Research map / Image generation / Editing and style transfer
Fix the Noise: Disentangling Source Feature for Controllable Domain Translation
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 |
|---|---|---|---|
| Decrease noise interpolation weight α from 1 to 0 | FixNoise · FFHQ → MetFaces | FID: 40.37 -> 19.68 | Helped |
| Decrease noise interpolation weight α from 1 to 0 | FixNoise · FFHQ → AAHQ | FID: 31.65 -> 5.10 | Helped |
| Decrease noise interpolation weight α from 1 to 0 | FixNoise · LSUN Church → WikiArt Cityscape | FID: 27.64 -> 11.49 | Helped |
| Decrease noise interpolation weight α from 1 to 0 | FixNoise · FFHQ → MetFaces | LPIPS: 0.412 -> 0.536 | Hurt |
| Decrease noise interpolation weight α from 1 to 0 | FixNoise · FFHQ → AAHQ | LPIPS: 0.316 -> 0.510 | Hurt |
| Decrease noise interpolation weight α from 1 to 0 | FixNoise · LSUN Church → WikiArt Cityscape | LPIPS: 0.521 -> 0.679 | Hurt |
| Apply feature matching in intermediate feature space H rather than RGB or image space (ablation) | FixNoise · FFHQ → MetFaces | Target-domain adaptation | Helped |
| Fix the noise input to an anchor point for feature matching instead of matching across the entire noise space (ablation) | FixNoise · Cross-domain feature control | Controllability | Helped |
| Use randomly sampled noise for feature matching rather than a fixed anchor (ablation) | FixNoise · Cross-domain feature control | Target-feature learning | Hurt |
Closest papers by meaning
- A Unified Feature Disentangler for Multi-Domain Image Translation and Manipulation (NeurIPS 2018 · similarity 0.89)
- Benign Examples: Imperceptible Changes Can Enhance Image Translation Performance (AAAI 2020 · similarity 0.88)
- Image-to-image translation for cross-domain disentanglement (NeurIPS 2018 · similarity 0.88)
- Multi-mapping Image-to-Image Translation via Learning Disentanglement (NeurIPS 2019 · similarity 0.87)
- Image is All You Need to Empower Large-scale Diffusion Models for In-Domain Generation (CVPR 2025 · similarity 0.86)
- Few-shot Hybrid Domain Adaptation of Image Generator (ICLR 2024 · 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.