Dynamic Range vs. Exposure Range, and why the difference matters

[Rienzphotoz]

The real difference between the two bodies most frequently referred to in this thread (5DmkIII/D800) is the irrelevant rating from DxO vs field use. This is because DxO evaluates and scores the image data before it is processed by each manufacturers noise reduction, demosaicing algorithms, and color/tone balancing. It is image data that the user of the camera does not have access to - therefore is not transferable to the field. There are examples shown by other supposedly in-depth equipment review services that the D800 really only has maybe one stop over the 5DmkIII and that is at the shadow range, and no higher in highlights (where it really counts the majority of the time). So unless you literally intend to overexpose shadows in all your shots, you gain nothing except digital resolution for cropping - maybe - as long as you stay under ~400 ISO..

+1

[SiliconVoid]

(hjulenissen)
The grain I referred to is what Nikon is doing in their latest bodies. (Just an aggravation jab as I was hoping to upgrade my D700 with video.. heh) However even Nikon knows that with current tech you cannot add more mp and higher ISO without introducing more noise. Not saying it won't happen as tech advances, but it is the reality atm. Nikons approach is to pull out color noise, desaturate it, redigitize it, and redistribute it as 'grain' which they have spent the past few years trying to convince everyone is ok because it gives that 'film' look.. One of the first evolutions of digital was the absence of grain (at one ISO setting or another) to reintroduce it again to cover high gain noise is not a move forward I care to see. Does not matter whether you agree (no offense) it is what they are doing.
*Well they have been doing it for a while actually, just showing up more now with their higher mp bodies*

I did not assert the focus on shadow recovery as a claim, but you don't have to spend but 5-seconds online anywhere to see how many people focus on little else, especially if they are a Nikon user or hate Canon for some reason and making it the bases of comparison in superiority.

I agree with you regarding 'magic' in your images, my reference was to shadow exposure - not color.

Not sure what anthropomorphic qualities you read into my comment, but the sensor does indeed 'know' what the manufacturer wants it to be sensitive to. Canon has traditionally focused on protecting highlights, as did the industry in general for years, and Nikon on shadows.. For some reason today all the rage is about shadow recovery.. /shrug

Just used cropping as the most relevant example, that again 5-seconds online, you can see what most are using all those mp for..

As for DxO, I am sorry but you are completely wrong. DxO clearly states how their testing is done, at what stage the data is derived, and emphasizes that as the reason their data analysis is empirical - which it is, technically, because they are testing the hardware itself not how the manufacturer implements the hardware. The consumer however gets the manufacturers flavor of the data..

[J.R.]

Interesting discussion. One does wonder though what would happen if Canon came close to Nikon's DR in low ISO at the cost of losing a stop or two in High ISO.

Now who would want that camera?

Why at the cost of losing a stop or two High ISO? The two are achieved via different means. I don't think the mechanisms by which Canon could improve Low ISO performance would by necessity eliminate the gains they have made at High ISO. The best of both worlds could be had if Canon can figure out how to reduce their read noise.

Thanks ... I'm not too savvy with the tech of how this is possible - only just started learning so I'm just reading any stuff which is available on these matters as and when I have time.

Anyhow, my basic issue with DR complainers is that if somehow Canon offered a sensor much alike the Nikon - high DR at low ISO and poor High ISO performance - how many takers would they have? Would the people who are grumbling now be pleased then, losing the High ISO performance?

That's the essential difference between Canon and Nikon sensors and personally, I'd choose High ISO performance any day over Nikon's DR, which, I may hasten to add, I'd love to have in my Canon cameras

[Rienzphotoz]

That's the essential difference between Canon and Nikon sensors and personally, I'd choose High ISO performance any day over Nikon's DR, which, I may hasten to add, I'd love to have in my Canon cameras

+1

[3kramd5]

I have absolutely no intentions of sitting down at the computer and bringing the shadows up so they are any different than what I saw and what made me want to take the picture to begin with.

I don't think that's really part of the discussion.

More like showing detail which you did see but which is occluded either by shadows or by blown highlights in a digital image.

For example, I betcha that when the photographer looked up, he didn't see that blotch of white.


And I similarly expect that the photographer here could see cobblestones at the bottom left.


Granted, in the second case, the clipped shadows look good, but is what she saw? I doubt it.

What would be great is if cameras and media could match the DR of human vision, and clipping black or white was a creative choice, not a technological necessity.

