Short version: Topaz Gigapixel AI is a single-purpose desktop app that enlarges images by up to 16x using AI models trained to reconstruct plausible detail instead of stretching pixels. If your job regularly involves turning small, soft, or compressed source files into print-sized output, it earns its place; if you only occasionally resize an image, the recurring subscription is harder to justify.
What the software actually does, and why the model choice matters
Gigapixel is not a general photo editor. It does one thing: it increases the pixel dimensions of an image while trying to invent detail that looks like it belonged there. Conventional resampling (bicubic and its relatives) can only average existing pixels, so enlargements go soft and mushy. Gigapixel instead runs the image through neural models that were trained on large sets of high- and low-resolution image pairs, so it can hallucinate texture, edges, and structure that plausibly match the subject.
The practical consequence of that approach is that which model you pick matters as much as the software itself. Topaz ships several enhancement models, and they are specialised on purpose. On the current version the app exposes models including Standard and Standard MAX, High Fidelity, Low Resolution, Text & Shapes, Art & CG, Recover, Redefine, and Face Recovery, plus generative models branded Wonder 2 and Wonder 3 (with an earlier Wonder 1 available through cloud access only). That spread exists because a heavily compressed phone snapshot, a clean high-resolution studio file, a line-art illustration, and a scanned archival print all fail in different ways when enlarged. High Fidelity aims to stay faithful to a good original; Low Resolution and Recover are built for salvaging weak sources; Art & CG and Text & Shapes avoid smearing the hard edges that photographic models tend to soften. The generative Wonder models push further into inventing detail, which is powerful for badly degraded input but also the point where output can drift away from what was really in the frame.
That tradeoff is the single most important thing to understand before buying. AI upscaling is a form of educated guessing. On a decent source it is genuinely convincing; on a tiny, mangled JPEG the software will confidently manufacture detail that was never captured, and faces, text, and fine repeating patterns are where invented detail is easiest to spot. Gigapixel gives you the tools to manage this, but it does not remove your responsibility to check the result.
Face recovery, batch processing, and the workflow features
A dedicated Face Recovery model addresses the most common failure mode of upscalers: eyes, mouths, and skin turning waxy or distorted when a portrait is enlarged from a small crop. This is the feature most likely to matter if you shoot events, do restoration work, or pull faces out of group shots. It is not magic on very low-resolution faces, but it meaningfully reduces the plastic look that plagues naive enlargement.
Batch processing is the other feature that separates a professional tool from a novelty. Upscaling is compute-heavy, so being able to queue a folder of images with a chosen model and let the machine grind through them unattended is what makes Gigapixel usable on a real shoot rather than one hero image at a time. The app supports unlimited local rendering, and offers cloud rendering as an alternative when your own hardware is the bottleneck. There is also a Lossless Zoom feature for inspecting and cropping without throwing away quality during evaluation.
It runs on Windows 10 and 11 and on macOS 12 through macOS 15, on both Apple Silicon and Intel Macs, though Topaz notes limitations for Intel Mac users on the generative models. That caveat is worth taking seriously: the newest, most detail-inventing models lean on modern GPU hardware, so an older machine may be limited to the more conservative models.
Fitting it into an existing edit
Most professionals do not want a standalone step bolted onto the side of their workflow, so Gigapixel integrates as a plug-in for Adobe Photoshop (through the Automate menu, up to a 30,000px dimension limit) and for Lightroom Classic. The plug-in route means you can upscale as part of a round-trip rather than exporting, re-importing, and losing your place. One documented limitation: cloud rendering is not available when running in plug-in mode, so heavy generative jobs may still want the standalone app. You can compare this against other options in the same space on the AI photo enhancer category.
Where it fits in real work
The clearest wins are in print and large-format output. Preparing a landscape or portrait for a gallery print, a poster, or a trade-show banner is exactly the scenario Gigapixel was designed for, because viewers stand back from big prints and the eye forgives invented detail more readily than it does on a screen at 100%. Archival and restoration work is a second strong fit: old scans, salvaged family photos, and low-resolution digital archives can be brought up to a usable size, with the Recover and Face Recovery models doing the heavy lifting.
