Topaz Gigapixel Review (Formerly Gigapixel AI)

What Gigapixel is called now, and who owns it

Topaz's pricing page lists the product as Topaz Gigapixel, and its Gigapixel product page calls Gigapixel AI the predecessor (first sold as an AI upscaler in 2019) without describing a rename. Topaz's Photo docs cover the Photo AI to Topaz Photo change and make no statement about Gigapixel. Adobe's announcement of 23 September 2026 says Adobe completed the acquisition of Topaz Labs, that Topaz continues as a standalone brand with apps and models still available as standalone products, and that Topaz technology is available in Adobe Firefly and Photoshop to upscale and enhance images and videos; it does not name which Photoshop feature. That overlap is why Gigapixel vs Photoshop Super Resolution is its own page; the deal itself is covered in the Adobe-Topaz explainer.

What Topaz says Gigapixel does

Topaz's pricing page lists Gigapixel as "image upscaling up to 16x pixels" with Recover, Redefine and Face Recovery image models, and Topaz Photo as the broader app with Wonder, Denoise and Sharpen models (Topaz pricing). Topaz Photo also lists Upscale (Topaz Photo page); how that compares with Gigapixel's upscaler: not found in sources reviewed. Topaz's Gigapixel product page headlines "Upscale almost any image 6x" while its navigation and structured data say "up to 16x"; the page does not reconcile the two (conflicting sources within Topaz), so treat 6x as the figure the page leads with and confirm the cap in the app. The batch documentation says the MAX option upscales to the smaller of 32,000 pixels on the longest side or 6x the original, which is consistent with the 6x headline but is a cap on that one option, not a statement of the maximum factor. The page names Lightroom Classic and Photoshop as plugin hosts and lists model families including Standard, High Fidelity, Low Res, Recover, Redefine, Wonder and Face Recovery. Supported output formats: not found in sources reviewed.

Lightroom Classic, Photoshop and batch handling

Topaz's Lightroom Classic plugin documentation describes the route as Photo > Edit in > Topaz Gigapixel (Studio), then choosing Edit a Copy (sends the image without Lightroom adjustments) or Edit Original (overwrites the selected file). Documented limits: the file is converted to TIFF before it is sent, so a TIFF comes back; sending RAW directly is not recommended because it locks the RAW noise into the file; cloud rendering is not available through the plugin; and only Lightroom Classic is supported, not Lightroom. For a batch, select several images in the Library grid or Filmstrip and use Edit In from the right-click menu; the Develop module's version sends only the image on screen. The product page's structured data adds that the Photoshop plugin sits under the Automate menu, takes up to 30,000 pixels and does not accept PSD files.

In the standalone app, the batch documentation says you set the output as the same dimensions or the same scale factor before importing, and that Auto mode picks a model and settings per image while Manual applies one model to everything selected. Batch size is limited by your hardware. The page does not describe per-image overrides in Manual mode, so mixing a face-heavy crop and a landscape in one Manual batch applies one model to both. For a wedding workflow that means grouping like-with-like (tight face crops together, wide scenes together) before the batch.

Hardware limits

Topaz's system requirements list, for Windows, 16 GB RAM (24 GB or more recommended), a GPU with 6 GB VRAM (8 GB or more recommended) and 55 GB free on C:, and for macOS an Apple M-series chip with 16 GB RAM on macOS 13 or later. The same page says large upscales need 24 GB of RAM and outputs of 256 megapixels or more, for example 16,000 x 16,000 pixels, count as "very large" and need 128 GB of RAM at minimum. The Gigapixel product page adds that Intel Macs run core models locally but need cloud rendering for Wonder 2 and 3. Example (arithmetic from pixel counts): a 24-megapixel frame (6,000 x 4,000) enlarged 4x per side is 24,000 x 16,000, or 384 megapixels, which exceeds the 256-megapixel "very large" threshold; a 2x enlargement of the same file is 96 megapixels. Large wall prints can therefore hit the hardware wall before the software cap.

Vendor claims

If your enlargements are occasional, Adobe says Super Resolution on desktop enlarges low-resolution images for print (Adobe features); see best AI image upscalers for how to choose.

Where your files go

Topaz's pricing page says desktop apps run on a Mac or PC with unlimited local rendering, and lists browser apps separately. The Gigapixel product page says "Your files never leave your computer with local rendering" and also offers cloud rendering (two images at a time on Personal, four on Pro), and the batch documentation says generative models can render in the cloud or locally. The Lightroom Classic plugin has no cloud rendering. What happens to an image sent to the cloud, including retention and training: not found in sources reviewed. Topaz's privacy policy and terms: source inaccessible (the addresses tried returned 404). No conclusion about client confidentiality can be drawn from what was readable; keep to local rendering for sensitive files and read Topaz's current terms.

