Comments on: Topaz DeNoise AI vs. Neat Image – Battle of the giants

By: TommyK

TommyK — Wed, 21 Jun 2023 00:30:28 +0000

I thought the Topaz (free demo) over did it. The software limited me from making adjustments like you pointed out. I’ve been using Neat Image V8 and it does a good job with overdoing it. Thanks for doing the comparison. I’m validated to stick with Neat Image. Sometimes if Neat Images does not remove all the specs (if that is what you call them, similar look to a dead pixel), the blur tool in Photoshop solves the problem. Again, thanks for this post. Tommy

By: Joe Eisen

Joe Eisen — Sun, 30 Oct 2022 11:21:51 +0000

Thank you for the informative article and the provision of the photos.
You have a great camera which produces such low noise at this ISO.

I took the liberty of comparing this photo with GIMP-ML and GMIC, which I think also produce good results at low ISO noise. And without artifacts.

Please see https://ibb.co/ZTC27Hm

By: Aditya Sridhar

Aditya Sridhar — Fri, 11 Jun 2021 03:45:31 +0000

In reply to David Pastern.

Thanks so much for reading and for the kind comment David!

NI v8 is certainly pretty good at reducing noise, but the learning curve can be slightly steep. DeNoise AI, on the other hand, is much easier and also quicker. The main issue with DeNoise AI is the lack of user control over the noise reduction process, though. Considering your interest in photographing deep sky objects, I presume your primary consideration would be whether you require full manual control over the process vs. a simple, non-complex software that provides you with just a few basic NR sliders.

I would also consider testing out the trial version of DxO PureRAW. I played around with it recently and it does seem pretty promising. It’s quite similar to DeNoise AI (with even fewer control over the whole NR process!) but appears to have a more robust algorithm.

Thanks again for the comment and apologies for the late reply! I hope this was helpful. If you have any more questions, feel free to leave a comment and I’ll try to get back to you ASAP 🙂

By: David Pastern

David Pastern — Mon, 24 May 2021 13:03:29 +0000

Many thanks Aditya. I have an (older) NI v6 pro licence and was thinking if it was worth moving to Topaz denoise AI etc. I’m going to become more involved with deep sky object imaging, so most of my processing of the data will be with PixInsight, with some minor work in Photoshop CS6. Possibly some noise reduction with NI (standalone or plugin), although PixInsight has some great noise removal tools such as multiscalelineartransform and tgvdenoise (it’s a very complicated application, makes Photoshop look easy lol!).

I think I will upgrade to v8 pro (stand alone and CS6 plugin). I don’t know how large the improvements are from v6 to v8. I haven’t used NI in probably 10 or so years, as I’m doing less terrestrial photography (actually, macro photography and a bit of birding photography, although my latter is nowhere near as remotely good as yours!).

Thanks for your thoughts.

Cheers,

Dave

By: Axel Elfner

Axel Elfner — Wed, 12 Aug 2020 18:28:34 +0000

Thank you for the analysis. I would have used the Topaz software, it having been endorsed by a professional photographer I respect, but it does not run on Linux, and that’s a non-starter for me. Like you, I have been happy with NeatImage 8 Pro. As you say, a bit of a learning curve, but lots of fine control, and you can use the generic or self-calibrated profiles if you don’t want to custom work each image.

That said, Topaz’s separate sharpen module that claims to reconstruct artifacts instead of just edge sharpening does look impressive.

By: Jehan Tillekeratne

Jehan Tillekeratne — Tue, 04 Aug 2020 04:38:44 +0000

Thanks for putting together this direct comparison! I picked up Neat Video for video needs, but was curious about AI for stills, and improving speed of workflow. Won’t be deciding for a while, but I was also concerned about the confusion of sharpening settings within their DN tool (Their site and product offerings in general are just a bunch of overlap and confusion really). Hope AI does improve and offer a streamlined approach when needed though.

By: Aditya Sridhar

Aditya Sridhar — Tue, 23 Jun 2020 12:57:39 +0000

In reply to Dick Locke. Thank you so much for visiting and commenting, Dick! I'm glad you found this post helpful! :)

By: Dick Locke

Dick Locke — Fri, 12 Jun 2020 12:18:19 +0000

Thanks for the comparison. I’ve been using Neat Image for years and wondered if I was missing out with Topaz. I’ll be in no hurry to change.

By: Aditya Sridhar

Aditya Sridhar — Wed, 26 Feb 2020 16:19:58 +0000

In reply to Alex Becker. Thanks a lot for visiting, Alex! I was curious to see how far AI has come in terms of reducing the need for manual input in post-processing. It certainly is a pain to have to draw layer masks when editing multiple photos and is quite time-consuming. I think DeNoise AI has potential and will improve over time. Hopefully, they'll sort out the artifact and sharpening issues soon, but it's a promising start. I, too, am impressed by the quality of the output that these products deliver from time to time. The future's looking bright, that's for sure :)

By: Alex Becker

Alex Becker — Tue, 25 Feb 2020 21:51:32 +0000

Nice post Aditya, and I agree with your conclusions and pros vs cons. In your sandpiper image I still see noise on the beak area especially on the Topaz image that isn’t present with the NI frame, likely related to some of the additional sharpening. The Topaz falcon image does have a lot of artifacts when you look closely as well — I guess you could brush them out, but then that isn’t much of a time saving compared to layers. I try these products every now and then and often make more work for myself trying to clean up various aspects than if I hadn’t tried, but every now and then I am very impressed by them too.