| title | Unlocking Complex Visual Generation via Closed-Loop Verified Reasoning |
|---|---|
| authors | Hanbo Cheng, Limin Lin, Ruo Zhang, Yicheng Pan, Jun Du |
| date | 2026-05-14 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2605.14876 |
| affiliation | University of Science and Technology of China |
| upvotes | 2 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Despite rapid advancements, current text-to-image (T2I) models predominantly rely on a single-step generation paradigm, which struggles with complex semantics and faces diminishing returns from parameter scaling. While recent multi-step reasoning approaches show promise, they are hindered by ungrounded planning hallucinations lacking verification, monolithic post-hoc reflection, long-context optimization instabilities, and prohibitive inference latency. To overcome these bottlenecks, we propose the Closed-Loop Visual Reasoning (CLVR) framework, a comprehensive system that deeply couples visual-language logical planning with pixel-level diffusion generation. CLVR introduces an automated data engine with step-level visual verification to synthesize reliable reasoning trajectories, and proposes Proxy Prompt Reinforcement Learning (PPRL) to resolve long-context optimization instabilities by distilling interleaved multimodal histories into explicit reward signals for accurate causal attribution. Furthermore, to mitigate the severe latency bottleneck caused by iterative denoising, we propose Δ-Space Weight Merge (DSWM), a theoretically grounded method that fuses alignment weights with off-the-shelf distillation priors, reducing the per-step inference cost to just 4 NFEs without requiring expensive re-distillation. Extensive experiments demonstrate that CLVR outperforms existing open-source baselines across multiple benchmarks and approaches the performance of proprietary commercial models, unlocking general test-time scaling capabilities for complex visual generation. |
cite
@misc{cheng2026unlocking,
title={Unlocking Complex Visual Generation via Closed-Loop Verified Reasoning},
author={Hanbo Cheng and Limin Lin and Ruo Zhang and Yicheng Pan and Jun Du},
year={2026},
eprint={2605.14876},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.14876}
}