DxO_OpticsPro 11 & 9 noise reduction issues

gordonpritchard said:

I've tried the DxO_OpticsPro 11 & 9 Prime noise reduction filter on a RAW file by turning off all filters except for the Prime noise reduction filter which was set to Auto.

The result is underwhelming in that the image ends up with artefacts where strong light areas abut dark. The same RAW file processed with PShop is smooth. Below is what this looks like (view at original size - original at top blow up sections below):


PShop left - DXO right

The smoother PShop version allows for image sharpening and cropping but the DXO version, IMHO, does not.

Anyone else experience this? Is there a fix or is this DXO's normal?

BTW, this is the case no matter what format I export the image in from DXO.

Click to expand...

Landscape & Wedding Photographer - Jacques Cornell

Jacques Cornell Photography - landscape, travel, portrait and art reproduction

jacquescornell.photography

Happening Photos - Event Photography in New York City

Event Photography in New York City and Around the World - conferences, sales meetings, fundraisers and more.

happening.photos

Jacques Cornell said:

gordonpritchard said:

I've tried the DxO_OpticsPro 11 & 9 Prime noise reduction filter on a RAW file by turning off all filters except for the Prime noise reduction filter which was set to Auto.

The result is underwhelming in that the image ends up with artefacts where strong light areas abut dark. The same RAW file processed with PShop is smooth. Below is what this looks like (view at original size - original at top blow up sections below):


PShop left - DXO right

The smoother PShop version allows for image sharpening and cropping but the DXO version, IMHO, does not.

Anyone else experience this? Is there a fix or is this DXO's normal?

BTW, this is the case no matter what format I export the image in from DXO.

Click to expand...

The PS version doesn't look "smooth" to me - it looks blurry. The DxO version retains more fine detail.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

gordonpritchard said:

Jacques Cornell said:

gordonpritchard said:

I've tried the DxO_OpticsPro 11 & 9 Prime noise reduction filter on a RAW file by turning off all filters except for the Prime noise reduction filter which was set to Auto.

The result is underwhelming in that the image ends up with artefacts where strong light areas abut dark. The same RAW file processed with PShop is smooth. Below is what this looks like (view at original size - original at top blow up sections below):


PShop left - DXO right

The smoother PShop version allows for image sharpening and cropping but the DXO version, IMHO, does not.

Anyone else experience this? Is there a fix or is this DXO's normal?

BTW, this is the case no matter what format I export the image in from DXO.

Click to expand...

The PS version doesn't look "smooth" to me - it looks blurry. The DxO version retains more fine detail.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

Look closely. The DXO version has artefacts - not fine detail - that the PShop version doesn't have. The PShop "blurriness" disappears with sharpening (which I didn't do). Sharpening the DXO version just exaggerates the artefacts. The DXO version seems to have have some kind of sharpening applied when DXO processed the image.

Here's what sharpening (same amount to both) does (PShop left/DXO right):

Click to expand...

Jacques Cornell said:

This looks like user error to me. DxO-converted RAWs need far less USM than RAWs converted in other apps. How familiar are you with DxO? I process low-light ISO 3200 MFT RAWs all the time and never see anything near this bad. If you want to post the original, I'll have a go with DxO OP 10.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

gordonpritchard said:

gordonpritchard said:

This looks like user error to me. DxO-converted RAWs need far less USM than RAWs converted in other apps. How familiar are you with DxO? I process low-light ISO 3200 MFT RAWs all the time and never see anything near this bad. If you want to post the original, I'll have a go with DxO OP 10.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

I did the sharpening to exaggerate the artefact issue.

Click to expand...

gordonpritchard said:

It seems like DXO is applying some kind of sharpening when it processes the image. I don't know how to turn that off.

I've put 2 RAW images into Dropbox that you can download:

https://www.dropbox.com/sh/zz01a48i2fj045h/AAB6VUc0I6vEAoDBqoNnLM0ja?dl=0

Both images end up with similar artefacts.

It would be great if you could process them in DXO and post the .dop file and/or the settings you use. If you can post the high res version of your result and/or put it into my dropbox folder that would be great.

You could email me the .dop file here: pritchardgordon (@) gmail (dot) com

Thanks for your efforts.

Click to expand...

Jacques Cornell said:

gordonpritchard said:

Jacques Cornell said:

This looks like user error to me. DxO-converted RAWs need far less USM than RAWs converted in other apps. How familiar are you with DxO? I process low-light ISO 3200 MFT RAWs all the time and never see anything near this bad. If you want to post the original, I'll have a go with DxO OP 10.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

I did the sharpening to exaggerate the artefact issue.

Click to expand...

The problem is, you exaggerated the artifact issue. Applying the same sharpening to both is inappropriate.

Jacques Cornell said:

It seems like DXO is applying some kind of sharpening when it processes the image. I don't know how to turn that off.

