Can Lightroom's New Denoise AI Replace DxO PureRaw, Topaz DeNoise AI, and ON1 NoNoise AI?

Denoise software has made exponential growth over the past several years and arguably become a necessity for many genres of photography. The market has been highly competitive and typically always required extra software which might be completely shaken up by Adobe. 

Last year i started dabbling a bit into wildlife photography and found myself shooting at such ISO levels I never thought I would set my camera too as a landscape photographer. Through that journey, I learned not only how important but how incredible software such as Topaz, DxO, and ON1 can be for reducing noise and enhancing details. The results were so good, they changed how much I worried about taking higher ISO images in the future.

Thus when Lightroom's new update included Denoise AI, I was intrigued but skeptical. Throughout this video, I compare two completely different genres of images using Topaz Labs Denoise AI, DxO PureRaw 3, and ON1 NoNoise AI 2022. Personally in my testing last year I preferred the results of DxO PureRaw the best overall, and you'll clearly see why within the video. Many of these programs do more than just reduce noise which is hard to gauge in a direct comparison. 

The question really came down to can Adobe's implementation within Lightroom Classic compete? I'd love to know your thoughts and hands-on experience down below. Have you tried this for yourself yet? 

Alex Armitage's picture

Alex Armitage has traveled the world to photograph and film some of the most beautiful places it has to offer. No matter the location, perfecting it's presentation to those absent in the moment is always the goal; hopefully to transmute the feeling of being there into a visual medium.

Log in or register to post comments
16 Comments

The new LRC denoise module takes 9x longer to process an image than Topaz deNoise RAW model (which has better results for me than its other models). Topaz DNG files are 6x larger than the original RAW files (50Mb -> 291Mb). LRC DNGS only blow up to 119Mb. Final product seems to be about the same. Topaz seems to retain more detail, but have added rare artifacts on one or two of my images (out of hundreds).

Thank you for the in-depth comment. Being my snarky self, I would reply "you can't batch process the AI noise"....yet ;)

Basically the same findings here. Also applying to much noise reduction in LRC leads to slightly waxy pictures.

thank you!

I haven't found this to be the case at all. In fact, LR beats Topaz by a significant margin. I'm guessing you're using inferior hardware?

If it were inferior hardware then it would effect both softwares don’t you think? Lightroom on my machine works well but the denoising takes for ever.

I agree.
I've had good results from Adobe Denoise AI in ACR so images look cleaner but softer too. But I also have issues with it taking around 10 minutes to process an image (Sony a1 50mp raw) versus around 20 seconds using Topaz plug-in on Photoshop - on my PC i7, 32gb ram. So I'll be sticking with Topaz for now for my work flow, it's great it's been added to Adobe though and hopefully it will speed up for us old PC users and get further options in the future 👍

The advantage of Topaz AI is the masking feature. Sometime, you don't want to sharpen the entire image. Still the winner in my workflow.

You can batch process in LRC however all images will get the same amount of denoise applied. If working within a scene and similar settings this will work fine but if different scenes and settings then you may wish to have a different amount applied to different images.

Kinda weird everyone all of a sudden comparing denoise software! Every since Topaz came out with all its AI software suddenly everyone has tried to match or make better. I used Topaz Denoise programs years before the AI's. All programs C1, Dxo, ON1 Photo RAW, Aurora, Luminar ect. have tried it over the years. First noise problems are mainly from pixel peppers! Next Lr has the best and complete lens correction were other do not keep up so where are you going to start! Again Topaz is a link in most post programs now. Yes Lr is very slow but works as the many resent tools that make it more like PS and easier than PS. The cool things of Topaz is you can use it several times on an image and get a brighter image the second time through, also you use the good or bad face but click on the bad it asks what went wrong, updates are very often.....

I like the new LR denoise and I think it compares just fine to DXO XD. I suspect it takes a bit longer but I can't deny the convenience factor of having everything in one program. It's exactly what LR was lacking for a long time, now we just need a solid AI sharpening/deblur tool and we're set.

I also think using AI to get better auto settings or cull (ala Aftershoot) would be a SOLID addition to lightroom. Can you imagine a few clicks and you're done editing and culling? Import, cmd-u, export. The dream.

I have been using Topaz DeNoise AI and Topaz Sharpen AI about a year now and really like the results. So I updated LRC just now to see how this new DeNoise Ai works. I have a Apple M1 MacBook. The picture i used foe now was shoot indoors with a Nikon D750 no flash ISO set to Auto -> 4500, tamron dx 18-400mm set to 35mm, f5.6 and speed to 1/320 sec (all manuel except iso). noise wasn't to bad on the picture, but there was some. Denoise manual in LRC removed most of the noise, but made picture soft as well. New LRC DeNoise AI removed Noise very welland fast, but still a little soft when zooming in. Topaz DeNoise AI removed Noise and contained sharpness a touch better then LRC DeNoise AI. So if i compare all three Topaz is still the winner because it not only removed noise but add sharpness to the picture and i can change settings if i don't like the results. LRC doesn't give you much settings to change. Speed wise i would say LRC DeNoise AI was a little faster, but not better in the end result. So this was not a a picture with bad noise and i know that Topaz DeNoise AI does a really good jobeven with a lot of noise and i do use the Sharpen AI sometimes as well and works great. By the way LRC crashed my MacBook twice I mean completely nothing not even the mouse worked and then restarted by itself. That hasn't happened before the update ever and i eas just zooming in and out to compare the results.

I like that Topaz have different models that you can switch between to see which works best and I love that it has masking, but to be fair it also includes sharpening by default.

This makes if not an immediately fair comparison.

I’ve been using DXO for denosing and happy with the results. Tried the LRC Denoise to compare and noticed that sometimes it washes the colour away around subjects leaving a gray area around the mask. I photograph dance performances, in very low light conditions so noise is inevitable.
LRC worked well when the stage was well lit, but overall DXO is doing a better job for me.

The biggest problem with these types of articles is that they bog us down in minute differences when the truth is, all "AI" software of this sort is good enough for most people and most uses.

These things are fun to view and fun to make. But I do think they ultimately lead to false assessments of overall value. I recently bought On1 Raw 2023 and I am quite satisfied with it's ability to control noise and sharpness in high ISO ORF files. That doesn't mean I'm going to get rid of my Topaz tools though, they are paid for, and they might have some value for SOME photos.

Both On1 and the Topaz tools have the ability to selectively mask areas of the photos for NR and sharpness, and then subsequently modify overall sharpness and NR if it isn't perfect. I have recently learned that Lr's noise reduction has this ability as well.

PureRaw does a fine job, but does its "take it or leave it" approach to these issues really reduce its overall value? THAT is something each of us ultimately must decide for ourselves. But for me, it is less desirable than an NR tool that offers me more localized control.

it's good but it's not topaz good.