Topaz Gigapixel AI

Topaz Gigapixel AI is an image upscaler. You give it a picture that is too small and it gives you back a larger one, with edges, textures and faces rebuilt by a set of AI models instead of stretched pixel by pixel. It runs as a desktop app on Mac and Windows, as a plugin for Photoshop and Lightroom Classic, and with optional cloud rendering for the heavier models. Topaz has renamed it a few times, so you will see both Gigapixel AI and plain Gigapixel in search results. It is the same product.

Marketing teams meet it at a predictable moment. The print shop wants 300 dpi and the only product shot is a web image from 2019. A partner sends a logo that is 200 pixels wide. An image from a generator looks great on screen and falls apart on a poster. Gigapixel exists for that gap between the file you have and the file you need.

This post covers what the tool does, where it helps a marketing team, how to run a job without wasting time, how the pricing changed, what upscaling means for rights, and the checks that stop an upscaled image from looking strange once it is printed two metres wide.

What Topaz Gigapixel AI does

Classic resizing guesses each new pixel from its neighbours. The result is bigger and softer. Gigapixel uses models trained on many pairs of small and large images, so it tries to predict what the sharp version would look like. Topaz advertises upscaling of up to 6x, and you pick a model depending on what is in the picture.

• Standard and High Fidelity models. The everyday options for photos. High Fidelity tries to keep fine detail from good source files.
• Low Res. For small, compressed images such as old web graphics and screenshots.
• Text & Shapes and Art & CG. For logos, illustrations, interface shots and rendered images, where clean edges matter more than texture.
• Face Recovery. A separate pass that rebuilds faces, which are usually the first thing to look wrong after a big enlargement.
• Recover and Redefine. Generative models that add more invented detail. They can rescue very poor files, and they are also the ones most likely to change what the picture shows.

The generative models are heavier. Topaz lists higher hardware requirements for them, and they do not run on older Intel Macs. That is one reason cloud rendering exists: the job runs on Topaz servers instead of your laptop. The model list changes with each release, so check the current product page before you write a process doc around one model name.

What Topaz Gigapixel AI is good at for marketing teams

The best uses all share one thing. You already have the right image. It is just too small. Gigapixel is a rescue tool, not a creative one.

• Print from web assets. Product shots, team photos and event pictures that were exported for a website and now need to go on a roll up banner or a brochure.
• Archive images. Old campaign photos where the original raw files are lost and only a small JPEG survived.
• Partner and customer logos. The low resolution logo on a case study slide, as a stopgap until the partner sends a vector file.
• AI generated images. Many generators export at screen size. An upscale makes a good result usable for print or a large hero image.
• Crops. A tight crop from a wide shot loses most of its pixels. Upscaling the crop gets it back to a usable size.

It is less useful when the image is wrong rather than small. A blurry photo of the wrong product is still the wrong product, just larger. If the brief needs a new picture, generate or shoot a new picture. Our guide to the best AI image generator for marketing covers that side.

How to use Topaz Gigapixel AI step by step

1. Start from the best file you have. A PNG or a high quality JPEG upscales far better than a screenshot of a screenshot.
2. Decide the final size first. Work back from the placement: print size and dpi, or pixel width for a screen. Do not upscale to 6x just because you can.
3. Pick a model that matches the content. Photos, text and graphics, and illustrations each have their own model.
4. Preview a detail at 100 percent, not the whole image. Look at edges, skin, hair and any text.
5. Turn on Face Recovery only when there are faces, and compare it with the plain result. Sometimes it changes a person too much.
6. Use the generative models last, for files that nothing else can save, and check them the hardest.
7. Export, place the file in the real layout, and look at it at the final size before you send it anywhere.

For many files at once, there is batch processing, and the Photoshop and Lightroom plugins let a designer upscale without leaving the tool they already work in. Save the settings that work for your product photography so the next person gets the same result.

Topaz Gigapixel AI pricing

The pricing model changed. For years Topaz sold perpetual licences: you paid once and kept the version you bought. In late 2025 the company moved new versions of its apps, Gigapixel included, to subscriptions. People who bought a perpetual licence before the change keep the version they own.

