Research map / Image generation / Editing and style transfer
A Diffusion Model Translator for Efficient Image-to-Image Translation
Mengfei Xia, Yu Zhou, Ran Yi, Yong‐Jin Liu, Wenping Wang
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 DMT translator on top of Pix2Pix | Pix2Pix · Stylization | FID: 19.14 -> 10.81 | Improved |
| add DMT translator on top of TSIT | TSIT · Segmentation-to-image | FID: 40.59 -> 36.78 | Improved |
| add DMT translator on top of Pix2Pix | Pix2Pix · Colorization | FID: 17.50 -> 17.44 | No clear effect |
| add DMT translator on top of TSIT | TSIT · Sketch-to-image | FID: 76.80 -> 74.37 | Improved |
| use DMT with TSIT instead of Palette | Palette · Inference time | seconds: 21.63s -> 0.48s | Improved |
| use DMT instead of SPADE | SPADE · Segmentation-to-image | FID: 66.55 -> 36.78 | Improved |
| use DMT instead of QMUPD | QMUPD · Stylization | FID: 12.81 -> 11.01 | Improved |
| increase preset timestep t from 0 to 200 (ablation) | TSIT-DMT · Segmentation-to-image | FID: 59.51 -> 36.78 | Improved |
| increase preset timestep t from 0 to 200 (ablation) | TSIT-DMT · Segmentation-to-image | SSIM: 0.522 -> 0.446 | Hurt |
| increase preset timestep t to 900 (ablation) | TSIT-DMT · Segmentation-to-image | FID: 36.78 -> 213.81 | Hurt |
| increase preset timestep t from 0 to 20 (ablation) | TSIT-DMT · Stylization | FID: 22.38 -> 11.01 | Improved |
| vary t near t* of 200 (ablation) | TSIT-DMT · CelebA-HQ segmentation | FID: 36.78 vs 36.04 at t=220 | 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.
- TSIT — TSIT: A Simple and Versatile Framework for Image-to-Image Translation (ECCV 2020) · better on 5 of 6 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
- High-Fidelity Diffusion-Based Image Editing (AAAI 2024 · similarity 0.89)
- Simpler Diffusion: 1.5 FID on ImageNet512 with Pixel-space Diffusion (CVPR 2025 · similarity 0.89)
- Denoising Diffusion Bridge Models (ICLR 2024 · similarity 0.89)
- Diffusion Bridge Implicit Models (ICLR 2025 · similarity 0.89)
- Exploring the Design Space of Diffusion Bridge Models (NeurIPS 2025 · similarity 0.89)
- Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference (NeurIPS 2024 · 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.
