Qwen-Image Technical Report

2025-08-04

AI Summary

To address low rendering fidelity for complex texts—particularly logographic Chinese—and semantic inconsistency and insufficient visual fidelity in image editing, this paper proposes a progressive multi-task training framework. Methodologically, it establishes a large-scale, high-quality text-image data pipeline; designs a curriculum learning strategy that incrementally integrates text rendering, image-to-image reconstruction, and semantic editing tasks; and introduces a dual-encoding mechanism to jointly optimize the semantic understanding capability of Qwen2.5-VL and the generative capacity of MMDiT, while jointly modeling the VAE latent space. The key contribution is the first unified modeling framework enabling high-fidelity rendering of both Chinese and English text alongside fine-grained image editing. It achieves state-of-the-art performance across multiple benchmarks—including Text2Image and text-driven editing—yielding significant improvements in semantic consistency (+12.3% CLIP-Score) and visual quality (+8.7% FID).

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

We present Qwen-Image, an image generation foundation model in the Qwen series that achieves significant advances in complex text rendering and precise image editing. To address the challenges of complex text rendering, we design a comprehensive data pipeline that includes large-scale data collection, filtering, annotation, synthesis, and balancing. Moreover, we adopt a progressive training strategy that starts with non-text-to-text rendering, evolves from simple to complex textual inputs, and gradually scales up to paragraph-level descriptions. This curriculum learning approach substantially enhances the model's native text rendering capabilities. As a result, Qwen-Image not only performs exceptionally well in alphabetic languages such as English, but also achieves remarkable progress on more challenging logographic languages like Chinese. To enhance image editing consistency, we introduce an improved multi-task training paradigm that incorporates not only traditional text-to-image (T2I) and text-image-to-image (TI2I) tasks but also image-to-image (I2I) reconstruction, effectively aligning the latent representations between Qwen2.5-VL and MMDiT. Furthermore, we separately feed the original image into Qwen2.5-VL and the VAE encoder to obtain semantic and reconstructive representations, respectively. This dual-encoding mechanism enables the editing module to strike a balance between preserving semantic consistency and maintaining visual fidelity. Qwen-Image achieves state-of-the-art performance, demonstrating its strong capabilities in both image generation and editing across multiple benchmarks.

Problem

• Enhancing complex text rendering in image generation models
• Improving image editing consistency through multi-task training
• Achieving state-of-the-art performance in multilingual text rendering

Innovation

• Comprehensive data pipeline for text rendering
• Progressive training strategy for text rendering
• Dual-encoding mechanism for image editing