Today you can subscribe to Gigapixel on its own or get it inside Topaz Studio, a bundle with the photo and video apps. Plans differ in cloud rendering limits and in how much commercial use they allow, and the prices quoted on review sites do not agree with each other. At the time of writing, check the official pricing page rather than any number in a blog post, including this one. If you only need to upscale images, the standalone plan is usually enough. The bundle starts to make sense if your team also restores video.

Rights: upscaling does not change who owns the image

This is the part teams skip. A bigger file is not a new image in the legal sense. If you did not have the right to use the small version, you do not have the right to use the large one. A stock photo licensed for web use is not suddenly licensed for a billboard because you upscaled it. A partner logo is still their trademark. A photo of a person still needs the same release it needed before.

The tool licence matters too. Some Topaz plans list limited commercial use and others full commercial use, so check that the plan you pay for covers client work and paid campaigns. For AI generated images, the rules of the generator still apply after the upscale. Our post on AI compliance has a short checklist for keeping a record of where each asset came from.

Limits and what to check

An upscaler invents pixels. Most of the time the invention is invisible. Sometimes it is not, and at print size the mistakes are large.

• Text. Small lettering on packaging can turn into shapes that look like letters. Retype or replace any text that has to be readable.
• Faces. Face Recovery can make a real person look like a slightly different person. Compare against the original for anyone who will be recognised.
• Product details. Stitching, labels, ports and buttons can be smoothed away or redrawn. The product in the ad has to match the product in the box.
• Texture. Skin, fabric and food can come out waxy or over sharpened. Look at 100 percent.
• Logos. An upscaled logo is a stopgap. Ask for the vector file.
• Colour. The upscale itself does not fix colour. Check the file against your brand colours and the printer's profile.

Treat the generative models with extra care. They are the most impressive on a bad file and the most likely to add something that was never there. For product and people photos, the plainer models are usually the safer choice.

Topaz Gigapixel AI vs generative upscalers

There are two kinds of upscaling now. Gigapixel's core models try to stay faithful to the source. Tools like Magnific lean into generation: they add new detail on purpose and let you control how much the image changes. That is great for concept art and for AI images where nobody minds a richer texture. It is risky for a product shot that has to match reality.

A simple rule works for most teams. If the picture is evidence, such as a product, a person or a place, use a faithful model. If the picture is mood, such as a background, a concept or an illustration, a creative upscaler is fair game. For shop and catalogue images, our guide to AI product photography for ecommerce goes deeper on keeping products accurate.

Where it fits in a week of work

Gigapixel sits at the end of the pipeline, right before export. The image is approved, the layout is set, and the file is too small. Put it there and it saves reshoots. Put it at the start, and teams start to treat every thumbnail as raw material, which is how a blurry 2019 photo ends up on a trade show wall.

The bigger fix is upstream. Keep originals in one place, at full size, with the licence attached, and most upscaling jobs never happen. In Content24, AI Images produces campaign visuals next to the copy that goes with them, and Brand Kit keeps the logo files and colours so nobody has to rescue a 200 pixel logo from a slide deck. When you do need a generator, Nano Banana and Adobe Firefly are two we have covered.

Frequently asked questions

Is Topaz Gigapixel AI free?

No. You can try it, but full use needs a paid subscription, either for Gigapixel alone or as part of the Topaz Studio bundle. Older perpetual licences still work for the version they cover.

Can I use upscaled images commercially?

Two things have to be true. Your Topaz plan has to allow commercial use, and you need the rights to the original image for that use. The upscale does not create new rights.

Does it work on AI generated images?

Yes, and it is one of the most common uses. The Art & CG model suits illustrations and renders. Check hands, text and small details, because generators and upscalers can make the same kind of mistake twice.

How big can it go?

Topaz advertises up to 6x. In practice the useful limit depends on the source. A clean file holds up well at 2x to 4x. A tiny, compressed file will look invented long before 6x.

Do I need a powerful computer?

For the standard models, a recent laptop is fine. The generative models need more graphics memory, do not run on older Intel Macs, and are often easier to run with cloud rendering.

Conclusion

Topaz Gigapixel AI is a good rescue tool for marketing teams. It turns the file you have into the file the printer asked for, and it does that better than a simple resize. Pick the model that matches the picture, check faces, text and product details at full size, and remember that a bigger file comes with the same rights as the small one. Get Started with Content24 to keep your originals and brand files in one place, so fewer images need rescuing in the first place.