[Aglet]

When printing or adapting for electronic display, it’s the shadow areas that get lifted, moreso than the highlites getting lowered, that compresses the DR of the image to fit the output medium.
Therefore, having clean shadow performance from the camera and sensor system can be quite important, especially when presenting on large prints or displays.  If you like to portray most of those darker shades as indistinguishable, which I see many do, and i often find distasteful, that’s your choice.
I prefer to have a system that gives me more options in post.
And I’ll define CLEAN once again:  the absence of fixed pattern noise is what’s more important than overall DR. Random noise, that looks like film-grain, does not create obvious distractions for the viewer.  Plaid-like patterns and strong vertical striping surely does.
If these points are important to you, choose your gear and-or limit your post-processing accordingly.
If not, don’t try to convince others, who may not share your viewpoint, that technical advantages like more DR and no FPN are irrelevant and it’s only the shooter’s skill that matters.  If that were true, why would anyone upgrade to better gear, especially the pro’s? We could all continue to compensate, and limit ourselves to fit the constraints of the gear we have, and we’d all be so happy with it.

[drphilgandini]

...
What would be great is if cameras and media could match the DR of human vision, and clipping black or white was a creative choice, not a technological necessity.

I don't know the neuroscience of human vision, but my experience tells me that the brain does a lot of processing to create a virtual dynamic range much greater than what the eye is capable of seeing. For example, in the photo of the alley, the cobblestones in the sunlight are much brighter than those in the shadows. Possibly by 7 or 8 EV, I don't know. When a human focuses on the stones in the light, it sees detail, but does not simultaneously see detail in the shadows, as the eye is not focused there. When the eye moves to focus on the shadow, the human adjusts to see detail there, but again, not simultaneously focusing on the highlighted stones. The brain processes these local contrast elements of the scene with detail. When looking at the scene as a whole, the brain now combines the global scene with the details of the highlights and shadows and creates a clear "picture" in the viewer's mind. But this is not a print, or anything equivalent to a print. Or even a scene in a movie or video. It's some image in a person's mind. I've concluded that it can't actually be recreated in the physical world via print or video. I also believe that the reason some people like HDR images is because they try to reproduce the "wider DR" we see in our minds, with both highlight detail and shadow detail.

[jrista]

Interesting discussion. One does wonder though what would happen if Canon came close to Nikon's DR in low ISO at the cost of losing a stop or two in High ISO.

Now who would want that camera?

Why at the cost of losing a stop or two High ISO? The two are achieved via different means. I don't think the mechanisms by which Canon could improve Low ISO performance would by necessity eliminate the gains they have made at High ISO. The best of both worlds could be had if Canon can figure out how to reduce their read noise.

Thanks ... I'm not too savvy with the tech of how this is possible - only just started learning so I'm just reading any stuff which is available on these matters as and when I have time.

Anyhow, my basic issue with DR complainers is that if somehow Canon offered a sensor much alike the Nikon - high DR at low ISO and poor High ISO performance - how many takers would they have? Would the people who are grumbling now be pleased then, losing the High ISO performance?

That's the essential difference between Canon and Nikon sensors and personally, I'd choose High ISO performance any day over Nikon's DR, which, I may hasten to add, I'd love to have in my Canon cameras

I think your overestimating what "poor high ISO performance" means in the context of any current Nikon camera. They are maybe 1/3rd to 1/2 of a stop worse than the 5D III. DR and noise at high ISO is ultimately limited by physics, and the design of a sensor can only have a relatively small impact (unless someone figures out a way to preserve a significantly greater percentage of the light entering the lens and reaching the sensor...in which case we might see a fairly substantial boost to high ISO noise for that particular brand.) The difference between Nikon and Canon is noticeable, but, it is still better than past-generation Nikon cameras. Nikon cameras don't really have "poor" high ISO performance. They are simply at a slight disadvantage relative to Canon. Again, either way, low ISO DR on Canon or high ISO on Nikon...both brands make excellent camera.

As for the rest of your post...personally, I too would take better high ISO performance...even if it is only a third to half a stop better. That is because I shoot at high ISO most of the time. I do, however, also take photos of landscapes, in which case ISO 100 is king and better DR is the most important thing. I am not one to push around shadows by four stops...I think that makes a photo look terrible. I would, however, like to be able to push my shadows around as much as I need without having to bother worrying about banding noise, or have to deal with running my RAWs through a tool like Topaz DeNoise 5 to get rid of any banding that exists in the shadows. I can recover a LOT of DR in my Canon images these days with a tool like Topaz DeNoise (at least a stop, if not two with some careful and meticulous tweaking), but...its extra work. It takes extra time, and isn't quite as good as a better sensor that doesn't have banding noise in the first place.

That said...clean, random noise in the shadows (i.e. once you have removed banding) can, as Ctein's article states, actually help improve "DR", or what he calls Exposure Range. Adding a little bit of noise back into shadows when you have to apply heavy debanding can actually help restore detail and extract even a little bit more dynamic range. (That was really the point I wanted to make by posting the articles, guess I should have known it would create a monster DR debate.)