I've put 2 RAW images into Dropbox that you can download:

https://www.dropbox.com/sh/zz01a48i2fj045h/AAB6VUc0I6vEAoDBqoNnLM0ja?dl=0

Both images end up with similar artefacts.

It would be great if you could process them in DXO and post the .dop file and/or the settings you use. If you can post the high res version of your result and/or put it into my dropbox folder that would be great.

You could email me the .dop file here: pritchardgordon (@) gmail (dot) com

Thanks for your efforts.

Click to expand...

Given that this image is super-noisy, looking at more than 100% seems inappropriate unless you plan to print it really big - 16"x24" or larger.

I worked the image in both LR and DxO, trying to find a satisfying balance between NR and detail retention. Below are the best results I could eke out of both. For the LR conversion I used the settings shown below. For the DxO conversion I simply applied PRIME NR at a value of 30.

The results are very, very close, and choosing between them is really a matter of taste. The DxO image took almost no effort to produce, while the LR image required quite a bit of experimentation to find the best settings.


Image processed in LR with the settings shown below.




Image processed in DxO Optics Pro 10 Elite with PRIME NR set to 30.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

gordonpritchard said:

Jacques Cornell said:

gordonpritchard said:

gordonpritchard said:

This looks like user error to me. DxO-converted RAWs need far less USM than RAWs converted in other apps. How familiar are you with DxO? I process low-light ISO 3200 MFT RAWs all the time and never see anything near this bad. If you want to post the original, I'll have a go with DxO OP 10.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

I did the sharpening to exaggerate the artefact issue.

Click to expand...

The problem is, you exaggerated the artifact issue. Applying the same sharpening to both is inappropriate.

gordonpritchard said:

It seems like DXO is applying some kind of sharpening when it processes the image. I don't know how to turn that off.

I've put 2 RAW images into Dropbox that you can download:

https://www.dropbox.com/sh/zz01a48i2fj045h/AAB6VUc0I6vEAoDBqoNnLM0ja?dl=0

Both images end up with similar artefacts.

It would be great if you could process them in DXO and post the .dop file and/or the settings you use. If you can post the high res version of your result and/or put it into my dropbox folder that would be great.

You could email me the .dop file here: pritchardgordon (@) gmail (dot) com

Thanks for your efforts.

Click to expand...

Given that this image is super-noisy, looking at more than 100% seems inappropriate unless you plan to print it really big - 16"x24" or larger.

I worked the image in both LR and DxO, trying to find a satisfying balance between NR and detail retention. Below are the best results I could eke out of both. For the LR conversion I used the settings shown below. For the DxO conversion I simply applied PRIME NR at a value of 30.

The results are very, very close, and choosing between them is really a matter of taste. The DxO image took almost no effort to produce, while the LR image required quite a bit of experimentation to find the best settings.


Image processed in LR with the settings shown below.




Image processed in DxO Optics Pro 10 Elite with PRIME NR set to 30.

--
The way to make a friend is to act like one.
http://jacquescornell.photography
http://happening.photos

Click to expand...

Like I said earlier, I applied the same sharpening to both to exaggerate the artifact issue so that it would be clear to see. It is not what I would normally do with an image. You're getting the same artefacts that I was getting so I guess I wasn't doing anything wrong after all.

Below is my version on the left as processed in PShop beside the version you did in DXO. You can see that despite the noise PShop gave a better result despite the Prime noise reduction in DXO which they tout as being hands down the best. I think not. Very disappointing.


PShop left - DXO right

--
My photos: http://www.gordonpritchard.blogspot.com/
My sketches: http://thesoftpen.blogspot.com/

Click to expand...

Landscape & Wedding Photographer - Jacques Cornell

Jacques Cornell Photography - landscape, travel, portrait and art reproduction

jacquescornell.photography

Happening Photos - Event Photography in New York City

Event Photography in New York City and Around the World - conferences, sales meetings, fundraisers and more.

happening.photos

Jacques Cornell said:

Your Photoshop version looks better than I could get out of either DxO or LR. What did you do to it and how long did it take?

--

Click to expand...

gordonpritchard said:

Jacques Cornell said:

Your Photoshop version looks better than I could get out of either DxO or LR. What did you do to it and how long did it take?

--

Click to expand...

I can only tell you in general terms as I don't remember exactly what I did in this case.

I process RAW files in PShop in order to get a clean, artifact-free, starting point to work with the image (that's where DXO failed). I don't try to make a pretty picture using the RAW image filter. So no clarity/saturation/vibrance/sharpening, etc adjustments. I just restore highlights, open midtone/shadows, set white balance, fix chromatic aberrations, and reduce noise.

For this image I likely would have then done some dodging and burning to get better tonal separation in the faces and then used the high pass filter to sharpen.

