CoCo : Co de as Co T for Text-to-Image Preview and Rare Concept Generation
Haodong Li 1,2 , Chunmei Qing 1,∗ , Huanyu Zhang 3 , Dongzhi Jiang 5 , Yihang Zou 1 , Hongbo Peng 2 , Dingming Li 2 , Yuhong Dai 2 , ZePeng Lin 1 , Juanxi Tian 4 , Yi Zhou 1 , Siqi Dai 2 , Jingwei Wu 2 1 South China University of Technology 2 StepFun 3 Institute of Automation, Chinese Academy of Sciences 4 Nanyang Technological University 5 The Chinese University of Hong Kong ∗ * Corresponding Author · 2026
Read CoCo : Co de as Co T for Text-to-Image Preview and Rare Concept Generation by Haodong Li 1,2 , Chunmei Qing 1,∗ , Huanyu Zhang 3 , Dongzhi Jiang 5 , Yihang Zou 1 , Hongbo Peng 2 , Dingming Li 2 , Yuhong Dai 2 , ZePeng Lin 1 , Juanxi Tian 4 , Yi Zhou 1 , Siqi Dai 2 , Jingwei Wu 2 1 South China University of Technology 2 StepFun 3 Institute of Automation, Chinese Academy of Sciences 4 Nanyang Technological University 5 The Chinese University of Hong Kong ∗ * Corresponding Author (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-08652.txt.
Description
Haodong Li and colleagues reason about image making in code rather than in loose words. Word-only plans lack the precision that tangled layouts, structured visuals, and dense text demand. Their CoCo frame writes runnable code for the scene layout and runs it in a sandbox to draft the picture. Fine edits then polish that draft, beating plain generation on three test benches.
Catalog record
| CREATOR | Haodong Li 1,2 , Chunmei Qing 1,∗ , Huanyu Zhang 3 , Dongzhi Jiang 5 , Yihang Zou 1 , Hongbo Peng 2 , Dingming Li 2 , Yuhong Dai 2 , ZePeng Lin 1 , Juanxi Tian 4 , Yi Zhou 1 , Siqi Dai 2 , Jingwei Wu 2 1 South China University of Technology 2 StepFun 3 Institute of Automation, Chinese Academy of Sciences 4 Nanyang Technological University 5 The Chinese University of Hong Kong ∗ * Corresponding Author |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33457 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.08652