I just made a video in which I compared Topaz Denoise AI with Capture One and darktable. I mainly tested it for my own curiosity. I have used quite a few Raw developers and image processing programs and I wondered if Topaz Denoise really performed better than average.
My conclusion is that Topaz Denoise produced the best result, with all three results being close. For the noisiest of images, it’s hard to beat Topaz Denoise among the three candidates, but for images with more mild noise, Topaz does not provide a noticeable advantage over more traditional methods. Topaz has two main advantages over traditional methods: it is very good at retaining detail even with heavy noise removal, and it operates relatively automatically without the need for manual local editing. I am guessing this is accomplished by it being trained on a large number of images and thus being more content aware than indiscriminate denoising.
I am including the unedited shot along with the images processed by the three respective programs so readers can compare the results. Please feel free to download these images for personal use to compare the results. This comparison was not sponsored by Topaz Denoise and I didn’t get a free copy of it either. Also, with more work, it probably would have been possible to improve the results in both Darktable and Capture One, but I thought it was a fairer comparison to spend a few minutes with each program and see what was possible.
Before Denoising

Topaz Denoise Result

Note: there is a watermark on this image because I am using the trial version of Topaz Denoise for this test.
Darktable Result

Capture One Result

Topaz vs Darktable

