Upscale many images in one run, on your own GPU. Pick the scale and tier once; every image is enlarged with the same settings and returned in a ZIP. The per-item time estimate tells you what a 200-photo batch will really cost before you are ten minutes into it.
How it works
- Drop an image or video, or many images, or a ZIP, for a batch.
- Pick the scale and tier. The estimate tells you how long an export will take before you commit.
- Download the enlarged result, or the whole batch as a ZIP.
Questions
- How many files can one batch take?
- Up to 500 files or 2 GB of input per run, dropped as loose files or a single ZIP. Items process one at a time on your device, nothing is uploaded, and the progress strip shows a per-item ETA. Bigger jobs: split into two drops.
- What do I get back?
- One ZIP: every processed file, named exactly like the single-file download would be, plus _report.txt and _report.json listing each input and what happened to it, including any that failed and why. A cancelled batch still downloads the items finished so far, with the report saying so.
- Does my file get uploaded?
- No. The models and all processing run inside your browser on your own device. You can verify this yourself: open your browser’s DevTools, watch the Network tab, and process a file: there are no uploads. After the first visit most tools also work fully offline.
- Which tier should I use?
- Balanced is right for almost everything: it is a 5 MB neural network that runs in seconds. Maximum uses a larger model and is worth it for print or archival work. Fast is instant and fine for quick looks.
- Does it work on old or blurry photos?
- Upscaling sharpens edges and adds plausible detail, and the balanced tier includes denoising. It cannot recover detail that was never captured, so faces enlarged from very small sizes will look smoothed, not restored.