Nikon Z 7II ISO 10000 & 16000

[Abitconfused]

I have actually enjoyed using higher ISO settings now that my wealthy uncle gave me a Nikon Z 7II. I explore here dealing with noise in post-production. I am certain much can be added to the discussion.
https://edruthphoto.com/iso-10000/

[DanK]

Interesting. At this size, these look very clean for the IS0 values used.

I have one question and one suggestion.

the suggestion is that you provide links to the full size images, ideally both before and after the noise reduction. As Manfred and I pointed out in an earlier thread, downsizing masks all manner of sins, and these are downsized from the original (45MPX? I don't know the camera) to 2.67 MPX. If not full size, at least much larger. Without the larger images, three things are confounded: the level of noise in the original, the effect of noise reduction, and the masking effect of downsizing.

The question is this:

Here we employ DxO PhotoLab 6 Elite, which provides three functions that aid us in addressing noise during post-production. First, DxO’s superb lens evaluations and Raw file processing algorithms glean abundant detail from our Raw files. Second, DxO achieved a major milestone in post-production color management with its recently introduced “DxO Wide Gamut” color space. This color space is evolutionary in that it provides a “working color space” that exceeds the color gamut of the popular Adobe RGB color space and embraces all of that which is useful in the very large ProPhoto RGB color space. Third, under “Detail > DxO Denoising Technologies” DxO PhotoLab 6 Elite provides “DeepPRIME XD.”

Two questions, actually. Can you explain why using a wider gamut helps with noise reduction? I'm not saying it doesn't, but I don't understand why it would. And is this any broader than the ProPhoto gamut that is available in Photoshop and some other software, or the "Melissa" variant of ProPhoto that is always used in Lightroom?

[Manfred M]

Further to Dan's comments; if I read your article correctly, all of these images have gone through some heavy duty noise reduction via the DxO Photolab's noise reduction algorithms. Sorry, if you want to write about how clean the camera's sensor is, you had better post it with NO noise reduction as well, otherwise there is no way of knowing whether the image quality comes from a great sensor or if it is the result of heavy-duty noise reduction. As Dan has mentioned, full size images (or a crop of the image that shows 100% full size) are needed to give the viewer a true indication of image quality.

The new colour space DxO uses appears to be ISO 22028-2:2013 ROMM RGB; a narrower gamut than the more commonly used ProPhoto RGB. It would be interesting to understand DxO's rationale for using it. It will have to be converted to a colour space that other software supports.

Further to Dan's comment, there is no link between noise reduction and colour space

[Abitconfused]

For some time, I did my own scripting in HTML 4 and 5 but switched to WordPress as I could not keep up with coding for all the handheld devices. In WordPress, I must struggle through the process to supply images larger than the highly compressed WEBP images which are default on my site host. Gone are the carefree days of HTML 4. But I have posted under: https://edruthphoto.com/reference/ three images, however miniaturized, that do betray the noise being addressed as stated above. Please bear with me.

As to DxO Wide Gamut, you are correct to inquire further. As is said in court, "Objection Your Honor, lack of foundation, argumentative." I can only say that I BELIEVE I saw an improvement. Additionally, Wide Gamut, if I recall correctly, is as large as ProPhoto but without the "imaginary" colors that ProPhoto projects beyond the RGB color model. Lightroom is, to a snob like me, Photoshop on trainer wheels. So, while I know of Melissa, we don't visit.

[Manfred M]

As to DxO Wide Gamut, you are correct to inquire further. As is said in court, "Objection Your Honor, lack of foundation, argumentative." I can only say that I BELIEVE I saw an improvement. Additionally, Wide Gamut, if I recall correctly, is as large as ProPhoto but without the "imaginary" colors that ProPhoto projects beyond the RGB color model. Lightroom is, to a snob like me, Photoshop on trainer wheels. So, while I know of Melissa, we don't visit.

With a wide gamut screen, you will be able to see Adobe RGB colours or P3 RGB colours (depending on the panel the manufacturer used), but anything wider you can't see unless you are using a higher end photo printer, which on glossy photo papers has a gamut into some of the ProPhoto hues.

