What we are building
PicToolBox is a collection of narrow, single-purpose image tools. Each one does one job, does it immediately, and does it without asking you to create an account first. There is no dashboard, no file manager and no subscription wall in front of basic functionality.
The toolkit currently covers eight jobs: removing shadows across a scene, cleaning up background shadows on product shots, removing a single colour, removing white backgrounds, general background removal, unpixelating blocky images, blurring backgrounds, and resizing to an exact size. More are added as we find genuinely common tasks that existing tools handle badly.
Why everything runs locally
The single most consequential design decision on this site is that your images are never uploaded. Every operation is performed by JavaScript running on your own device, using the browser's Canvas API.
That choice has three consequences worth stating plainly:
- Privacy is structural, not a promise. We cannot leak, sell or lose your photos because we never receive them. This is not a policy we could change without rebuilding the product from scratch.
- Speed. There is no queue, no upload time and no server round-trip. A 12 MP photo is usually processed in well under a second.
- Honest limitations. Local processing cannot use large AI models. So the tools do deterministic image processing, and we tell you exactly what that means for each one — including what it will not do. The unpixelate tool, for instance, does not claim to reconstruct a face from twelve pixels, because nothing running privately on your laptop can actually do that.
How the site is funded
Credits, not ads and not subscriptions. Previews are free and unlimited; a clean full-size download costs one credit, sold in small packs that never expire. There is no recurring charge to forget about. That is the whole business model, and there is no data brokerage in it — which is precisely why the local-only architecture is worth the limitations it imposes.
How we test
Every tool is checked against real photographs before it ships: portraits in window light, product shots on a paper sweep, phone photos with flash falloff, scanned documents, heavily compressed JPEGs and low-resolution screenshots. Where a tool performs badly on a category of image, that limitation is documented on the tool page rather than hidden.
Get in touch
Found a bug, a photo that produces a bad result, or a job you wish existed here? Tell us. Reports that include the image type involved are the most useful ones.
Last updated: October 2026.