ON1 NoNoise AI comments and comparison

Yes, agree.

ON1 NoNoise AI posts a Disclosure upon opening images that the product is “optimized” for raw images. It’s not clear to me (as a plugins-only user for any company’s products) that if my raw image starts in Ps 2021 whether it will be passed as a raw into NoNoise AI (or, as I am nicknaming it after trying a few pics… “NoNo AI”) or if it will do whatever the heck DAI is doing & just “passing pixels” into the program.

I didn’t watch their tutorials yet. I like to start using any program w/out tutorials. To see how intuitive it is out the gate. I only see a side x side slider comparison view in NoNo AI - but maybe there’s other options in there.

Unfortunately (???) my cameras rarely shoots noisy images - so I work with what I have to test.

I tried a CR2 (Canon proprietary) raw image 1st in NoNoise AI (ON1) then in this Topaz DAI beta.

Original view in NNAI (ON1) interface:

DeNoised preview in NNAI (ON1) interface:

Split Preview - ORIG/Denoised Preview in NNAI (ON1) interface:

Split Preview - ORIG/Clear Mode DeNoised Preview in DAI beta (Topaz) Standalone interface:

I finally got Affinity to work with DeNoise AI. Working with RAW files has to be converted to 16bit for DN to work. Now its working in Affinity.

Before DeNoise but processed with RAW setting

After working in DeNoise and exported into Affinity.

I too have done a quick comparison, but I DO have a lot of high ISO images.
This one was shot at ISO 12800.
I used Auto in all three cases (with DxO Pure Raw there’s no option anyway) but as it was a tad underexposed for the shadows, I edited all three further in On1 Photo Raw. No questions which is best!

With regard to this beta, there is still a lot of work to be done to get blotches created by

colour noise removed!

1 Like

Looks to me that Topaz DN AI wins out. I would think that On1 will add their DeNoise to their suite when they have another update. I just got the On1 email for a 14 day trail, not very long.

1 Like

Because ON1 put the banner Disclosure up saying their product is only optimized for raw images I felt compelled to immediately try a .jpg to see what (if anything) would happen.

I only ran all these using the Standalone products for ON1 & Topaz.

I hate the idea of my denoising processing software being limited to only do its best on raw images of any ilk - proprietary camera mfr., proprietary Adobe (dng), etc. I have too many old images that were just jpgs that I might wanna denoise. And, tons of .tiff’s.

Here’s a comparable comparison to what I did with the CR2 raw image above - this time for a landscape scene .jpg

ON1 NoNoise AI - Removes critical landscape details (tree branches, etc.) that makes my image look as if it was run thru Topaz Simplify. Split preview of Orig/denoised:

Topaz DAI - Clear Mode Preview - Set up to look like the ON1 in terms of side x side view (vs comparison 4-way view):

BTW - I liked “Clear” mode (DAI) best of all the Topaz DAI modes for all the tests I ran to do these prod. comparisons.

Very clearly so in all the tests I just ran on raw & jpg images!

And - unless I’m missing something (which is certainly possible) they have less flexibility for comparisons in the ON1 UI.

I don’t go nuts sharpening inside a denoising program b/c I do that sharpening in a program I figure is optimized for sharpening. So my tests may look a tad soft. I test as I would work in an actual workflow - sharpening later in the process after I do all my “creative” “artsy” stuff.

Test 1: (Very Noise RAW Image)

RAW: On1 NoNoise AI vs Topaz De Noise AI (Severe) vs DXO DeepPRIME

Test 2: (Extreme Noise RAW Image)

RAW: On1 NoNoise AI vs Topaz De Noise AI (Severe) vs DXO DeepPRIME

Test 3 (Very Noise JPG)

JPG: On1 NoNoise AI vs Topaz De Noise AI (Severe) vs Topaz De Noise AI (Clear)

Test 4 : (Extreme Noise JPG)

JPG: On1 NoNoise AI vs Topaz De Noise AI (Severe)vs Topaz De Noise AI (Clear)

I hope everyone is interested to see discussions about a beta product and how it compares to ON1 in the general discussion…

Continuing Comparison Tests of ON1 NoNoise AI vs Topaz DeNoise AI 3.2.02b from yesterday (7/6/21) — Note: my earlier posts were moved here by someone else (above) from their original post site.

