DXO Prime vs. Topaz Denoise AI

Started 10 months ago | Discussions
Sagittarius Veteran Member • Posts: 6,450
DXO Prime vs. Topaz Denoise AI
2

1

2

ISO 51200. Same image has been processed in DXO PL prime and Topaz Denoise AI. Which one do you think is better #1 or #2?

-- hide signature --

Best regards

 Sagittarius's gear list:Sagittarius's gear list
Nikon D810 Nikon D500 Nikon AF-S Nikkor 70-200mm f/2.8G ED VR II Nikon AF-S Micro-Nikkor 105mm F2.8G IF-ED VR Nikon AF-S Nikkor 50mm F1.4G +4 more
stormyboy Contributing Member • Posts: 848
Question about your skills with the programs:
17

Can we assume that you have fiddled with all the possible adjustments enough to truly understand them, and have arrived at each version being the absolutely best version you can possibly make it? If not this is pretty meaningless and we would be better served if you uploaded your raw file so everyone could experiment. I'm sorry to be negative, but these sort of comparisons are so open to variable changes that I don't have much faith in them.

NAwlins Contrarian Veteran Member • Posts: 3,997
Different raw conversions, settings issues, and taste / trade-offs
10

ISO 51200. Same image has been processed in DXO PL prime and Topaz Denoise AI. Which one do you think is better #1 or #2?

I realize these sorts of comparisons are difficult to do well, but there's an overarching issue that the two images are just different in ways that can and probably will affect the appearance of noise:

* Image 2 is noticeably brighter than image 1.

* Image 2 is noticeably more saturated and/or shifted toward red compared to image 1.

* The sharpening differs between the two. Overall image 2 appears maybe more sharpened. To some extent this is a matter of taste, and upon brief review I'm not prepared to say either is clearly better-sharpened than the other, but in USM terms the radius / amount / threshold settings appear different.

Also, the results you get depend heavily on how you set the noise reduction controls. I presume (but don't know) that Topaz Denoise AI is similar, but with DxO PhotoLab Elite's Prime noise reduction, there are five different controls (Luminance, Chrominance, Low frequency, Dead pixels, and Maze). Personally I find I can usually produce results much more subjectively pleasing to me by changing the settings from the defaults--often changing them a good bit.

With those caveats and FWIW, my opinions of your tests:

* Image 1 has more coarse / low-frequency noise.

* Image 2 has more fine / high-frequency noise.

* Overall, subjectively, I think I prefer image 2, but that's mostly due to the distracting noise in the air / shadows in image 1, and image 1 is arguably more pleasing in the detailed image areas. Also, I think I might well change preferences based on magnification / print size.

But I might well reach different conclusions if given the opportunity to change the controls and settings. I know with DxO they make a huge difference, and I suspect with Topaz they do too.

 NAwlins Contrarian's gear list:NAwlins Contrarian's gear list
Nikon Coolpix S30 Canon PowerShot S110 Sony Alpha DSLR-A580 Sony DT 16-50mm F2.8 SSM Tamron SP 70-300mm F/4-5.6 Di USD +5 more
tokumeino Senior Member • Posts: 2,950
Re: DXO Prime vs. Topaz Denoise AI
2

#2 without question. I didn't even look a the full size image : the thumbnails is enough to judge.

Note that I'm not sure whether it can help to qualify a SW among the other, since many thing rely on parameters.

 tokumeino's gear list:tokumeino's gear list
Fujifilm XF 35mm F1.4 R Fujifilm XF 27mm F2.8 Canon G7 X II Fujifilm X-T3 Samyang 8mm F2.8 UMC Fisheye +6 more
jan pawlak Contributing Member • Posts: 619
Re: DXO Prime vs. Topaz Denoise AI
1

Sagittarius wrote:

1

2

ISO 51200. Same image has been processed in DXO PL prime and Topaz Denoise AI. Which one do you think is better #1 or #2?

No 1.

