Research map / Image generation / Editing and style transfer
EGSDE: Unpaired Image-to-Image Translation via Energy-Guided Stochastic Differential Equations
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 energy guidance from source+target-domain feature extractors to SDEdit-style SDE | SDEdit · Cat->Dog | FID: 74.17 -> 65.82 | Helped |
| add energy guidance | SDEdit · Wild->Dog | FID: 68.51 -> 59.75 | Helped |
| add energy guidance | SDEdit · Male->Female | FID: 49.43 -> 41.93 | Helped |
| add energy guidance | SDEdit · Cat->Dog | L2 distance: 47.88 -> 47.22 | Helped |
| increase lambda_s (realism expert) from 0 to 500 (ablation) | EGSDE · Cat->Dog | FID: 73.85 -> 62.44 | Helped |
| increase lambda_s from 0 to 500 (ablation) | EGSDE · Cat->Dog | L2 distance: 47.87 -> 51.02 | Hurt |
| increase lambda_i (faithful expert) from 0 to 2 (ablation) | EGSDE · Cat->Dog | L2 distance: 47.87 -> 44.23 | Helped |
| increase lambda_i from 0 to 2 (ablation) | EGSDE · Cat->Dog | FID: 73.85 -> 77.05 | Hurt |
| use energy on perturbed source x_t instead of noise-free x0 (ablation) | EGSDE · Cat->Dog | FID: 79.01 -> 65.82 | Helped |
| use energy on perturbed source x_t instead of noise-free x0 (ablation) | EGSDE · Male->Female | FID: 41.86 -> 41.93 | No clear effect |
| increase initial time M from 0.3T to 0.7T (ablation) | EGSDE · Cat->Dog | FID: 97.02 -> 53.01 | Helped |
| increase initial time M from 0.3T to 0.7T (ablation) | EGSDE · Cat->Dog | L2 distance: 33.39 -> 65.61 | Hurt |
Closest papers by meaning
- Unpaired Image-to-Image Translation via Neural Schrödinger Bridge (ICLR 2024 · similarity 0.88)
- SCott: Accelerating Diffusion Models with Stochastic Consistency Distillation (AAAI 2025 · similarity 0.87)
- Deterministic Image-to-Image Translation via Denoising Brownian Bridge Models with Dual Approximators (CVPR 2025 · similarity 0.87)
- Simpler Diffusion: 1.5 FID on ImageNet512 with Pixel-space Diffusion (CVPR 2025 · similarity 0.87)
- Plug-and-Play Diffusion Distillation (CVPR 2024 · similarity 0.87)
- Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference (NeurIPS 2024 · 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.