Research map / Image generation / Super-resolution
PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models
Sachit Menon, Alexandru Damian, Shijia Hu, Nikhil Ravi, Cynthia Rudin
Check 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 |
|---|---|---|---|
| Use PULSE instead of FSRGAN | Face super-resolution · CelebA HQ, 8×, 128 × 128 | MOS: 2.92 -> 3.60 | Improved |
| Use PULSE instead of FSRNet | Face super-resolution · CelebA HQ, 8×, 128 × 128 | MOS: 2.77 -> 3.60 | Improved |
| Use PULSE instead of bicubic upscaling | Face super-resolution · CelebA HQ, 8×, 128 × 128 | MOS: 1.34 -> 3.60 | Improved |
| Use PULSE instead of bicubic upscaling | Face super-resolution · CelebA HQ, 64×, 1024 × 1024 | NIQE: 7.06 -> 2.47 | Improved |
| Use PULSE instead of nearest-neighbor upscaling | Face super-resolution · CelebA HQ, 64×, 1024 × 1024 | NIQE: 12.48 -> 2.47 | Improved |
| Change the hypersphere radius used for latent search (ablation) | PULSE with StyleGAN-FFHQ · FairFace, ×64 | success rate | No clear effect |
| Relax StyleGAN's constraint by allowing its 18 latent vectors to vary while encouraging them to be approximately equal (ablation) | PULSE with StyleGAN-FFHQ · Face super-resolution | realism | Improved |
| Loosen StyleGAN's latent-space constraints further (ablation) | PULSE with StyleGAN-FFHQ · Face super-resolution | realism | Hurt |
| Use PULSE with bicubic downscaling as a stand-in when the true degradation includes noise or blur | PULSE · Unknown degradation operators | downscaling correctness | Improved |
Later papers that compared with this method
10 later accepted papers report PULSE as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- NaP-Control: Navigating Diffusion Prior for Versatile and Fast Character Control (ECCV 2026 · better on 5 of 11 shared results)
- One-for-All: Towards Universal Domain Translation With a Single StyleGAN (TPAMI 2025 · better on 3 of 3 shared results)
- Steered Diffusion: A Generalized Framework for Plug-and-Play Conditional Image Synthesis (ICCV 2023 · better on 1 of 2 shared results)
- Robust Unsupervised StyleGAN Image Restoration (CVPR 2023 · better on 1 of 1 shared result)
- VQFR: Blind Face Restoration with Vector-Quantized Dictionary and Parallel Decoder (ECCV 2022 · better on 5 of 6 shared results)
- Faithful Extreme Rescaling via Generative Prior Reciprocated Invertible Representations (CVPR 2022 · better on 3 of 3 shared results)
- RestoreFormer: High-Quality Blind Face Restoration From Undegraded Key-Value Pairs (CVPR 2022 · better on 1 of 1 shared result)
- GLEAN: Generative Latent Bank for Image Super-Resolution and Beyond (TPAMI 2022 · better on 1 of 1 shared result)
2 more not listed.
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.
- FSRNet — FSRNet: End-to-End Learning Face Super-Resolution With Facial Priors (CVPR 2018) · better on 1 of 1 shared result
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
- Low-Res Leads the Way: Improving Generalization for Super-Resolution by Self-Supervised Learning (CVPR 2024 · similarity 0.88)
- When Latents Forget Pixels: Restoring Fidelity in Diffusion Transformer Super-Resolution (NeurIPS 2026 · similarity 0.88)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.88)
- Latent Space Super-Resolution for Higher-Resolution Image Generation with Diffusion Models (CVPR 2025 · similarity 0.88)
- Towards Realistic Data Generation for Real-World Super-Resolution (ICLR 2025 · similarity 0.87)
- Arbitrary-Scale Image Generation and Upsampling using Latent Diffusion Model and Implicit Neural Decoder (CVPR 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.
