PiSA-SR — Image Upscaler AI Model
PiSA-SR upscales images 2× or 4× and rebuilds fine detail, with pixel-level and semantic enhancement on separate controls — the sharpest of the on-device upscalers, and the slowest of them: about 50 seconds.
Also runs in Compare Three Image Upscalers · One Prompt, Three 4K Images
What it does
PiSA-SR turns a blurry, low-resolution image into a sharp 2x or 4x one. It runs two modules: one at the pixel level to clear noise and artifacts, one at the semantic level to add perceptual detail.
The balance between them is yours at inference time — lean toward faithful cleanup or toward richer invented detail — from a single trained model, with no retraining either way.
When to use it
- A low-resolution image from the web, brought up to a usable size
- A screenshot or video thumbnail, upscaled and clarified
- An old camera or phone photograph with age and degradation on it
- A file compressed to save space, with quality recovered
- Real-world degradation — blur, noise and compression at once
Details
- CategoryImage Upscaler
- Year2025
- LicenseApache-2.0
Compliance & Provenance
- ProviderOpen-source · Specialized
- EU AI Act RiskMinimal Risk
- Art. 50 TransparencyNot applicable
Inputs & Outputs
- ImageInputimage
Low-resolution source image
- ImageOutputimage
Upscaled high-resolution image
Parameters
- scale_factorintegerdefault 2options: 2, 4
Upscale ratio (2/4)
Enlargement factor. 2 (default) — fastest, good for moderate upscaling. 4 — strongest enhancement but much slower and more VRAM-intensive.