Research map / Image generation / Text-to-image diffusion
PosterOmni: Generalized Artistic Poster Creation via Task Distillation and Unified Reward Feedback
Sixiang Chen, Jianyu Lai, Jialin Gao, Hengyu Shi, Zhongying Liu, Tian Ye, Junfeng Luo, Xiaoming Wei, Lei Zhu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| distill local and global experts into one student instead of mixed training (ablation) | PosterOmni-SFT · PosterOmni-Bench-en (Extend / Layout) | score: 4.33 / 3.72 -> 4.43 / 3.89 | Improved |
| use individual local expert only (ablation) | Qwen-Image-Edit [2509] · PosterOmni-Bench-en (Layout-driven) | score: 3.44 -> 2.79 (layout) | Hurt |
| use individual global expert only (ablation) | Qwen-Image-Edit [2509] · PosterOmni-Bench-en (Extend) | score: 4.28 -> 3.35 (extend) | Hurt |
| add auxiliary text-rendering loss to task distillation (ablation) | PosterOmni-SFT · PosterOmni-Bench-en (Extend / Layout) | score: 4.39 / 3.82 -> 4.43 / 3.89 | Improved |
| replace VLM reward with unified Romni reward (ablation) | Omni-Edit RL · PosterOmni-Bench-en (Extend / Layout) | score: 4.58 / 3.97 -> 4.76 / 4.20 | Improved |
| use Omni-Edit RL (DiffusionNFT) instead of FlowGRPO (ablation) | Romni-rewarded RL · PosterOmni-Bench-en (Extend / Layout) | score: 4.65 / 4.08 -> 4.76 / 4.20 | Improved |
| apply full PosterOmni pipeline to base editor | Qwen-Image-Edit [2509] · PosterOmni-Bench-en (Overall) | score: 3.51 -> 4.27 | Improved |
| apply full PosterOmni pipeline to base editor | Qwen-Image-Edit [2509] · PosterOmni-Bench-cn (Overall) | score: 3.46 -> 4.37 | Improved |
| PosterOmni vs proprietary Seedream-4.0 | PosterOmni · PosterOmni-Bench-en (Rescaling / Filling) | score: Rescaling 4.00 vs 3.97; Overall 4.28 vs 4.27 | Mixed |
Closest papers by meaning
- PosterCraft: Rethinking High-Quality Aesthetic Poster Generation in a Unified Framework (ICLR 2026 · similarity 0.92)
- POSTA: A Go-to Framework for Customized Artistic Poster Generation (CVPR 2025 · similarity 0.90)
- P2P: Automated Paper-to-Poster Generation and Fine-Grained Benchmark (ICLR 2026 · similarity 0.87)
- PosterO: Structuring Layout Trees to Enable Language Models in Generalized Content-Aware Layout Generation (CVPR 2025 · similarity 0.86)
- Imagine Before You Draw: Visual Prompt Engineering for Image Generation (NeurIPS 2026 · similarity 0.86)
- Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
