Topaz Labs AI

[Budgie1]

A couple of nights ago I came across a YouTube article/review by @Luke Newbould on his channel about the Topaz Labs AI software. 

I hadn't heard about this software and looked quite interesting for reducing noise and sharpening the image. So I thought I'd give it a go and have to say it's brilliant. 

Like with other software, you can over-do the correction but it is controllable and leaves a much improved final image. It's not cheap but I felt it was worth the cost, so I bought the Image Quality Bundle, which includes Sharpen AI, Gigapixle AI & DiNoise AI.

I'm not going to do a review or anything (I'll let Luke post a link to his video as he's covered it all in that ) and I'm not connected to this software, I just though it was worth a mention because I was so impressed with the results so far. I know it's not new to the market but it doesn't seem to be mentioned that much and  I hope it's useful to those who may want to use the software.

So, here's a couple of examples of what it can do. These were both done with one sweep of GigaPixle AI, top is the before and below is the after.

[Clarkey]

8 minutes ago, Budgie1 said:

So I thought I'd give it a go and have to say it's brilliant. 

I recently brought the software and I have to agree with you. Used carefully it is really very good for noise reduction. I have found you need to be careful not to over-do-it or you can get some funny star artefacts.

[Budgie1]

11 minutes ago, Clarkey said:

 I have found you need to be careful not to over-do-it or you can get some funny star artefacts.

It can also do funny things with galaxies, adding detail that isn't there by over-sharpening the image. So you do have to be careful.

Edited November 21, 2021 by Budgie1

[Lee_P]

I use Topaz DeNoise for almost all of my images. It can be really good. I've found it works best on starless images, and then I add the stars back in afterwards. 

[ONIKKINEN]

I too use Topaz DeNoise frequently. I find that it really depends on the quality of the data how well it works, but generally there is always an improvement. If the data is really bad, it wont do any magic tricks and the final result will look like a painting if overdone, but with decent data i find that it works really well.

[Luke Newbould]

Hey there @Budgie1 ! - I'm glad you found my little video overview useful, and thank you for the shout out too! 🙂 

I whole-heartedly agree with the points made in this thread, when used carefully it's a thing to behold! - taken too far, artefacting can occur as @Clarkey mentions, or things go a bit like a painting as @ONIKKINEN mentioned!

@Lee_P brings up a great point there about using it on starless images, then adding them back in afterwards - when used on a starless image it's always been at it's best for me, much easier to control the outcome!

I wonder at where this software will be in a few more years, it's boggling!

Clear skies 🙂