Topaz Labs Review: Adobe Bought It. Here's What Survives.

In June 2026, Adobe announced it was buying Topaz Labs, a Dallas company that had spent two decades turning down venture money and building a photo and video AI business on its own terms. Topaz Labs is an AI photo and video enhancement suite built around upscaling, denoising, and sharpening, the tools photographers and video editors reach for when a camera got something wrong. This review covers what its five apps cost today, what they're genuinely good at, and what an Adobe-owned Topaz means for anyone about to buy in.

Short answer: Topaz Labs is worth it if you regularly need to rescue real, imperfect photos or footage, low light, low resolution, motion blur, and don't mind a subscription for models that are genuinely ahead of the pack. It's not worth it if you only need occasional light upscaling; a tool you already pay for probably covers that.

Quick Verdict

What works well:

• Dedicated AI models trained for specific problems (old footage, low light, faces, text) rather than one generic upscaler doing everything

• Runs locally on your own machine, not just in a browser, so you're not capped by upload limits or a slow connection

• Plugs directly into Photoshop, Lightroom Classic, Premiere Pro, and DaVinci Resolve instead of forcing a separate workflow

• NeuroStream compression means the newer models run on consumer GPUs, not just studio-grade hardware

What to watch:

• No single subscription covers everything. Photo AI, Video AI, and Gigapixel AI are sold separately unless you buy the bundle

• Perpetual licenses ended in October 2025. Every new customer is now on a subscription, and that's a real change for people who liked owning software outright

• Push any model too hard and results can look over-sharpened or artificial

• Desktop apps want a real GPU. On weaker laptops, renders slow down noticeably

What Is Topaz Labs, and Why Did Adobe Want It?

Albert Yang, a computer-vision PhD with a background in video and audio signal processing, started Topaz Labs in Dallas in 2005. His son Eric joined the company in its early years, left, then came back as CEO in 2018 to rebuild the product line around Photo AI and Video AI. Neither ever took venture funding to do it. Topaz Labs is, by its own account, a bootstrapped business that has quietly built roughly $48 million in annual revenue and more than 3 million customers, from working photographers to Fortune 500 production teams, without a single funding round to point to.

That changed twice in the last year. In October 2025, Topaz ended perpetual licenses entirely and moved every product to subscription pricing, a decision that split its long-time user base (more on that below). Two months later, it won a 2025 Technology & Engineering Emmy, shared with MTI Film, for AI-driven restoration work on high-quality TV catalog footage. Then in June 2026, Adobe announced a definitive agreement to acquire the company outright. No price has been disclosed, the deal is still awaiting regulatory approval, and it's expected to close in the second half of 2026.

The subscription change is worth pausing on, because it genuinely annoyed a chunk of Topaz's oldest customers. For years, buying Gigapixel or Photo AI meant paying once and owning that version forever, only paying again if you wanted an upgrade. Since October 2025, that option is gone for new buyers. Existing perpetual license holders keep what they already own, but they're frozen out of new models and updates unless they move to a subscription too. It's a familiar trade in software right now, more frequent updates in exchange for giving up ownership, and Topaz's own community forums show plenty of users still working through whether they're fine with it.

Adobe's own reasoning is specific: it wants Topaz's NeuroStream technology, which cuts the memory a large AI model needs by up to 95%, roughly the difference between needing a server rack and needing a laptop. The plan is to fold that into Firefly, Firefly Services, and the wider Creative Cloud lineup. For now, Adobe says Topaz's standalone apps keep shipping and Eric Yang stays on as CEO. Topaz isn't the first AI creative tool to end up owned by a bigger design company either, Adobe has never met a good creative tool it didn't eventually want to own. Leonardo AI made a similar trip into Canva's hands in 2024, and the pattern is becoming familiar: build something genuinely good, and one of the platform giants eventually buys the distribution.

The Apps in the Topaz Suite

Topaz isn't one app, it's five, each solving a different version of the same problem: making a real photo or video look like the best possible version of itself. Think of it less like a Swiss Army knife and more like a specialist's toolbox, you reach for the exact blade for the exact job instead of one tool trying to do everything adequately.

• Photo AI. Runs denoise, sharpen, upscale, and face recovery in one automated pass, plus dedicated tools for dust and scratch removal, object removal, and fixing images that just missed focus. Works standalone or as a plugin for Photoshop, Lightroom Classic, Capture One, and Affinity Photo.

