CoCo: Code as CoT for Text-to-Image Preview and Rare Concept Generation
2026-03-09 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors present CoCo, a new method for turning text into images by writing and running code that plans the picture step-by-step. Instead of using vague descriptions, their system creates precise, executable code to design a draft image layout, then improves it into a final detailed picture. They built a special dataset to teach the model how to make drafts and refine them. Their experiments show CoCo makes better and more controllable images than previous methods that only use language-based planning.
Unified Multimodal ModelsText-to-Image GenerationChain-of-Thought ReasoningExecutable CodeStructured LayoutImage EditingDatasetVisual RefinementControllable GenerationSandbox Environment
Authors
Haodong Li, Chunmei Qing, Huanyu Zhang, Dongzhi Jiang, Yihang Zou, Hongbo Peng, Dingming Li, Yuhong Dai, ZePeng Lin, Juanxi Tian, Yi Zhou, Siqi Dai, Jingwei Wu
Abstract
Recent advancements in Unified Multimodal Models (UMMs) have significantly advanced text-to-image (T2I) generation, particularly through the integration of Chain-of-Thought (CoT) reasoning. However, existing CoT-based T2I methods largely rely on abstract natural-language planning, which lacks the precision required for complex spatial layouts, structured visual elements, and dense textual content. In this work, we propose CoCo (Code-as-CoT), a code-driven reasoning framework that represents the reasoning process as executable code, enabling explicit and verifiable intermediate planning for image generation. Given a text prompt, CoCo first generates executable code that specifies the structural layout of the scene, which is then executed in a sandboxed environment to render a deterministic draft image. The model subsequently refines this draft through fine-grained image editing to produce the final high-fidelity result. To support this training paradigm, we construct CoCo-10K, a curated dataset containing structured draft-final image pairs designed to teach both structured draft construction and corrective visual refinement. Empirical evaluations on StructT2IBench, OneIG-Bench, and LongText-Bench show that CoCo achieves improvements of +68.83%, +54.8%, and +41.23% over direct generation, while also outperforming other generation methods empowered by CoT. These results demonstrate that executable code is an effective and reliable reasoning paradigm for precise, controllable, and structured text-to-image generation. The code is available at: https://github.com/micky-li-hd/CoCo