Research map / Image generation / Editing and style transfer
Source Prompt Disentangled Inversion for Boosting Image Editability with Diffusion Models
Ruibin Li, Ruihuang Li, Song Guo, Lei Zhang
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 |
|---|---|---|---|
| replace DDIM inversion with fixed-point loss search (SPDInv) on top of P2P | P2P editing engine · PIE-Bench | DINO distance x10^3: 69.43 -> 8.81 | Improved |
| SPDInv versus AIDI fixed-point iteration under MasaCtrl | MasaCtrl engine · PIE-Bench | DINO distance x10^3: 55.93 -> 20.48 | Improved |
| apply SPDInv to PNP engine vs DirectInv | PNP engine · PIE-Bench | PSNR: 22.43 -> 26.72 | Improved |
| integrate SPDInv into ELITE customized generation | ELITE · Localized editing | PSNR: 14.83 -> 24.14 | Improved |
| SPDInv inversion cost | P2P · PIE-Bench | Inversion time (s): 11.55 -> 27.04 | Hurt |
| increase max optimization rounds K from 5 to 50 (ablation) | SPDInv · PIE-Bench subset | CLIP: PSNR 31.49 -> 32.12; CLIP 26.70 -> 26.32 | Mixed |
| increase learning rate eta to 0.1 (ablation) | SPDInv · PIE-Bench subset | PSNR: 31.55 -> 22.82 | Hurt |
| tighten stopping threshold delta below 5e-6 (ablation) | SPDInv · PIE-Bench subset | PSNR: 31.61 -> 31.65 | No clear effect |
| SPDInv reduces noise code dependence on source prompt vs DDIM inversion | inverted noise code · 100 generated images | CLIP score between noise and source prompt: 14.55 -> 13.18 | Improved |
| SPDInv on mask-less addition or portrait editing | SPDInv with P2P/MasaCtrl/PNP · failure cases | qualitative | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ELITE — ELITE: Encoding Visual Concepts into Textual Embeddings for Customized Text-to-Image Generation (ICCV 2023) · 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
- EditInfinity: Image Editing with Binary-Quantized Generative Models (NeurIPS 2025 · similarity 0.92)
- PnP Inversion: Boosting Diffusion-based Editing with 3 Lines of Code (ICLR 2024 · similarity 0.91)
- Inversion-Free Image Editing with Language-Guided Diffusion Models (CVPR 2024 · similarity 0.91)
- Editable Noise Map Inversion: Encoding Target-image into Noise For High-Fidelity Image Manipulation (ICML 2025 · similarity 0.90)
- Object-Aware Inversion and Reassembly for Image Editing (ICLR 2024 · similarity 0.90)
- DCI: Dual-Conditional Inversion for Boosting Diffusion-Based Image Editing (NeurIPS 2025 · similarity 0.89)
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.
