Research map / Probabilistic and scientific modeling / Variational and generative sampling
Stochastic Interpolants for Revealing Stylistic Flows across the History of Art
Pingchuan Ma, Ming Gui, Johannes Schusterbauer, Xiaopei Yang, Olga Grebenkova, Vincent Tao Hu, Björn Ommer
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 |
|---|---|---|---|
| fine-tune Stable Diffusion 1.5 on the art dataset with year prompts | SD 1.5 · Feature synthesis, 16th century | FID: 131.1 -> 111.7 | Improved |
| fine-tune Stable Diffusion 1.5 on the art dataset with year prompts | SD 1.5 · Feature synthesis, 20th century | FID: 180.4 -> 150.7 | Improved |
| fine-tune SDEdit on the art dataset | SDEdit · Time matching, Rococo artworks | Recall@1: 6.3 -> 37.9 | Improved |
| fine-tune SDEdit on the art dataset | SDEdit · Time matching, Realism artworks | Recall@1: 3.7 -> 27.5 | Improved |
| use latent style-space stochastic interpolant bridge instead of fine-tuned SDEdit | SDEdit (fine-tuned) · Time matching, Rococo artworks | Recall@1: 37.9 -> 72.5 | Improved |
| use latent style-space bridge instead of fine-tuned SDEdit | SDEdit (fine-tuned) · Time matching, Impressionism artworks | compactness: 0.272 -> 0.024 | Improved |
| use latent bridge instead of SDEdit | SDEdit · Time matching, Impressionism artworks | triplet consistency (%): 34.5 -> 75.8 | Improved |
| use latent bridge instead of RF-inversion | RF-inversion · Time matching, Rococo artworks | triplet consistency (%): 83.7 -> 84.8 | Mixed |
| use the model with year conditioning instead of text-conditioned SDXL | SDXL · Feature synthesis, 19th century | FID: 114.8 -> 72.0 | Improved |
| use the model instead of FLUX.1-schnell | FLUX.1-schnell · Feature synthesis, 20th century | FID: 126.1 -> 78.9 | Improved |
| set integration threshold t_th too low (ablation) | stochastic interpolant bridge · qualitative | content preservation | Hurt |
| raise integration threshold t_th toward 1 (ablation) | stochastic interpolant bridge · qualitative | semantic identity | Improved |
Closest papers by meaning
- Gender Artifacts from Art History to Text-to-Image Generation (NeurIPS 2026 · similarity 0.89)
- ArtBank: Artistic Style Transfer with Pre-trained Diffusion Model and Implicit Style Prompt Bank (AAAI 2024 · similarity 0.87)
- Stylebreeder: Exploring and Democratizing Artistic Styles through Text-to-Image Models (NeurIPS 2024 · similarity 0.86)
- Alchemist: Turning Public Text-to-Image Data into Generative Gold (NeurIPS 2025 · similarity 0.86)
- Z*: Zero-shot Style Transfer via Attention Reweighting (CVPR 2024 · similarity 0.85)
- Conditional Balance: Improving Multi-Conditioning Trade-Offs in Image Generation (CVPR 2025 · similarity 0.85)
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.
