Overview
Gigapixel exists for the situation every archive and every print shop runs into: the only copy of an image is too small for the output it has to fill. Conventional resampling makes a soft, larger version of the same picture. This rebuilds plausible detail at the new size, which is a different result entirely.
Several models ship in the build and choosing between them is most of the skill. Standard covers general photography, high fidelity preserves fine texture in a clean source, low resolution handles small web images, compressed targets material wrecked by heavy encoding, art is tuned for illustration and flat colour, and the face recovery pass rebuilds features on subjects too small for the main model to hold.
The application runs as a batch processor as much as a single image tool. Point it at a folder, pick a scale factor or a target output size, set the model and the sharpening and noise controls, and leave it. Output goes to standard image formats with colour profiles preserved, and there is a plugin path so it can be called from an editor without leaving the document.
What it does well
Model selection
Separate models for general photography, high fidelity sources, small web images, compression damaged files and illustration.
Face recovery
A dedicated pass that rebuilds facial detail on subjects rendered too small for the main model to resolve, with a strength control.
Batch processing
Run a folder through one setting set with output naming, format and scale rules, and leave it working.
Comparison view
Split and side by side previews at pixel level so a model choice is judged on the actual result rather than a thumbnail.
Editor plugin
Call it from a host image editor and return the enlarged result into the open document.
Changes in this build
- New recovery model that holds edges better on text and line art.
- Face recovery strength is now adjustable rather than on or off.
- GPU processing rebalanced, cutting time on large batches.
- Preview generation made faster so model comparison is interactive.
- Colour profile handling fixed for wide gamut source files.
What is in the package
- Gigapixel AI Pro installer, 64-bit
- Full bundled model set
- Editor plugin components
- Batch processing presets
- Sample images for model comparison
System requirements
| Processor | Intel or AMD 64-bit with AVX2 |
| Memory | 16 GB recommended for large enlargements |
| Graphics | GPU with 6 GB VRAM, more for batches at high scale factors |
| Storage | 6 GB free plus output space |
| Display | 1920 x 1080 minimum |
| System | Windows 10 or Windows 11, 64-bit |
Installation
- Extract the archive to a local drive.
- Run the installer and let the model files unpack, which takes longer than the application itself.
- Launch once and let it detect the GPU before processing anything.
- Register the editor plugin from the preferences panel if you want it in a host application.
- Block the updater so the build and the models stay aligned.
Before you start
Model files are large. Moving the install folder afterwards breaks model loading.
Very high scale factors invent detail. Judge the result at print size, not on screen at full zoom.
GPU memory is the limiting factor on batch throughput, lower the batch concurrency if it stalls.
Questions about this title
How far can it enlarge?
Up to six times linear in one pass, with quality depending far more on the source than the factor.
Does it work offline?
Yes, the models are bundled and run on the machine.
Can it process a folder?
Yes, batch mode with output naming and format rules is built in.
Does it keep colour profiles?
Yes, embedded profiles are preserved through processing.
About this listing
This entry was checked on a clean install of Windows 10 / 11 (64-bit) before it was published, and it is rechecked whenever the package is rebuilt. The figures on this page come from the site index rather than from the publisher, so the download count is what people here have actually pulled.
If a mirror stops answering, use the contact form and it gets replaced. Reports about a specific title are handled faster than general messages because they name the file, the mirror and the point at which the transfer stopped.
Requests for a different version, a different language build or an older release go on the requests page. Older versions of a title are usually still held even when only the current one is listed.