Research map / Image generation / Editing and style transfer
OrionEdit: Bridging Reference and Source Images for Generalized Cross-Image Editing
Zeyu Jiang, Lai Man Po, Xuyuan Xu, Yexin Wang, Guoping Gong, Haoxuan Wu, Chenbo Yan, Kun Li, Yuyang Liu
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 |
|---|---|---|---|
| add orthogonal subspace constraint between branch adapters (ablation) | OrionEdit (Qwen-Image-Edit-2509) · OrionEditBench ablation | DPG: 53.81 -> 76.35 | Improved |
| add orthogonal subspace constraint between branch adapters (ablation) | OrionEdit (Qwen-Image-Edit-2509) · OrionEditBench ablation | DINO: 0.738 -> 0.752 | Improved |
| add reverse-causal information-flow mask (ablation) | OrionEdit (Qwen-Image-Edit-2509) · OrionEditBench ablation | DPG: 53.81 -> 72.74 | Improved |
| add random attention mask instead of reverse-causal mask (ablation) | OrionEdit (Qwen-Image-Edit-2509) · OrionEditBench ablation | DPG: 53.81 -> 49.37 | Hurt |
| combine orthogonality and reverse-causal mask (ablation) | OrionEdit (Qwen-Image-Edit-2509) · OrionEditBench ablation | DPG: 76.35 -> 83.83 | Improved |
| increase rank of B from 128 to 256 (ablation) | OrionEdit-qwen · OrionEditBench ablation | DPG: 83.83 -> 85.24 | Improved |
| use OrionEdit on Qwen backbone instead of plain Qwen-Image-Edit | Qwen-Image-Edit-2509 · OrionEditBench average | DPG: 66.08 -> 87.02 | Improved |
| use Flux-Kontext backbone instead of Qwen-Image-Edit for OrionEdit | OrionEdit · OrionEditBench average | DPG: 87.02 -> 81.27 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DreamOmni2 — DreamOmni2: Multimodal Instruction-based Generation and Editing (CVPR 2026) · better on 1 of 1 shared result
- OmniStyle — OmniStyle: Filtering High Quality Style Transfer Data at Scale (CVPR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FreeEdit: Mask-Free Reference-Based Image Editing With Multi-Modal Instruction (TPAMI 2025 · similarity 0.89)
- PIXELS: Progressive Image Xemplar-based Editing with Latent Surgery (AAAI 2025 · similarity 0.88)
- UltraEdit: Instruction-based Fine-Grained Image Editing at Scale (NeurIPS 2024 · similarity 0.88)
- Zero-shot Image Editing with Reference Imitation (NeurIPS 2024 · similarity 0.87)
- PairEdit: Learning Semantic Variations for Exemplar-based Image Editing (NeurIPS 2025 · similarity 0.87)
- FireEdit: Fine-grained Instruction-based Image Editing via Region-aware Vision Language Model (CVPR 2025 · similarity 0.87)
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.