It bothers me when people suggest ProPhoto shows "imaginary colours" as this is simply not technically the case. ProPhoto can handle colours that are outside of human visual range, but that does not make them imaginary. Just to complicate matters even more, I am not aware of any device that can reproduced these hues. If we can't see them or can't reproduce them, who really cares.

Sorry to disagree with you again, Lightroom and Adobe Camera Raw are siblings. The same engine drives both and most of the Lightroom Develop Module is duplicated in Camera Raw / Photoshot Camera Raw Filter. ACR / LR are parametric editors and PS is a pixel based editor; they work in completely different ways.

[Abitconfused]

I don't really disagree with your assessment. I imagine at some far off juncture all visible colors will be available from camera to print. I think DxO did something bold here and it seems to fit the thrust of their interpretation of a Raw processor. True Lightroom and Photoshop are siblings but Photoshop features such as masking, Edit > Transform, and layers are so very convenient.

[Abitconfused]

An interesting post on the subject here: https://www.dpreview.com/forums/post/66530350

[DanK]

Thanks for posting the noisy versions. Unfortunately, I couldn't find comparisons at my end, as I wasn't able to find any saved images with ISO>3200.

However, your noisy images reminded me of a fourth factor that you have confounded: exposure level. In general, the lower the exposure, the lower the S/N ratio, that is, the greater the noise, and as you increase ISO, this becomes more apparent. Your Dagny's image is severely underexposed, other than the light fixture. (Look at the histogram.) That's why it looks so much noisier than the others you posted.

Here are a couple of images from a series I posted here years ago to illustrate this. I used a first-generation Canon 7D, which was notorious for getting noisy fast as ISO was increased. All of these are at ISO 3200. Unfortunately, I don't have them at a higher ISO, but they illustrate the point. you will need to enlarge these in the lightbox by double clicking.

First, here's one that's very well exposed, with the histogram extending almost to the right-hand edge. The lightbox will take you to approximately the same size you posted.



Here's the identical shot, 4 stops underexposed (similar to most of your Dagny's shot), then brightened to be comparable to the previous.



Now, let's look at a 100% crop from the well-exposed one, to show that there is noise present that was masked by downsizing. You'll need to double-click to get the lightbox to show it full size.



The contrasts would be more striking if I increased ISO 2 or 3 more stops, as you have in your photos.

I don't think Manfred's point was that LR and Photoshop are related. I believe it was that if you are using Photoshop to render raw images, you are effectively using LR, as ACR and LR are the same rendering engine. They have a different interface, but they are doing precisely the same thing, unless you set ACR to use 8-bit processing.

I'm not sure why DXO's offering a gamut wider than Adobe RGB is so bold, when other software has been offering that for many years. I think it's a great step, however, for people who are using DXO for other reasons.

I think the most useful way to teach people how to handle noise is to separate the various things they can do. They need to know the limitations of their particular camera (not clearly illustrated in your post because of the confounding), but there isn't much they can do about that other than try to avoid ISOs that are problematic for that camera. But beyond that, in many cases, they can use exposure to minimize noise, as the images above show. I do a lot of night photography, and I almost never use noise reduction other than in-camera long-exposure noise reduction, which is a whole different issue than the random noise in your post. Sometimes, however, there is no choice but to push ISO, and then the final step is to learn how to use various types of noise reduction if there's no avoiding it. The student should learn to control these variables.

For example, here's a 7-minute exposure at midnight with no noise reduction:



I only had the option of a 7-minute exposure because it was a windless night. But in urban night photography, one usually has stationary subjects, and wind isn't usually an issue. Here's a 20-second exposure with no noise reduction:



In urban night photography, the exposure issue is often very different: too large a dynamic range because of artificial lighting. That's the issue with your Dagny's photo: it's well exposed for the light, which would be blown out if you had exposed for the walls. I would normally not expose for the light and then hope for good noise reduction. I would expose for the shadows and either let the light blow out or bracket exposures. Here's an example with both: I allowed the street lights to blow out but used exposure bracketing to get a reasonable range of exposure for the rest. Again, no noise reduction.

