AI image upscaler · in development
IncreaseAI upscales photos, product shots and illustrations 2× or 4×, then sharpens and denoises them, so small images become print- and zoom-ready.
How it works
Upload a JPEG, PNG or WebP: an old scan, a small product shot, a screenshot or a piece of artwork.
Choose the scale factor and a model tuned for photos or for anime and illustration. Optionally add denoise or sharpening.
Check the result side by side with the original, then export as PNG, JPEG or WebP at the new size.
Features
Resizing in an image editor adds pixels by averaging their neighbours. IncreaseAI is being built to add the detail those pixels should have had.
Increase pixel dimensions by two or four times along each side, so a 1000 px image becomes 2000 or 4000 px wide.
A trained model predicts plausible edges and texture instead of just stretching pixels, which avoids the soft, smeared look of simple resizing.
Reduce sensor noise and blocky compression artifacts before upscaling, so the model does not enlarge the damage.
Adjustable sharpening on the output, with a strength setting so edges get crisper without halos.
Separate models for natural photos and for flat-colour line art, because they fail in very different ways.
Export at the pixel size you need for a target print size and DPI, in a lossless or web-friendly format.
Use cases
Bring small scans, early digital camera shots and images saved from chat apps up to a usable size.
Meet marketplace minimums for zoom-capable product images without reshooting.
Enlarge fan art, wallpapers and sprites while keeping line art clean and flat colours flat.
Get enough pixels for a poster, canvas or photo book at the print size you actually want.
Status: launching soon
We are building the upscaler now. The plan for launch is a free tier for small images with no signup needed. There is no waitlist form; bookmark increaseai.xyz and check back. In the meantime, our guides explain how to get sharp results from any upscaler.
FAQ
Not yet. The product is in development and this site is where it will launch. Until then, the blog covers how upscaling works and how to get the best results with any tool.
Yes, that is the plan: a free tier for small images, with no signup required. Paid options for larger images and batch work may come later; nothing is priced yet.
The width and the height are each multiplied by the factor. A 4× upscale of a 500×400 image produces 2000×1600, which is 16 times as many pixels.
Not literally. A model reconstructs plausible detail based on what similar images look like. For most photos and artwork that looks convincing, but it is not a forensic recovery of the original scene, so do not rely on it for things like reading licence plates.
The plan is JPEG, PNG and WebP for input and output, with AVIF under consideration. See our guide on choosing an export format for the trade-offs.
The intended design is to process images only to produce your result and delete them afterwards, without using them for training. Final details will be published in the privacy policy before launch.
From the blog
5 min read
A practical comparison of JPEG, PNG, WebP and AVIF for upscaled images: compression, transparency, colour depth, compatibility and which to pick for print, web and marketplaces.
5 min read
How to size, crop, clean up and compress product images for marketplaces and online stores so they pass listing rules, support zoom and load fast.
5 min read
Why anime and illustrations need different upscaling than photos, which tools and models suit them, and a step-by-step workflow for clean lines and flat colours.