Licence, price and exit terms, as of 8 November 2026

Topaz's pricing page lists Gigapixel Personal at US$149 per year, US$19 per month labelled "annual commitment, billed monthly" and US$29 per month without an annual-commitment label; Gigapixel Pro at US$499 per year and US$54 per month (Topaz pricing). The US$12 and US$42 per month figures on the page are the annual-billing equivalents of the US$149 and US$499 annual prices, not extra options. Only part of Topaz's pricing page was read (the first 100,000 of 144,501 characters), and the one price conflict found is Studio Pro at US$75 versus US$79 a month. Confirm everything at checkout.

Example (arithmetic from the cited US$149 annual price): spread over 10, 25 or 45 weddings a year it is about US$14.90, US$5.96 or US$3.31 per wedding. That assumes every wedding is covered by one subscription; Gigapixel needs only be applied to the few frames you enlarge. For bundle versus singles, see Topaz Studio vs single apps.

Print and album scenarios

Gigapixel earns its place only when pixel count, not taste, limits the print. Example (arithmetic, assuming you target 300 pixels per inch; check your lab): a 16x20 inch print needs 4,800 x 6,000 pixels, about 28.8 megapixels; a 24x36 needs 7,200 x 10,800, about 77.8 megapixels. A 6,000 x 4,000 pixel file covers an 8x10 at that target but falls short of 16x20: cropped to a 4:5 shape it is 4,000 x 5,000 (20 megapixels), so it needs about 1.2x enlargement per side, or a lower ppi target. The cases where a specialist upscaler comes into question are heavy crops, large wall prints and cropped telephoto frames for album spreads. Topaz says "up to 16x pixels"; the wording suggests a pixel-count multiple (about 4x per side) but Topaz does not define the unit, so confirm in the app.

Where it sits in the workflow

Use it at the end of the chain, per print, after you have chosen the crop and finished the edit. Keep the enlarged master separate from the delivery gallery so a very large file does not slow your catalogue. Whether to denoise before or after enlarging is not settled by any source read; try both on one file. Check the file type, size and colour handling Gigapixel returns, because album software and labs have their own limits. Through the Lightroom Classic plugin a TIFF comes back, per Topaz's documentation; standalone export formats: not found in sources reviewed. Because the plugin will not take a RAW cleanly, a practical order is to finish the edit, export a full-resolution TIFF of the chosen crop, then enlarge that, which is reasoning from the documented limit and not a Topaz recommendation.

When the licence stops

Behaviour after cancellation: not found in sources reviewed. Keep every enlarged master you have sold and note the tool, version and settings so a reprint order does not depend on a subscription you cancelled. Topaz is subscription only, and the one-computer rule is documented for Photo, so confirm Gigapixel's rule before relying on it for a print deadline.

Honest downsides

• No perpetual licence for new buyers.
• The one-computer rule is documented for Photo; Gigapixel's own terms were not found.
• A trial: not found in sources reviewed on the Gigapixel product page.
• The Lightroom Classic route converts to TIFF and cannot use cloud rendering, and RAW is not recommended as input.
• Large enlargements have a hardware floor: 24 GB of RAM for large upscales and 128 GB for outputs of 256 megapixels or more.
• Topaz's own pages give two different headline factors (6x and 16x).
• Privacy policy and terms: source inaccessible.
• A ceiling like "16x pixels" says nothing about how a given file will look at that factor.

Who should pick it

• Pick Gigapixel if enlargement is a recurring paid service and you want a dedicated tool.
• Skip it for now if print sizes are modest, or you hold Topaz Photo and its Upscale may cover you; test that first.
• Consider Studio only if you need several Topaz apps.

Check this in your trial

1. Confirm a trial exists and its length before relying on one.
2. Enlarge a file you have already sold as a print to your largest delivered size; view the print at normal distance.
3. Check faces, lace, hair and signage for artefacts.
4. Record time, memory use and file size on your computer at 2x, 4x and 6x.
5. Check whether cloud rendering is used and whether you can turn it off.
6. Send a three-image batch from Lightroom Classic with Edit a Copy, confirm a TIFF returns, and check where it lands in the catalogue and how large it is.
7. Calculate the pixel dimensions your largest print needs and start the test there.
8. Run the enlargement in one pass and two passes and compare.

Original test pending: we would enlarge a RAW set at 2x, 4x and 6x, record file size and processing time and view prints, because the print decision depends on results at size that no page here measures.