[Abitconfused]

Thank you, a very well done and educational addition to the discussion. Although I like DxO PhotoLab Elite quite a bit, I also use the "Camera Raw Filter" in Photoshop specifically for the Clarity and Dehaze functions. Often it is what we become most familiar with that we applaud. It may be best to switch to Manual and closely monitor the histogram when shooting at night. In fact, I may have been doing that and fumbled the underexposure. I do try to expose to the right as prudent. DxO did a masterful job of saving that underexposure. I suppose it is premature to lament "Alas, poor Yorick!" regarding ProPhoto RGB. But DxO continues to amaze me with their dedication to innovation.

[pnodrog]

For some time, I did my own scripting in HTML 4 and 5 but switched to WordPress as I could not keep up with coding for all the handheld devices. In WordPress, I must struggle through the process to supply images larger than the highly compressed WEBP images which are default on my site host. Gone are the carefree days of HTML 4. But I have posted under: https://edruthphoto.com/reference/ three images, however miniaturized, that do betray the noise being addressed as stated above. Please bear with me.

As to DxO Wide Gamut, you are correct to inquire further. As is said in court, "Objection Your Honor, lack of foundation, argumentative." I can only say that I BELIEVE I saw an improvement. Additionally, Wide Gamut, if I recall correctly, is as large as ProPhoto but without the "imaginary" colors that ProPhoto projects beyond the RGB color model. Lightroom is, to a snob like me, Photoshop on trainer wheels. So, while I know of Melissa, we don't visit.

Sorry I don't think it's a matter of being a snob. It's just a lack of appreciation of what the two software applications are intended for.

[Abitconfused]

Truly, upon reflection, I am not a snob! A tad vain perhaps...

[pnodrog]

Truly, upon reflection, I am not a snob! A tad vain perhaps...

I assumed the comment was a bit tongue in cheek but I felt obliged to severely reprimand you...

[stuck]

...The new colour space DxO uses appears to be ISO 22028-2:2013 ROMM RGB; a narrower gamut than the more commonly used ProPhoto RGB...

Please correct me if I'm wrong but I thought ISO 22028-2:2013 ROMM was the same as ProPhoto, see here:
https://www.color.org/chardata/rgb/rommrgb.xalter
and here:
https://en.wikipedia.org/wiki/ProPhoto_RGB_color_space
and here, where a staff member at Affinity Photo is posting:
https://forum.affinity.serif.com/ind...comment-159117

...It would be interesting to understand DxO's rationale for using it....

Indeed but DxO are saying almost nothing about their wide gamut colour space. Consequently there is a lot of confused and, as far as I can make out, much misguided discussions about colour management in general and DxO's wide gamut in particular on the DxO forum. If you have a spare lifetime you can try and wade through some of that chatter, e.g. here:
https://feedback.dxo.com/t/question-...or-space/28617
https://feedback.dxo.com/t/yet-more-...usion/27879/40
https://feedback.dxo.com/t/dxo-pl6-color-space/29035
DxO's aim seems to be to cover Pointer's Gamut and to do that that they might be using something that is close to Rec. 2020:
https://feedback.dxo.com/t/dxo-pl6-color-space/29035/6
https://en.wikipedia.org/wiki/Rec._2020
but without any definitive word from DxO then your crucial observation that:

...It will have to be converted to a colour space that other software supports..."

becomes wild guesswork.

[DanK]

Interesting. I don't know myself, but I think I recall Ted, who used to post here, noting that ROMM and ProPhoto are the same.

I do wonder about the practical differences among various wide gamuts. Wide-gamut devices don't precisely conform to any of these standards. For example, my NEC PA271 misses a bit of Adobe RGB (they claim 98.1% coverage), but it also does display some hues that are outside Adobe RGB but within ProPhoto RGB. And in any event, that doesn't match the gamut of my printer exactly.

NEC's advice, if I recall, is to edit with the full range of the monitor, not restricting it to Adobe RGB. In theory, soft proofing can address differences between the monitor gamut and the gamut of the printer/paper combination.

I haven't tested carefully, but I think it's a relatively small proportion of my images for which the wider gamut matters, but when it does matter, it can matter quite a bit.

In any case, not a matter germane to noise reduction, I think.

[stuck]

...soft proofing can address differences between the monitor gamut and the gamut of the printer/paper combination...