Using Win 10 Pro PC.

Standalone versions of both programs.

Worked with a single Lumix raw image with ISO = 25,600 (night photo with heavy noise - incl. color noise)

Focus Today for me = Masking Functions/User Interface in the 2 products.

Bottom Line: ON1 NoNoise AI’s masking outperformed Topaz DeNoise AI’s masking functionality and interface options in my tests. The ON1 NoNoise AI masking experience is more feature-rich, and offers more flexible and easy-to-use masking functionality than Topaz DAI NoNoise AI’s masking more closely resembles the types of masking features/functions present in Topaz Studio. The one exception - I like the color paint on overlay feature that helps me see where I’m brushing in DAI.

Snips showing what I saw in the two competing products… ON1 NoNoise followed by Topaz DAI. Labelled in filenames.

Original Lumix Raw Image in ON1’s NoNoise AI Interface (see above for image specs):

Preview of Globally Denoised Raw Image in ON1’s NoNoise AI Interface:

Split Side x Side Preview (Orig/Denoised) in ON1 NoNoise AI UI:

Enlargeable BW Mask - Easy to View + Very Accurate “Edge Aware” Brush Tool in ON1 NoNoise AI masking UI (enabled by creating a separate layer + ability to lock denoising/sharpening values for masking in the NNAI masking interface):

Raw image with Selective Masking (sky was masked to show denoising, all ground elements were unmasked - to retain orig noise - for test purposes) in the ON1 NoNoise AI interface:

  • From a global image perspective, Topaz DAI using “Severe Noise” mode + add’l. Color Noise slider adjustment was comparable to ON1’s NoNoise AI result with this Lumix raw image. All tests were viewed at 100 & 200%. Brush masking was handled with full image elements visible in the UI; not zoomed in.

Topaz DAI 4-Way Comparison View of the Lumix Raw Image. Significant advantage to see all denoising options/modes together to pick one to work with:

Topaz DAI Side x Side Comparison View - ORIG & Severe Noise:

Topaz DAI - Masking Interface with Lumix Raw Image (for comparison vs ON1 NNAI:

Topaz DAI - Masking Brush Color Overlay Helps See Where Painting Mask over Image: [Edge Aware is set to on, but doesn’t conform well to edges - doesn’t matter if complex like leaves or smoother edges… unlike ON1 NNAI which seemed tuned to cling to appropriate edges]

Topaz DAI 3.2.0.b - Snip of Sky DeNoised and Ground Elements Masked to Retain Look of ORIG Image (for denoising comparison…):

** Disclaimer: I don’t like to over-sharpen in a denoising program. My goal is to get rid of unwanted noise without adding sharpening that would impact my creative post-processing workflow. After I add creative effects to images, then I sharpen - usually in a program like Topaz SAI or in Ps 2021 that has tools targeted to sharpening.

A comparison video that Matt Kloskowski just released today (Jy. 8, 2021).

FYI… Matt used to be a VP at ON1 for a short while. He’s also a user of Topaz DeNoise and has hawked it & gets promo code ‘benefits’ for doing so.

That said, in my own tests (and personal opinions) I think he & I are pretty close wrt observations and the opinions he expresses in his video (see link below).

Matt doesn’t address masking features/functions - as I did above. Probably b/c he is primarily a Lr (& Ps) user - he was one of Kelby’s original “Photoshop Guys”). If someone doesn’t use these products via a host program (Ps, Lr, Affinity, DxO, CaptureOne, etc.) or wants to work with a standalone program to handle raw images, then masking tool functionality & feature-richness - I believe - can be a big issue.