[3kramd5]

...
What would be great is if cameras and media could match the DR of human vision, and clipping black or white was a creative choice, not a technological necessity.

I don't know the neuroscience of human vision, but my experience tells me that the brain does a lot of processing to create a virtual dynamic range much greater than what the eye is capable of seeing. For example, in the photo of the alley, the cobblestones in the sunlight are much brighter than those in the shadows. Possibly by 7 or 8 EV, I don't know. When a human focuses on the stones in the light, it sees detail, but does not simultaneously see detail in the shadows, as the eye is not focused there. When the eye moves to focus on the shadow, the human adjusts to see detail there, but again, not simultaneously focusing on the highlighted stones. The brain processes these local contrast elements of the scene with detail. When looking at the scene as a whole, the brain now combines the global scene with the details of the highlights and shadows and creates a clear "picture" in the viewer's mind. But this is not a print, or anything equivalent to a print. Or even a scene in a movie or video. It's some image in a person's mind. I've concluded that it can't actually be recreated in the physical world via print or video. I also believe that the reason some people like HDR images is because they try to reproduce the "wider DR" we see in our minds, with both highlight detail and shadow detail.

Is my brain doing some sort of sampling? Maybe. But whatever the mechanisms at play, I see what I see. We may never get there with imaging technology, but it's a worthy goal.

[jrista]

...
What would be great is if cameras and media could match the DR of human vision, and clipping black or white was a creative choice, not a technological necessity.

I don't know the neuroscience of human vision, but my experience tells me that the brain does a lot of processing to create a virtual dynamic range much greater than what the eye is capable of seeing. For example, in the photo of the alley, the cobblestones in the sunlight are much brighter than those in the shadows. Possibly by 7 or 8 EV, I don't know. When a human focuses on the stones in the light, it sees detail, but does not simultaneously see detail in the shadows, as the eye is not focused there. When the eye moves to focus on the shadow, the human adjusts to see detail there, but again, not simultaneously focusing on the highlighted stones. The brain processes these local contrast elements of the scene with detail. When looking at the scene as a whole, the brain now combines the global scene with the details of the highlights and shadows and creates a clear "picture" in the viewer's mind. But this is not a print, or anything equivalent to a print. Or even a scene in a movie or video. It's some image in a person's mind. I've concluded that it can't actually be recreated in the physical world via print or video. I also believe that the reason some people like HDR images is because they try to reproduce the "wider DR" we see in our minds, with both highlight detail and shadow detail.

Is my brain doing some sort of sampling? Maybe. But whatever the mechanisms at play, I see what I see. We may never get there with imaging technology, but it's a worthy goal.

Because of how the eye/brain vision center processes, we can see up to as much as 25 stops. The brain is effectively a superresolution processor for the central 2° foveal spot (the highest concentration of rods and cones in the retina, namely green and red cones). The eye has a refresh rate of about 500 frames per second, and the brain processes that information on sort of a rolling buffer....the oldest "frames" bear the least weight, and the newest "frames" bear the most. That gives us a very high resolution, saturated, crisp view of the world around us. The outer region of our retina factors in at a less important "weight" for primary daylight vision, however it is important for motion sensing and low-light vision.

From a technological standpoint, we are getting fairly close to 25 stops in digial sensors. Red recently demonstrated a 20-stop 4k cinema sensor. That is actually a pretty amazing feat. The eye manages 25 stops or so thanks to what would effectively be HDR processing by the brain. For a CIS to achieve 20 stops without any HDR or superresolution is very interesting, and technically "better" than the eye in that it can do that for every frame.

[J.R.]

I do, however, also take photos of landscapes, in which case ISO 100 is king and better DR is the most important thing. I am not one to push around shadows by four stops...I think that makes a photo look terrible. I would, however, like to be able to push my shadows around as much as I need without having to bother worrying about banding noise, or have to deal with running my RAWs through a tool like Topaz DeNoise 5 to get rid of any banding that exists in the shadows. I can recover a LOT of DR in my Canon images these days with a tool like Topaz DeNoise (at least a stop, if not two with some careful and meticulous tweaking), but...its extra work. It takes extra time, and isn't quite as good as a better sensor that doesn't have banding noise in the first place.

That said...clean, random noise in the shadows (i.e. once you have removed banding) can, as Ctein's article states, actually help improve "DR", or what he calls Exposure Range. Adding a little bit of noise back into shadows when you have to apply heavy debanding can actually help restore detail and extract even a little bit more dynamic range. (That was really the point I wanted to make by posting the articles, guess I should have known it would create a monster DR debate.)

Agree ... If I were shooting primarily landscapes, I would choose the D800 over the Canon offerings.

BTW, i dont agree that this is a "monster DR debate". on the contrary, this is the most reasonable DR debate I've seen on this forum for a long time.

Thanks for posting the link!