No. 1 has lower (normal) saturation, more details and is sharper

-- hide signature --
 jan pawlak's gear list:jan pawlak's gear list
Sony FE 24mm F1.4 GM Sony a7R III Sony FE 12-24mm F4 Sony FE 70-200mm F2.8 GM OSS Zeiss Batis 40mm F2 CF
Digital Nigel Forum Pro • Posts: 10,500
Re: DXO Prime vs. Topaz Denoise AI
7

I've done similar tests, and shown the results here. Denoise alone produced strange results, and using Denoise AI directly on a RAW image is a bad idea.

I came to the conclusion that PRIME was much better for day to day use, but the occasional high ISO image benefited from being processed from RAW with PRIME, and then having Denoise AI applied. That's a tedious process, so only worth doing for the few images that really need it.

Here's an example:

https://www.flickr.com/photos/recliner/albums/72157708128725115

Someone else commented that Denoise AI was a useful way of 'polishing' very noisy images that had first been processed with PRIME. I think that's a good way of describing it, and that's what I now do.

 Digital Nigel's gear list:Digital Nigel's gear list
Panasonic FZ1000 Canon PowerShot G7 X Panasonic ZS100 Nikon Coolpix P900 Sony RX10 III +18 more
Ernie Misner
Ernie Misner Veteran Member • Posts: 5,795
Re: DXO Prime vs. Topaz Denoise AI

Digital Nigel wrote:

I've done similar tests, and shown the results here. Denoise alone produced strange results, and using Denoise AI directly on a RAW image is a bad idea.

I came to the conclusion that PRIME was much better for day to day use, but the occasional high ISO image benefited from being processed from RAW with PRIME, and then having Denoise AI applied. That's a tedious process, so only worth doing for the few images that really need it.

Here's an example:

https://www.flickr.com/photos/recliner/albums/72157708128725115

Someone else commented that Denoise AI was a useful way of 'polishing' very noisy images that had first been processed with PRIME. I think that's a good way of describing it, and that's what I now do.

I agree with your results and thanks for the flickr link!

-- hide signature --

Ernie Misner
http://www.flickr.com/photos/erniemisner/
"I bet we'll find hints of extraterrestrial life before 2030" - Jamie Shreeve, Popular Science

 Ernie Misner's gear list:Ernie Misner's gear list
Nikon D810 Nikon AF-S Nikkor 70-200mm F4G ED VR Nikon AF-S Nikkor 24-70mm f/2.8G ED Nikon AF-S Nikkor 50mm F1.4G Tamron SP 90mm F2.8 Di VC USD Macro +6 more
David5833 Senior Member • Posts: 1,319
Re: DXO Prime vs. Topaz Denoise AI

Photo 2 looks much better to me, no blobs or smearing of details.

 David5833's gear list:David5833's gear list
Nikon D810 Nikon D7500 Olympus OM-D E-M10 II Nikon AF Micro-Nikkor 200mm f/4D ED-IF Nikon 200-500mm F5.6E ED VR +12 more
Digital Nigel Forum Pro • Posts: 10,500
Re: Different raw conversions, settings issues, and taste / trade-offs

NAwlins Contrarian wrote:

I<snip>

Also, the results you get depend heavily on how you set the noise reduction controls. I presume (but don't know) that Topaz Denoise AI is similar, but with DxO PhotoLab Elite's Prime noise reduction, there are five different controls (Luminance, Chrominance, Low frequency, Dead pixels, and Maze). Personally I find I can usually produce results much more subjectively pleasing to me by changing the settings from the defaults--often changing them a good bit.

No, the controls are completely different in the Topaz AI products. Fundamentally, you are letting the AI system choose the method to apply, based on its training on many other noisy and noise-reduced images. In effect, it tries to produce the same effect as the best 'fixed' images it was trained on. For that reason, the results are very unpredictable, both from image to image, and even in different parts of the image.

