MOSS-Image improves text rich image generation with intent control
MOSS-Image Technical Report
Computer Vision and Pattern Recognition
Summary
Image generation systems today often struggle to create pictures that include a lot of text or complicated layouts. The authors developed MOSS-Image, which is an open-source tool that can make images with detailed text and precise arrangements. Their method includes training a large model to handle long text in two languages and a special module that understands what the user wants to include in the image. They also created tests to measure how well the system does at combining text and layout. MOSS-Image performs very well compared to other similar open models.
What this means in practice
- •For graphic designers: Generate images with complex text layouts and bilingual content that match user instructions precisely.
- •For digital marketing teams: Produce information-dense promotional graphics that require accurate text rendering in English and Chinese.
Authors
MOSI. AI
Abstract
Recent closed-source image generation systems can produce diverse visual content from simple compositions to information-dense graphics, yet open-weight text-to-image models still struggle with dense text and intricate layouts. To narrow this gap, we introduce MOSS-Image, an open generation system that combines high-fidelity visual synthesis with intent-aware composition. First, we pretrain a 5B single-stream diffusion transformer (DiT) from scratch with a carefully designed recipe, enabling bilingual long-form text rendering and precise layout control. Second, we develop an intention-aware agentic prompt enhancement (PE) module that translates diverse user requests into detailed content and layout specifications. Together, they enable MOSS-Image to generate text-rich and layout-coherent images. We further introduce LongLongTextBench (LLTB) for evaluating long-form English and Chinese text rendering, and RealLayoutArena (RLA) for end-to-end layout-rich generation. MOSS-Image achieves state-of-the-art performance among open-weight models on LLTB, with an average text coverage of 96.3%. On RLA, MOSS-Image ranks second in the end-to-end system comparison, while its backbone ranks first among generators evaluated with shared enhanced prompts. We will release the model, training recipe, and evaluation framework to support further research and promote the text-to-image community.