Yes but the new soft proofing feature in DxO PL6 is another 'interesting' feature

...In any case, not a matter germane to noise reduction, I think.

Indeed, I realise my comment is a bit of a distraction from the main topic of this thread.

[DanK]

Yes but the new soft proofing feature in DxO PL6 is another 'interesting' feature

Is it any better than the soft proofing that's been available for a long time in both Photoshop and Lightroom? Although some people look down their noses at this, I have to admit that I print almost entirely from Lightroom, not Photoshop. (At least I have Jeff Schewe saying that this is a sensible decision...) I've found that the soft proofing is less than perfect--it has to be, comparing an emissive surface to a reflective one--but serviceable.

Indeed, I realise my comment is a bit of a distraction from the main topic of this thread.

Not really, as the starting point was comments on E. James' thread on high-ISO photography and noise reduction. He's posted a number of draft web pages for comment, and one of my consistent ones is that they would be improved by stripping out aspects of his processing that aren't germane to the topic of the page. I think this is one.

[Abitconfused]

Is it any better than the soft proofing that's been available for a long time in both Photoshop and Lightroom? Although some people look down their noses at this, I have to admit that I print almost entirely from Lightroom, not Photoshop. (At least I have Jeff Schewe saying that this is a sensible decision...) I've found that the soft proofing is less than perfect--it has to be, comparing an emissive surface to a reflective one--but serviceable.



Not really, as the starting point was comments on E. James' thread on high-ISO photography and noise reduction. He's posted a number of draft web pages for comment, and one of my consistent ones is that they would be improved by stripping out aspects of his processing that aren't germane to the topic of the page. I think this is one.

A logical point. However, when we deal with noise, my philosophy is to attack it from all directions. If a processing alteration provides the illusion of less noise that is as good as actual noise reduction. So I may use Microcontrast to bring out a subdued highlight and give the illusion of resolution sans noise even if a lower level of noise overall is not achieved. All is fair in art.

[Manfred M]

Please correct me if I'm wrong but I thought ISO 22028-2:2013 ROMM was the same as ProPhoto, see here:
https://www.color.org/chardata/rgb/rommrgb.xalter
and here:
https://en.wikipedia.org/wiki/ProPhoto_RGB_color_space
and here, where a staff member at Affinity Photo is posting:
https://forum.affinity.serif.com/ind...comment-159117

Indeed but DxO are saying almost nothing about their wide gamut colour space. Consequently there is a lot of confused and, as far as I can make out, much misguided discussions about colour management in general and DxO's wide gamut in particular on the DxO forum. If you have a spare lifetime you can try and wade through some of that chatter, e.g. here:
https://feedback.dxo.com/t/question-...or-space/28617
https://feedback.dxo.com/t/yet-more-...usion/27879/40
https://feedback.dxo.com/t/dxo-pl6-color-space/29035
DxO's aim seems to be to cover Pointer's Gamut and to do that that they might be using something that is close to Rec. 2020:
https://feedback.dxo.com/t/dxo-pl6-color-space/29035/6
https://en.wikipedia.org/wiki/Rec._2020
but without any definitive word from DxO then your crucial observation that:

becomes wild guesswork.

This information comes from a link from some discussions on the DPReview site noted in #7.

I claim no inside knowledge, although at one point several years ago, when I was using PhotoLab I asked DxO whether they would go to native ProPhoto or even L*a*b* colour spaces. Their answer was non-committal (i.e. we will pass this on to our development team for consideration for future updates).

I do still have PhotoLab 4 installed, but need serious noise reduction very rarely so could not justify upgrading to newer versions. Had they implemented ProPhoto RGB, I might have upgraded, but the proprietary colour space does not tempt me at this point,

[stuck]

...I do still have PhotoLab 4 installed... ...I might have upgraded, but the proprietary colour space does not tempt me at this point,

That's logical. I'm on PL5 but see no need to upgrade to PL6 for similar reasons. Plus the new soft proof feature is incomplete / barely a beta feature.

By the way / in case you didn't know, DxO have a strict policy of only giving a discount on the upgrade price for users of the previous two versions, i.e. you will be eligible for a discount on the upgrade price from PL4 to PL6 but this time next year when they release PL7 you'll be expected to pay the full price for PL7.