These are the controls:

Topaz DeNoise AI controls

 Digital Nigel's gear list:Digital Nigel's gear list
Panasonic FZ1000 Canon PowerShot G7 X Panasonic ZS100 Nikon Coolpix P900 Sony RX10 III +18 more
OP Sagittarius Veteran Member • Posts: 6,450
Re: DXO Prime vs. Topaz Denoise AI

The same image has been sent from LR to both SW. #1 is by Topaz, #2 is by DXO PL1.

-- hide signature --

Best regards

 Sagittarius's gear list:Sagittarius's gear list
Nikon D810 Nikon D500 Nikon AF-S Nikkor 70-200mm f/2.8G ED VR II Nikon AF-S Micro-Nikkor 105mm F2.8G IF-ED VR Nikon AF-S Nikkor 50mm F1.4G +4 more
sobrien
sobrien Senior Member • Posts: 1,154
Re: DXO Prime vs. Topaz Denoise AI
1

#2 looks clearly better to me. Very good detail/noise ratio.

#1 has that smudgy look you get when NR is cranked up too high, IMO.

-- hide signature --

"The simple things are also the most extraordinary things and only the wise can see them."
https://www.flickr.com/photos/135843555@N03/

 sobrien's gear list:sobrien's gear list
Canon EOS R Canon EOS M50 Canon EF-S 18-135mm F3.5-5.6 IS STM Canon EF 40mm f/2.8 STM Canon EF 100-400mm F4.5-5.6L IS II +18 more
David5833 Senior Member • Posts: 1,319
Re: DXO Prime vs. Topaz Denoise AI

Sagittarius wrote:

The same image has been sent from LR to both SW. #1 is by Topaz, #2 is by DXO PL1.

What value was the "remove noise" slider set to in Topaz AI?

 David5833's gear list:David5833's gear list
Nikon D810 Nikon D7500 Olympus OM-D E-M10 II Nikon AF Micro-Nikkor 200mm f/4D ED-IF Nikon 200-500mm F5.6E ED VR +12 more
NAwlins Contrarian Veteran Member • Posts: 3,997
Re: Different raw conversions, settings issues, and taste / trade-offs
1

No, the controls are completely different in the Topaz AI products. Fundamentally, you are letting the AI system choose the method to apply, based on its training on many other noisy and noise-reduced images.

Okay, but it looks like even with Topaz Denoise AI, the user still can (and probably should) adjust / balance / make subjective judgments about the best relationship among control settings for noise, sharpness, and details--right?

 NAwlins Contrarian's gear list:NAwlins Contrarian's gear list
Nikon Coolpix S30 Canon PowerShot S110 Sony Alpha DSLR-A580 Sony DT 16-50mm F2.8 SSM Tamron SP 70-300mm F/4-5.6 Di USD +5 more
NAwlins Contrarian Veteran Member • Posts: 3,997
So not the current DxO Prime?

The same image has been sent from LR to both SW. #1 is by Topaz, #2 is by DXO PL1.

If I understand correctly, you used DxO PhotoLab 1, which was replaced by PhotoLab 2 eight months ago. DxO claims to have improved Prime noise reduction between PL1 and PL2; how much and how, I don't know. But it seems like that's another methodological issue.

 NAwlins Contrarian's gear list:NAwlins Contrarian's gear list
Nikon Coolpix S30 Canon PowerShot S110 Sony Alpha DSLR-A580 Sony DT 16-50mm F2.8 SSM Tamron SP 70-300mm F/4-5.6 Di USD +5 more
tokumeino Senior Member • Posts: 2,950
Re: DXO Prime vs. Topaz Denoise AI
1

jan pawlak wrote:

No 1.

No. 1 has lower (normal) saturation, more details and is sharper

Why even looking at details ? Isn't it obvious that all the background suffers a very heavy cotton pattern ?

 tokumeino's gear list:tokumeino's gear list
Fujifilm XF 35mm F1.4 R Fujifilm XF 27mm F2.8 Canon G7 X II Fujifilm X-T3 Samyang 8mm F2.8 UMC Fisheye +6 more
Digital Nigel Forum Pro • Posts: 10,500
Re: Different raw conversions, settings issues, and taste / trade-offs

NAwlins Contrarian wrote:

No, the controls are completely different in the Topaz AI products. Fundamentally, you are letting the AI system choose the method to apply, based on its training on many other noisy and noise-reduced images.