• Video AI. Upscales footage to 4K or 8K, removes compression artifacts, stabilizes shaky clips, interpolates frame rate up to 60 to 120 fps, and converts SDR footage to HDR. Ships as a Premiere Pro and DaVinci Resolve plugin as well as a standalone app.

• Gigapixel AI. The pure upscaling specialist. Scales up to 6x by default and 16x at the extreme, with separate models tuned for faces, text and line art, and CG or digital art.

• Bloom. A browser-based diffusion model built specifically for AI-generated and CG artwork rather than real photography. It upscales up to 8x or 100 megapixels, takes an optional text prompt to steer the added detail, and includes a free tier of 10 upscales a month.

• Astra. A usage-based, credit-metered cloud tier of the video models for people who want Topaz's quality without buying the full desktop app.

How It Performs

We didn't run a paid Topaz subscription for this review, so treat this section as what's documented and independently reported, not a first-hand test. On the documentation, the differentiation is real: Video AI's Starlight and Proteus models are trained specifically to restore old, low-bitrate, or heavily compressed footage rather than just resizing pixels, and Gigapixel's face recovery model is built to reconstruct facial detail without the waxy, over-smoothed look older upscalers are known for.

Independent reviewers back that up. G2 users rate the individual apps between 4.5 and 4.7 out of 5, and call out the noise reduction and detail recovery specifically as standouts. The recurring complaint across independent reviews is consistent too: push a model harder than the image needs and results can look over-processed or artificial, and the desktop apps genuinely need a capable GPU to run smoothly. On older or integrated graphics, renders slow down noticeably. If you're buying this to actually use it daily, budget for real hardware, not just the software.

There's also a middle ground worth knowing about: Video AI subscribers get a set number of cloud rendering credits each month (25 for Personal, 100 for Pro) on top of unlimited local rendering, so you're not fully dependent on your own GPU for every export. It's a sensible hedge, not unlimited cloud processing, but enough to cover overflow work without buying new hardware immediately.

Topaz Labs Pricing in 2026

Every Topaz app is sold on its own, or bundled into Topaz Studio. Prices below are current as of writing and verified on Topaz's own pricing page.

There's no free plan for Photo AI or Gigapixel AI. Video AI offers a full-featured trial that watermarks exports until you subscribe, and Bloom has a genuinely usable free tier at 10 upscales a month. Since October 2025, none of this is available as a one-time purchase anymore, only subscriptions, which is the single biggest change to how Topaz sells software in its 21-year history.

Is it worth the price? If you only need one app occasionally, paying $29 to $59 a month adds up fast for something you might use twice a month. If you're a working photographer or editor who needs two or more of these tools regularly, the Studio bundle at $399 a year works out cheaper than paying for Photo AI and Video AI separately, and is the better default choice.

Topaz Labs vs the Alternatives

We cover Photoshop's own AI toolset, including its newer multi-model Generative Fill, in full in our Adobe Photoshop review. If Photoshop's built-in upscaling already covers what you need, Topaz is a genuinely optional extra, not a replacement for it.

Who Should Actually Buy This

Best for:

• Photographers dealing with real, imperfect source files: noise, motion blur, missed focus, low resolution

• Video editors restoring old or low-bitrate footage, or needing clean upscales to 4K and beyond

• Studios and production teams that already budget for specialized software per role

Skip it if:

• You only need occasional, light upscaling, a feature already inside a tool you pay for will likely be enough

• You want a single flat subscription that covers everything, Topaz's per-app pricing works against you

• You're not running a machine with a dedicated GPU, since performance will disappoint you regardless of the software

The Verdict

Topaz Labs earned its reputation the slow way, two decades of specialized, locally-run AI models built by a company that never took outside money to get there. That's precisely why Adobe wanted it enough to buy it. For now, nothing changes for a new subscriber: the apps still run standalone, Eric Yang is still CEO, and the pricing above is what you'd actually pay. The real question isn't whether Topaz's models are good, independent reviewers and its own Emmy consistently back that up, it's whether you need more than one of these apps regularly enough to justify a subscription instead of leaning on the AI upscaling already built into software you own. If the answer is yes, Topaz Labs is one of the few AI tools in this space that earns the price tag.