Reading mode: Light Dark

D7000's dynamic range in the highlights

Go to last post
Just think you are making it out to be much more complex than it really is.....

What a meter says is 18% is highly prone to user error - the most significant variability in all picture taking.... all the business about the fact that the histogram may or may not be accurate depending on what "picture style" you are running, whether or not you ADL is active, and on and on account for an error of about 1/2 of a stop, after the fact, so I trust it more than any other single thing a modern camera does.... its just NOT that big a deal. Just avoid heavy clipping and you are good to go, you remap (or don't) in post - where you place your zones then simply becomes a matter of taste in post.

That is why the comment about a calibrated monitor.....

The point of the technology is that it is highly forgiving, especially in the case of the D7000 at base ISO......the thing to be absolutely aware of is that the basic tools (lets just consider the standard "levels" and "curves" tools) are tone remapping "devices". There is no such thing as a set 18% benchmark, either locally or globally, only your opinion of that mark.

"That's my story and I'm sticking to it....."
 
eNo said:
Then what would we check? I agree the JPEG isn't an accurate representation of whether the RAW actually clips or not, but out in the field, what exactly would you suggest we check as a barometer of whether the RAW is clipping?
Click to expand...
If you've tested the camera and have determined the difference between raw clipping and what meters as middle gray, you don't have to check anything. Meter the highlights (or at least the brightest area in which you wish to retain detail) and open up an amount slightly less than the difference.
eNo said:
If we understand that JPEG isn't accurate but from experience and trial-error learn the relative difference between what the JPEG is showing us through the histogram (for a given tone curve, etc.) and what the RAW actually gives you, you at least have an indicator that you can use to gauge fairly accurately whether the RAW data is clipping. It's not 100% bullet-proof, but it's better than shooting now and figuring out whether we have clipping later.
Click to expand...
If you know the difference between what the meter sees as middle gray and the highlight clipping point, why use the JPEG as an intermediary at all? It just seems to me like a lot more memorization (if using multiple picture controls) for less accuracy.
eNo said:
Take the sample images in the OP. If through experimentation I've learned that I can actually expose the whites on zone 8 rather than 7 and still recover highlight detail, I will look at histograms for situations like these with the understanding that moderate JPG clipping still leaves me with headroom I can use in PP. Do this enough, and you will be able to tell in the histogram how much clipping is okay and how much is over the top.

Does that make sense?
Click to expand...
Well, not really. If you've already determined you can expose highlights to a certain point with the ability to recover them later, why look at the histogram? It's redundant.
 