Click to expand...

Jacques Cornell said:

gordonpritchard said:

Jacques Cornell said:

Your Photoshop version looks better than I could get out of either DxO or LR. What did you do to it and how long did it take?

--

Click to expand...

I can only tell you in general terms as I don't remember exactly what I did in this case.

I process RAW files in PShop in order to get a clean, artifact-free, starting point to work with the image (that's where DXO failed). I don't try to make a pretty picture using the RAW image filter. So no clarity/saturation/vibrance/sharpening, etc adjustments. I just restore highlights, open midtone/shadows, set white balance, fix chromatic aberrations, and reduce noise.

For this image I likely would have then done some dodging and burning to get better tonal separation in the faces and then used the high pass filter to sharpen.

Click to expand...

Well, Lightroom did no better than DxO, and that uses the same Adobe Camera RAW engine as Photoshop.

Click to expand...

gordonpritchard said:

Jacques Cornell said:

gordonpritchard said:

Jacques Cornell said:

Your Photoshop version looks better than I could get out of either DxO or LR. What did you do to it and how long did it take?

--

Click to expand...

I can only tell you in general terms as I don't remember exactly what I did in this case.

I process RAW files in PShop in order to get a clean, artifact-free, starting point to work with the image (that's where DXO failed). I don't try to make a pretty picture using the RAW image filter. So no clarity/saturation/vibrance/sharpening, etc adjustments. I just restore highlights, open midtone/shadows, set white balance, fix chromatic aberrations, and reduce noise.

For this image I likely would have then done some dodging and burning to get better tonal separation in the faces and then used the high pass filter to sharpen.

Click to expand...

Well, Lightroom did no better than DxO, and that uses the same Adobe Camera RAW engine as Photoshop.

Click to expand...

Click to expand...

Alan H 101 said:

The A77ii raws are likely to suffer more from pixel level artifacting when pushed because of Sonys compressed raw format.

http://www.dpreview.com/articles/28...the-cooked-pulling-apart-sony-raw-compression

It's probably harder for DxOs algorithm to distinguish between real detail and false detail created by the compression as it depends on raw data to work.

Click to expand...

gordonpritchard said:

Alan H 101 said:

The A77ii raws are likely to suffer more from pixel level artifacting when pushed because of Sonys compressed raw format.

http://www.dpreview.com/articles/28...the-cooked-pulling-apart-sony-raw-compression

It's probably harder for DxOs algorithm to distinguish between real detail and false detail created by the compression as it depends on raw data to work.

Click to expand...

Perhaps. But the artifacts also happen when noise reduction is turned off in DXO. Neither PShop nor Affinity photo have this problem with Sony RAW files.

Click to expand...

gordonpritchard said:

Jacques Cornell said:

Your Photoshop version looks better than I could get out of either DxO or LR. What did you do to it and how long did it take?

Click to expand...

I can only tell you in general terms as I don't remember exactly what I did in this case.

Click to expand...

sybersitizen said:

gordonpritchard said:

Jacques Cornell said:

Your Photoshop version looks better than I could get out of either DxO or LR. What did you do to it and how long did it take?

Click to expand...

I can only tell you in general terms as I don't remember exactly what I did in this case.

Click to expand...

I would encourage you to try very hard to remember, or recreate the result and document it. The version on the left shows not just lower noise and lack of artifacts. It appears to have been made from a much higher resolution original image - at least double, or maybe quadruple the resolution of the one on the right. The difference in the rendering of the faces and clothing and bare branches is nothing short of shocking. The only 'weirdness' is in the rendering of lights in the background, which looks extremely artificial. In the absence of a detailed explanation, I would surmise that the image must have been treated with some miraculous state-of-the-art plug-in.

Click to expand...

gordonpritchard said:

So, I managed to recreate it the way I think I originally did.

AFAIK I opened the original RAW file in PShop and removed as much of the noise as I could there.

Click to expand...

gordonpritchard said:

Then I launched the Topaz deNoise plugin and further reduced the noise. (Apparently you're not supposed to do a double denoise like that when using Topaz). Then I sharpened the image using the High Pas filter technique (and denoised that filter layer when setting it to Overlay).

Click to expand...

gordonpritchard said:

Since artefacts were not introduced by either PShop or Topaz - the final image was quite "clean" so sharpening retained a smooth image. On the other hand, it seems that no matter what the settings, DxO does some harsh sharpening to the image even when all settings are off. That introduces the artefacts that I see and, to me, kills the app.

Click to expand...

sybersitizen said:

gordonpritchard said:

So, I managed to recreate it the way I think I originally did.

AFAIK I opened the original RAW file in PShop and removed as much of the noise as I could there.

Click to expand...

Is that the noise reduction offered within Adobe Camera RAW or outside of it? What version of Photoshop?

Click to expand...