Here’s the link to Matt’s comparison of ON1 NoNoise AI vs Topaz DeNoise AI:

I thought this video on YouTube offered an interesting comparison among NoNoise DeNoise and Pure Raw. He did not test the masking features or a shot taken at especially high ISO. He though all three programmes did a good job, with some differences where one was better than another in one area and not in others etc.

1 Like

This shot was taken indoors at ISO5000.It’s an ARW file and in all three instances it was converted to DNG for denoising.
As I have noted before (and no one else seems to have commented on this or explained why) the DNG created by DeNoise AI is almost twice the size of the other two: 114 v 87 and 78 MB.
I set Denoise to severe but maybe it would have done a better job in standard mode. Look at the freckles and the shininess around the nose!

1 Like

In the past, DeNoise struggled with black tonalities at high ISO. It’s got better, but still not great. Again, I might have achieved a better result with DeNoise if I’d chosen a different model. NoNoise also struggles with the colour noise, even though the slider is set in auto mode at 100%. Pure Raw is totally automated so there are no sliders at all. However, it did by far the best job in reducing colour noise as well as retaining the textures of the black fabric.

DeNoise using GPU is the fastest of the 3 programmes on my MacBook Pro, Pure Raw the slowest: but there isn’t a huge difference.

This next and the previous image are ISO12800 and were shot hand-held in low light. All 3 using the auto setting (in Pure Raw there is no other option)
NoNoise still has a lot of luminance noise, which might be improved if I had used the slider. DeNoise might also be improvable using a different model and adjusting the sliders, but in severe mode, the black tonality is off and the texture of the fabric is intermittent.

I am an On1 user, so if there are improvements to NoNoise when it is incorporated in to On1 Photo Raw 2022, I’ll be very happy.
I have been using the trial of DxO Pure Raw and have used it to reprocess all my images from the past 4 years that have an ISO greater than 1600. Since it’s a standalone, I can use it in batch mode at the start of my workflow and then continue my processing from there. Quite a lot of my images also need the use of Sharpen AI, due to hand-holding in low light or the use of a zoom lens.


Has ON1 said whether NoNoise will remain a separate product option once incorporated into Photo Raw?

I’m curious b/c I’m wondering what the update implications are for folks who buy it now to use standalone or in competing host programs (like Ps).

I tried the NoNoise AI plugin (in Ps) during trial experiments last night. 1st time not standalone. It took what felt like 10 yrs for my image to be sent from inside Ps to the NoNoise standalone via Ps’ ON1 filter menu listing…

On1 has stated that NoNoise will become part of PhotoRaw in the new 2022 version later this year. Your best way to get these benefits is to use the subscription which is $90/year. For that you get the 2021 Photoraw + NoNoise AI. Later in the year you get the new 2022 version, all included. When renewal comes up you can either stick with it or buy the 2022 version.

1 Like

I want to restate something I said previously (but of course was ignored). If your photo has only light to moderate noise you may be better off processing it in Topaz Sharpen AI first. Sharpen does a credible job of reducing noise and is superior at sharpening (but can leave halos on some lines). Then go on to process the result. This works either as a stand alone or plugin (I haven’t compared the two) and you can skip Denoise AI.

1 Like

In Matt Kloskowski video (thanks for the link) he showed that NoNoise was much better using the RAW file so I would not presently send it from Photoshop.

1 Like


But my question was whether it will also remain a separate stand on its own product that folks can purchase/use alone & with other hosts or if it’s going to be a just integrated into Photo Raw sales model. :blush:

Yeah. Trying to work with it using the Ps plugin was ridiculously slow.

The NoNoise standalone was the inverse of DAI. NoNoise was longer to load images but faster processing them…in my tests so far (mix of photo types - proprietary raw, jpg, tiff.

I don’t know for sure but since they are selling it separate they should support it. I think it will be a separate product for a long time. You might ask the question on their Facebook page.

1 Like