A third, more mundane use is rescuing the file you cannot re-shoot: the only surviving image of a product, an event, or a person, delivered at a size too small for the intended use. Here Gigapixel is a genuine problem-solver even when the result is imperfect, because the alternative is having nothing usable at all. Conversely, if you routinely capture high-resolution files and rarely need to exceed them, you may find you reach for this tool only occasionally.
What it costs
Topaz now sells Gigapixel as a subscription rather than the perpetual, one-time licence it was historically known for. Buyers who remember the old buy-once model should note that shift, because it changes the long-term cost calculation considerably. Current standalone pricing is published as follows.
| Plan | Monthly | Annual (billed monthly) | Annual (lump sum) |
|---|---|---|---|
| Personal | $29/mo | $19/mo | $149/yr |
| Pro | — | $54/mo | $499/yr |
The Personal plan is licensed for limited use inside organisations under $1M in annual revenue, while the Pro plan carries full commercial-use rights. Gigapixel is also bundled into Topaz Studio (which the vendor lists at $399/year lump sum for Personal, alongside monthly options), which packages it with the company's other photo and video apps. Whether the bundle is worthwhile depends entirely on whether you also need those other tools. The pricing page does not advertise a free trial or a perpetual licence, so treat this as a recurring cost, not a one-off. Prices can change, so confirm current figures on the vendor's site before committing.
Strengths worth paying for
- Detail reconstruction that clearly outperforms conventional resampling on suitable source images, which is the whole reason to buy it.
- A range of specialised models, so you can match the engine to the subject instead of forcing one algorithm onto everything.
- A dedicated Face Recovery model that targets the most visible weakness of enlargement.
- Batch processing plus unlimited local rendering, making high-volume jobs practical.
- Photoshop and Lightroom Classic plug-ins that keep it inside an existing editing workflow.
- Cross-platform support for current Windows and macOS, including Apple Silicon.
Honest drawbacks
- It is now subscription-only; there is no publicly documented perpetual licence or one-time purchase, which is a meaningful change from its earlier reputation.
- Results on very small or heavily compressed sources involve genuine invention of detail, and faces, text, and fine patterns can look artificial if you push too hard.
- It is a single-purpose tool. It upscales; it is not a substitute for a full editor, and you pay for a narrow (if deep) capability.
- The most capable generative models depend on modern GPU hardware, with documented limitations for Intel Macs.
- Cloud rendering is unavailable in plug-in mode, so the heaviest generative jobs may need the standalone app.
Who should buy it, and who should skip it
Gigapixel is a well-focused tool that does its narrow job better than general-purpose resampling, and the recent move to a subscription is the main thing prospective buyers need to weigh. If you enlarge images for print, restoration, or archival work often enough that the monthly or annual fee is a rounding error against the value of the output, it is an easy recommendation, particularly for portrait-heavy work where Face Recovery pays off. If your enlargements are occasional, or if you need broad editing rather than upscaling, the recurring cost is harder to defend and you should evaluate whether a bundle or an alternative better matches your volume. Either way, judge the tool on your own representative files during whatever evaluation the vendor currently offers, since upscaling quality depends heavily on the specific images you feed it. For broader comparison, browse the full tools directory or read related coverage on the blog.
Common questions
How much can Topaz Gigapixel AI enlarge an image?
The current version supports upscaling up to 16x. The practical ceiling for good-looking results is lower and depends on the quality of your source file and the model you choose.
Is there a one-time purchase or free trial?
The vendor's pricing page does not list a perpetual licence or a free trial for the current release; Gigapixel is offered on a subscription basis, with Personal plans starting at $149/year (lump sum) and Pro at $499/year.
Can I use it inside Photoshop or Lightroom?
Yes. Gigapixel integrates as a plug-in for Adobe Photoshop (via the Automate menu, up to a 30,000px limit) and for Lightroom Classic. Cloud rendering is not available when running in plug-in mode.
Which model should I use?
It depends on the image. High Fidelity suits strong originals, Recover and Low Resolution target weak sources, Art & CG and Text & Shapes preserve hard edges in illustrations and text, Face Recovery targets portraits, and the generative Wonder models invent the most detail for badly degraded input.
Does it work on Mac and Windows?
Yes. It supports Windows 10 and 11 and macOS 12 through macOS 15 on both Apple Silicon and Intel Macs, though Topaz notes limitations for Intel Mac users on the generative models.
