Research map / Image generation / Text-to-image diffusion
FE2E: From Editor to Dense Geometry Estimator
Jiyuan Wang, Chunyu Lin, Lei Sun, Rongying Liu, Lang Nie, Mingxing Li, Kang Liao, Xiangxiang Chu
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 |
|---|---|---|---|
| use an image editor (Step1X-Edit) instead of generator (FLUX) as foundation, direct adaptation (ablation) | FE2E · KITTI depth | AbsRel: 9.7 -> 9.5 | Improved |
| use Step1X-Edit instead of FLUX with improved protocol (ablation) | FE2E · ETH3D depth | AbsRel: 4.5 -> 3.9 | Improved |
| switch to consistent velocity objective (ablation) | FE2E · KITTI depth | AbsRel: 9.5 -> 8.8 | Improved |
| add fixed starting point (zero) in flow (ablation) | FE2E · KITTI depth | AbsRel: 8.8 -> 8.6 | Improved |
| use inverse quantization instead of uniform (direct) quantization (ablation) | FE2E · KITTI depth | AbsRel: 8.6 -> 6.9 | Improved |
| use logarithmic instead of inverse quantization (ablation) | FE2E · ETH3D depth | AbsRel: 4.6 -> 3.9 | Improved |
| add cost-free joint depth and normal estimation (ablation) | FE2E · KITTI depth | AbsRel: 6.8 -> 6.6 | Improved |
| add auxiliary dispersion loss (ablation) | FE2E · ETH3D depth | AbsRel: 4.8 -> 4.5 | Improved |
| swap editor to FLUX-Kontext (ablation) | FE2E · depth KITTI/ETH3D | AbsRel: KITTI 6.6 -> 6.7, ETH3D 3.8 -> 3.6 | Mixed |
| fine-tuned editor versus DepthAnything trained on 62.6M images | FE2E · ETH3D depth | AbsRel: 12.7 -> 3.8 | Improved |
| FE2E versus DepthAnything on ScanNet | FE2E · ScanNet depth | AbsRel: 4.3 -> 4.4 | Hurt |
Later papers that compared with this method
1 later accepted paper reports FE2E as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- MUSE: Unlocking Timestep as Native Task Steering for One-Step Dense Prediction (ECCV 2026 · better on 0 of 4 shared results)
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
- Exploiting Diffusion Prior for Generalizable Dense Prediction (CVPR 2024 · similarity 0.88)
- Lotus: Diffusion-based Visual Foundation Model for High-quality Dense Prediction (ICLR 2025 · similarity 0.86)
- Accelerating Text-to-Image Editing via Cache-Enabled Sparse Diffusion Inference (AAAI 2024 · similarity 0.86)
- Perturb-and-Revise: Flexible 3D Editing with Generative Trajectories (CVPR 2025 · similarity 0.86)
- WorDepth: Variational Language Prior for Monocular Depth Estimation (CVPR 2024 · similarity 0.86)
- EditAR: Unified Conditional Generation with Autoregressive Models (CVPR 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.
