Text Representation

2026-04-20

AI Summary

This work addresses the challenge that single-step image generation struggles to accommodate flexible text conditioning by integrating a highly discriminative large language model as the text encoder within the MeanFlow framework. Recognizing the stringent discriminative requirements imposed by one-step generation on textual representations, the authors redesign the generative pipeline to enable high-quality text-to-image synthesis in a single step for the first time. The proposed method not only achieves substantially improved generation quality on standard benchmarks but also demonstrates the generalizability of this strategy within diffusion models, offering a practical solution for efficient few-step and even single-step text-to-image generation.

Technology Category

• Natural Language Processing: Generation
• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Few-step generation has been a long-standing goal, with recent one-step generation methods exemplified by MeanFlow achieving remarkable results. Existing research on MeanFlow primarily focuses on class-to-image generation. However, an intuitive yet unexplored direction is to extend the condition from fixed class labels to flexible text inputs, enabling richer content creation. Compared to the limited class labels, text conditions pose greater challenges to the model's understanding capability, necessitating the effective integration of powerful text encoders into the MeanFlow framework. Surprisingly, although incorporating text conditions appears straightforward, we find that integrating powerful LLM-based text encoders using conventional training strategies results in unsatisfactory performance. To uncover the underlying cause, we conduct detailed analyses and reveal that, due to the extremely limited number of refinement steps in the MeanFlow generation, such as only one step, the text feature representations are required to possess sufficiently high discriminability. This also explains why discrete and easily distinguishable class features perform well within the MeanFlow framework. Guided by these insights, we leverage a powerful LLM-based text encoder validated to possess the required semantic properties and adapt the MeanFlow generation process to this framework, resulting in efficient text-conditioned synthesis for the first time. Furthermore, we validate our approach on the widely used diffusion model, demonstrating significant generation performance improvements. We hope this work provides a general and practical reference for future research on text-conditioned MeanFlow generation. The code is available at https://github.com/AMAP-ML/EMF.

Problem

• one-step generation
• text-to-image
• MeanFlow
• text conditioning
• discriminative representation

Innovation

• one-step generation
• text-conditioned image synthesis
• discriminative text representation
• MeanFlow
• LLM-based text encoder