Lightroom now uses Topaz AI Sharpen, following Adobe’s acquisition, but at 10 AI credits per operation!

This happened in the June 2026 Lightroom update, so that you can now use an AI Sharpen option powered by Topaz AI, and there’s an optional Topaz Denoise checkbox you can use alongside it. It shows Adobe’s determination to dominate the AI landscape, but at what cost? (Not to Adobe, of course, but to us.)

First things first. This new feature is for Lightroom only, not Lightroom Classic. Maybe it will find its way into Lightroom Classic in due course – I don’t know.

The Topaz sharpen and denoise feature seems a bit of an odd fit to me. First, Adobe’s existing AI Denoise seemed pretty good to me already. I prefer the results from DxO PureRAW, but it’s close enough that the convenience of Adobe’s built in denoising is compelling.

But what we’re actually talking about, of course, is AI sharpening. Now having tried out Lightroom’s Topaz powered AI sharpening, let’s just say I think its improvements are nuanced (see how polite I can be?). Maybe it sharpens better than the regular sharpen options, maybe if used in conjunction with Topaz denoise it produces a better outcome, but I’ve not seen anything yet to convince me.

Does the Topaz AI perform any miracles? I haven’t seen any yet, but I’ll keep looking. Others may feel it’s the best thing they’ve ever seen. I’ve tried it on shots which I think maybe lack just that last bit of crispness (I don’t keep blurred shots) and it does improve them noticeably, but then so does sharpening with regular tools, and I’m zoomed right in to 200% looking for differences, so it’s already becoming somewhat academic.

I do think noise and sharpness are closely related. Your softest shots tend to be the ones shot at the highest ISOs. I tested the Topaz AI and Sharpen process against DxO DeepPRIME on a Nikon NEF file shot handheld from a hotel window in Reykjavík at ISO 45,600. It was both noisy and rather soft, as you might expect. DxO DeepPRIME did a pretty remarkable job. The Topaz process didn’t even get close for noise OR sharpening.

If I’ve not been moved by the Topaz AI sharpening process, I have, however, seen two things that definitely count against it.

The first is that it costs 10 AI credits per operation. Yes, TEN. Consider that Adobe Firefly’s generative tools cost just one credit per operation, and that while Adobe’s all apps plan comes with 4,000 credits per month, the regular 1TB Photography Plan comes with 1,000, but users of the legacy 20GB plan (without the 1TB storage) get just 100 credits, and Adobe’s Lightroom Plan also comes with just 100. At 10 credits per use, you could use up your monthly allowance on just ten images.

Indeed, now that Adobe is working actively with parter AI models like ChatGPT and Google Gemini, you may already have noticed a steep increase in the cost of generative AI credits with third-party tools. These all operate within the Firefly framework. And naturally, Adobe will be glad to sell you additional Firefly Plans with tiered monthly AI credit allowances based on which tier you opt for. However, for many people it’s bad enough that Adobe software is on a subscription, but this brave new world of AI credits has added a whole new cost structure (and potentially a whole new layer of cynicism).

You don’t have to use these AI tools at all. Adobe’s own AI denoise is still free to use right now, and if you use it alongside Lightroom’s regular sharpening tools, I feel you can get results similar enough to those from Topaz AI that I wouldn’t bother swapping.

But then I mentioned two things I have against this new process. The second one is speed. It’s SLOOOW. 2 minutes and 12 seconds to process a 20MP RAW file? It’s a lot slower than Lightroom’s regular AI denoise and slower even than DxO’s DeepPRIME process. I guess it’s using off-device cloud-based AI processing. So… why?

I imagine the argument will be that only a cloud-based server will have the necessary processing power. How convenient. I can’t help noticing that AI providers can’t monetise on-device AI processing, but cloud-based processing is perfect for AI credit schemes and subscriptions.

Regular visitors to this site will know that I’m a big fan of DxO PureRAW. There are reasons for this. It comes with a perpetual license, so there’s no subscription, and it uses on-device processing so there are no AI credits. PureRAW isn’t cheap, but you pay once and you’re done. Adobe’s AI credit setup is too open-ended for me. You’re already paying to rent its software, and now you’re paying again to use its latest tools. If it was dramatically better I might be convinced, but it’s not even that.