Okay, but it looks like even with Topaz Denoise AI, the user still can (and probably should) adjust / balance / make subjective judgments about the best relationship among control settings for noise, sharpness, and details--right?

Yes, very much so. Maybe the AI will get good enough one day to pick its own optimum settings?

I just use trial and error, which can be rather a tedious process, given the very slow processing speeds even on my very fast workstation. It's hard to become expert in the software, given my infrequent use of it, and the big changes in each new release.  It feels almost like you're participating in an experiment.

 Digital Nigel's gear list:Digital Nigel's gear list
Panasonic FZ1000 Canon PowerShot G7 X Panasonic ZS100 Nikon Coolpix P900 Sony RX10 III +18 more
Lan Senior Member • Posts: 1,526
Re: DXO Prime vs. Topaz Denoise AI

ISO 51200. Same image has been processed in DXO PL prime and Topaz Denoise AI. Which one do you think is better #1 or #2?

I have a strong preference for #2, the most obvious difference is some really nasty wide chroma blotching in #1 which is visible even at thumbnail size, but there's also more detail in #2 when viewed at 100%.

tokumeino Senior Member • Posts: 2,950
Re: DXO Prime vs. Topaz Denoise AI

Lan wrote:

ISO 51200. Same image has been processed in DXO PL prime and Topaz Denoise AI. Which one do you think is better #1 or #2?

I have a strong preference for #2, the most obvious difference is some really nasty wide chroma blotching in #1 which is visible even at thumbnail size, but there's also more detail in #2 when viewed at 100%.

Did you look at the picture before zooming 2:1 ? The background is full of huge 'cotton' artifacts. How can people miss that ? Are you using a camera or a microscope ?

 tokumeino's gear list:tokumeino's gear list
Fujifilm XF 35mm F1.4 R Fujifilm XF 27mm F2.8 Canon G7 X II Fujifilm X-T3 Samyang 8mm F2.8 UMC Fisheye +6 more
Lan Senior Member • Posts: 1,526
Re: DXO Prime vs. Topaz Denoise AI

tokumeino wrote:

Lan wrote:

ISO 51200. Same image has been processed in DXO PL prime and Topaz Denoise AI. Which one do you think is better #1 or #2?

I have a strong preference for #2, the most obvious difference is some really nasty wide chroma blotching in #1 which is visible even at thumbnail size, but there's also more detail in #2 when viewed at 100%.

Did you look at the picture before zooming 2:1 ? The background is full of huge 'cotton' artifacts. How can people miss that ? Are you using a camera or a microscope ?

You mean the first thing I mentioned about the images

tokumeino Senior Member • Posts: 2,950
Re: DXO Prime vs. Topaz Denoise AI

Lan wrote:

tokumeino wrote:

Lan wrote:

ISO 51200. Same image has been processed in DXO PL prime and Topaz Denoise AI. Which one do you think is better #1 or #2?

I have a strong preference for #2, the most obvious difference is some really nasty wide chroma blotching in #1 which is visible even at thumbnail size, but there's also more detail in #2 when viewed at 100%.

Did you look at the picture before zooming 2:1 ? The background is full of huge 'cotton' artifacts. How can people miss that ? Are you using a camera or a microscope ?

You mean the first thing I mentioned about the images

Sorry, I've indeed not read past "I have a strong preference for #2". The thing is : blotching appears in #2, not #1. Didn't you mix the numbers ?

 tokumeino's gear list:tokumeino's gear list
Fujifilm XF 35mm F1.4 R Fujifilm XF 27mm F2.8 Canon G7 X II Fujifilm X-T3 Samyang 8mm F2.8 UMC Fisheye +6 more
Keyboard shortcuts:
FForum MMy threads