Topaz Gigapixel
Topaz Labs' desktop image upscaler, now subscription-only, with local and unlimited cloud rendering on Windows and macOS
Topaz Labs' desktop image upscaler, now subscription-only, with local and unlimited cloud rendering on Windows and macOS
Fast-read signals for fit, pricing, and trust.
Large, detailed enlargements from small or degraded files are what Topaz Gigapixel is for, and the photographer, print maker or restorer it suits has to accept a subscription that ends access when it lapses. Personal's commercial use stops at organisations of $1M annual revenue, the $19 and $54 monthly prices are 12-month commitments that cannot be cancelled mid-term without a fee, and Topaz's documentation rules out a trial, offering a 2-day money-back window instead.
After subscribing, the app installs from your Topaz account and activates by signing in with the checkout email. Import one image or a batch, choose a core or generative model, set the scale or target dimensions, compare the preview, render locally or send the job to the cloud, and export, or do the same from inside Photoshop or Lightroom Classic. The documentation gives a ceiling of 32,000 pixels on the longest side; Topaz's official pages state two different maximum scale factors.
The subscription terms deserve a careful read. 'Annual, Billed Monthly' is a 12-month commitment, and Topaz's refund policy says annual-commitment subscriptions cannot be cancelled or refunded during the term without an additional fee, that cancellation should be made at least 24 hours before renewal, and that cancelling ends access to the app. The app re-authenticates monthly and works up to one month offline.
Gigapixel is also sold inside the Topaz Studio bundle with Topaz Video, Photo, Mobile, Bloom and Web, at $69 a month or $399 billed annually; the bundle's annual-commitment-billed-monthly price is printed two ways on the product page. Topaz lists macOS 12 Monterey as supported on the product page and macOS 13 as the minimum in its system requirements, so check the requirements page for your Mac rather than relying on a version number.
Quick fit check against how you actually work.
Aggregate community sentiment, summarized — not native reviews.
Fifteen independent accounts, commenting on ten YouTube videos between May 2023 and May 2026, describe Topaz Gigapixel from three angles. The subscription switch drew the most heat: five accounts on four videos after the September 2025 relaunch object to subscription-only licensing, several saying they stopped updating or would choose a one-time-licence tool, while one Reddit user takes the counter-view that a subscription can be taken up and cancelled as needed. Seven accounts on four videos report slow local processing: several say a single upscale took minutes to hours and some tie it to older or weaker hardware, two of them adding that a high-end GPU or an M2 Pro Mac runs it acceptably. On output quality there is genuine disagreement: four accounts on four videos see smeared, plastic or cartoony textures or artifacts, while three others are positive, one about clean huge enlargements, one tying clean results to modest scale factors on good source files and one crediting what they call the Wonder 3 update.
What you can actually do with this tool.
Local core models handle everyday upscaling; the generative Wonder, Redefine and Recover models reconstruct detail and faces and carry per-model size limits (Wonder, for example, 128MP input and output). Batch processing is supported.
Both plans include unlimited local rendering and, in the vendor's word, 'Unlimited cloud image rendering' at no extra cost, with two concurrent cloud jobs on Personal and four on Pro. Cloud rendering applies to the generative models, is not available when Gigapixel runs as a plugin, and results must be downloaded within 7 days; local rendering keeps files on your computer.
The app runs standalone or from inside Adobe Photoshop and Lightroom Classic, except that on Windows Snapdragon machines the Photoshop plugins are not supported.
16 GB of RAM everywhere, then platform-specific minimums: Apple Silicon or a 2016-or-newer Intel Mac on macOS, with the generative models needing cloud rendering on Intel Macs, and Windows 10 or 11 with more disk space on Intel, AMD or Snapdragon PCs. Linux, virtual machines and eGPUs are unsupported.
Pricing tiers and what's included in each.
Common questions about this tool, answered.
Not on a trial: Topaz's documentation says there is no trial mode for Gigapixel. The substitute is a 2-day money-back guarantee, a full refund for a request within 2 days of purchase and none after. Because the 'Annual, Billed Monthly' plans are a 12-month commitment that cannot be cancelled during the term without an additional fee, the no-commitment monthly Personal plan at $29 is the lower-risk way to evaluate it.
Users disagree. Some describe plastic or cartoony textures or artifacts. Others are positive: one found huge enlargements free of artefacts, another links clean output to small scale factors from a good source file, and a third says the Wonder 3 update, as they call it, made it worth keeping. Results vary with the source image, the scale and the model, so compare the preview before rendering.
Every platform needs 16 GB of RAM. macOS needs an Apple Silicon or 2016-or-newer Intel Mac and 30 GB of storage, and Intel Macs must use cloud rendering for the generative models; Windows needs 55 GB on the C: drive, Windows 10 (latest) or later on Intel or AMD, and Windows 11 on Snapdragon, where the Photoshop plugins are not supported. The minimum macOS version is stated as 12 on the product page and 13 in the requirements documentation, so read the requirements page for your Mac. Meeting the minimum is not the same as speed: users describe slow local processing, some blaming weaker hardware, and cloud rendering is the included alternative.