Research map / Image generation / Image restoration
LATINO-PRO: LAtent consisTency INverse sOlver with PRompt Optimization
Alessio Spagnoletti, Jean Prost, Andrés Almansa, Nicolas Papadakis, Marcelo Pereyra
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 prompt optimization (SAPG) to LATINO | LATINO · AFHQ-512 val Gaussian deblur sigma=5 | FID: 20.03 -> 18.37 | Improved |
| add prompt optimization (SAPG) to LATINO | LATINO · AFHQ-512 val SR x16 | FID: 42.14 -> 30.40 | Improved |
| add prompt optimization (SAPG) to LATINO | LATINO · AFHQ-512 val SR x16 | PSNR: 20.05 -> 21.52 | Improved |
| add prompt optimization (SAPG) to LATINO | LATINO · AFHQ-512 val Gaussian deblur sigma=5 | PSNR: 26.25 -> 26.82 | Improved |
| add prompt optimization (SAPG) to LATINO | LATINO · FFHQ-512 val Gaussian deblur sigma=3 | FID: 33.94 -> 31.98 | Improved |
| add prompt optimization (SAPG) to LATINO | LATINO · FFHQ-512 val SR x8 | FID: 37.13 -> 40.95 | Hurt |
| add prompt optimization (SAPG) to LATINO | LATINO · FFHQ-512 val Motion deblur | PSNR: 26.88 -> 27.14 | Improved |
| use LATINO (8 NFE) instead of P2L (2000 NFE) | P2L · AFHQ-512 val Gaussian deblur sigma=5 | FID: 85.80 -> 20.03 | Improved |
| use LATINO (8 NFE) instead of TReg (200 NFE) | TReg · AFHQ-512 val SR x16 | FID: 37.13 -> 42.14 | Hurt |
| use LATINO (8 NFE) instead of P2L (2000 NFE) | P2L · FFHQ-512 val SR x8 | PSNR: 28.55 -> 26.22 | Hurt |
| use LATINO-PRO instead of P2L | P2L · FFHQ-512 val Gaussian deblur sigma=3 | PSNR: 26.97 -> 29.11 | Improved |
| use LATINO-PRO instead of LDPS | LDPS · FFHQ-512 val Motion deblur | FID: 58.66 -> 27.80 | Improved |
Closest papers by meaning
- Prompt-tuning Latent Diffusion Models for Inverse Problems (ICML 2024 · similarity 0.91)
- Solving Inverse Problems with FLAIR (NeurIPS 2025 · similarity 0.88)
- SITCOM: Step-wise Triple-Consistent Diffusion Sampling For Inverse Problems (ICML 2025 · similarity 0.88)
- Regularization by Texts for Latent Diffusion Inverse Solvers (ICLR 2025 · similarity 0.87)
- Adapt and Diffuse: Sample-adaptive Reconstruction via Latent Diffusion Models (ICML 2024 · similarity 0.87)
- Provably Robust Score-Based Diffusion Posterior Sampling for Plug-and-Play Image Reconstruction (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.
