Research map / Image generation / Text-to-image diffusion
Learn Once, Edit Anywhere: Visual Direction Transfer for Diffusion Models
Yusuf Dalva, Hidir Yesiltepe, Pinar Yanardag
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 L_latent to semantic loss (ablation) | ViDiT · Ablation (face directions) | DINO: 0.832 -> 0.891 | Improved |
| add L_latent to semantic loss (ablation) | ViDiT · Ablation (face directions) | LPIPS: 0.121 -> 0.093 | Improved |
| add L_latent to semantic loss (ablation) | ViDiT · Ablation (face directions) | CLIP-T: 0.409 -> 0.406 | No clear effect |
| increase training pairs from 10 to 1000 (ablation) | ViDiT · Ablation (face directions) | DINO: 0.869 -> 0.891 | Improved |
| increase training pairs from 100 to 1000 (ablation) | ViDiT · Ablation (face directions) | LPIPS: 0.136 -> 0.093 | Improved |
| increase training pairs from 10 to 100 (ablation) | ViDiT · Ablation (face directions) | CLIP-T: 0.403 -> 0.425 | Improved |
| increase training pairs from 10 to 100 (ablation) | ViDiT · Ablation (face directions) | LPIPS: 0.098 -> 0.136 | Hurt |
| use DDPM inversion instead of DDIM inversion (ablation) | ViDiT real image editing · real images | identity preservation (qualitative) | Improved |
| apply edit at 0.4T instead of full T (ablation) | ViDiT · generated faces | edit strength vs identity (qualitative) | Mixed |
| replace text prompt with learned visual direction vs InstructPix2Pix | ViDiT · Text-driven editing comparison (200 pairs, Asian and Smile) | LPIPS: 0.059 -> 0.030 | Improved |
| use ViDiT instead of Prompt2Prompt | ViDiT · Text-driven editing comparison (200 pairs, Asian and Smile) | DINO: 0.867 -> 0.929 | Improved |
| use ViDiT instead of Concept Sliders | ViDiT · Text-driven editing comparison (200 pairs, Asian and Smile) | CLIP-T: 0.325 -> 0.407 | Improved |
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.
- InstructPix2Pix — InstructPix2Pix: Learning To Follow Image Editing Instructions (CVPR 2023) · better on 4 of 5 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
- Localizing and Editing Knowledge In Text-to-Image Generative Models (ICLR 2024 · similarity 0.92)
- Textualize Visual Prompt for Image Editing via Diffusion Bridge (AAAI 2025 · similarity 0.91)
- RelationAdapter: Learning and Transferring Visual Relation with Diffusion Transformers (NeurIPS 2025 · similarity 0.90)
- NoiseCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directions in Diffusion Models (CVPR 2024 · similarity 0.90)
- An Item Is Worth a Prompt: Versatile Image Editing with Disentangled Control (AAAI 2025 · similarity 0.90)
- Progressive Text-to-Image Diffusion with Soft Latent Direction (AAAI 2024 · similarity 0.90)
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.