binary_eye said:
binary_eye said:
If we understand that JPEG isn't accurate but from experience and trial-error learn the relative difference between what the JPEG is showing us through the histogram (for a given tone curve, etc.) and what the RAW actually gives you, you at least have an indicator that you can use to gauge fairly accurately whether the RAW data is clipping. It's not 100% bullet-proof, but it's better than shooting now and figuring out whether we have clipping later.
Click to expand...
If you know the difference between what the meter sees as middle gray and the highlight clipping point, why use the JPEG as an intermediary at all?
Click to expand...
Again, I ask, what would you check instead? It just so happens that when we bought our cameras, we paid for this histogram feature that we can use as an indicator. I guess we can choose to ignore it to avoid all the memorization, but it sounds to me like we're left with guessing. Oh, yes, in the old days all anyone needed was a meter, but I just happen to think the new days where we can double-check our work are much nicer.

~~~~~~~~~~~~~~
Seeking the heart and spirit in each image



Gallery and blog: http://imagesbyeduardo.com
Flickr stream: http://www.flickr.com/photos/22061657@N03
 
Fred Mueller said:
why soooo much discussion?
  • shoot raw, just don't clip, or clip as little as possible - then adjust tonality, curves, color response etc later in post. If you are pressed for time, do the same wiih jpg's - you will just have less leeway.....
who cares what is exactly 18% and how many stops above till clip and on and on.... use your eyes....no need for obscure discussions about "zones" and linear sensor response.....none of that actually matters - all of it is adjustable ex-post-facto....a calibrated monitor is far more important.

This is hardly a new issue, and while the D7000 has a little extra dynamic range at base iso, the idea of not clipping highlights and lifting shadow detail "to taste" in post is simply SOP for digital photography - probably the first thing to learn about post process....
Click to expand...
True, but with D80 it usually meant very poor colors and a lot of noise in lifted areas, one went for a compromise and let some HLs blow. It's nice to know the new models can perform much better in that regard, actually amazingly better given the last post by eNo on shadows lifting floating around:

http://imagesbyeduardo.com/main/2011/05/20/d7000s-dynamic-range-for-shadows/

Cheers!!

--
Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)
 
And you claim all Nikons you've tested that way always show 2.5 stops over center of exp scale at clip point?

Sounds a bit strange, but could be the calibration Nikon imposes on their meters. I've checked both D80 and D7k in that regard, and even though both start showing some clipping at 2.3-2.5 stops (blinkies, but that's jpegs), when I open the files in CNX2 and try to bring that back HLs from clipping I see better results from D7k, indicating that in the RAW file, actually, the D7k's image is less clipped.

Don't worry, I'm not into extra tech-testing myself, I just need to know how camera performs in practical terms, and my simple checks show some differences, maybe for some other reasons than clipping actually, which is not relevant, again, in practice.

Thanks.
panos_m said:
panos_m said:
rhlpetrus wrote:
No, I mean how did you measure your camera's RAW headroom
Click to expand...
You need one shot of a evenly lit surface with the meter centered and one completely saturated. I believe 5 stops over every camera's spot meter is enough but to be sure add more. From the resulting raw files, 4channels program (which included in libraw binaries distributions) and Photoshop or Gimp you can measure the RAW headroom. I am leaving for the weekend. If you are interested I can provide more detailed instructions when I come back.

--
Panagiotis
Click to expand...
--
Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)
 
Renato said:
"True, but with D80 it usually meant very poor colors and a lot of noise in lifted areas, one went for a compromise and let some HLs blow. It's nice to know the new models can perform much better in that regard, actually amazingly better given the last post by eNo on shadows lifting floating around"
Click to expand...
Of course! You had a little bit less DR to work with but do you really think this is going to change a whole lot with the D7000 - I doubt it. I have owned a D40, then a D7000, and now a D700. In most of that time I also shot my son's D300, often. For me there was practically NO apparent difference in the "feel" for what you had to do to get a decent exposure, or preserve highlights.

All of these cameras are perfectly capable of blowing the heck out of highlights. In fact, the most reasonable comment I can make is how exactly the same I shoot each one of them with regard to retaining highlights, and for all of them by far the best method is to shoot and then trust the histogram, especially the RGB histogram.

The whole idea that you can "spot meter zone 8 or zone 7 and then open up your exposure setting by some exact fractional number of stops, and therefore get a perfect exposure is pretty specious, and if you have time to do all of that, hell, just bracket!

sorry to throw a skunk into the party.....

Fred
 
Hi Eduardo:

Thought I would chime in, hopefully with something useful. I believe that user panos_m pointed out earlier that for a midtone based on 18% refllectance, there are only 2.5 stops of highlights regardless of camera. Another user did the math based on doublings of light showing this to be true. I have to agree with this based on the following.

Best as I can tell from several sources, reflected light is linear, so that 50% reflection represents half of 100% (seems intuitive as it should be on a percent scale). So, 36% is double the reflected light of 18%, 72% is double 36%, and 100% is about 2.5 doublings or stops.

From wiki,the reflected light equation is Nsquared/t = LS/k

where
N = the fstop
t = shutter speed
L = luminance of the scene
S = ISO
k = a meter calibration constant

If you rearrange the equation to solve for the luminance L, and use an fstop of 2, shutter speed of 1, ISO of 100, and a constant for Nikon of 12.5, you get a luminance value of 0.04 (forget the units). If you change the fstop to 1.4, assuming this is what the meter suggested, you get a luminance of 0.02, half as bright as we would expect if we had to open up 1 stop.

So, the above shows that the reflected light equation is linear.

I believe that Nikon bases their meter calibration on 12% reflectance, not 18%. That means that there are about 3 doublings or stops to get to 100%: 24%, 48%, 96%. If all Nikons used this meter calibration, then all would have about 3 stops of highllights above a 12% reflected light midtone.

User panos_m also noted that the extra stops available are below mid gray. This is true and is a result of decreasing the read noise of the sensor. The overall amount of photons collected can be higher for different sensors and thus any given tonal value will have more photons converted to electrons associated with that level, so a mid gray on the D7000 will have more photons or electrons than some earlier cameras with older technology (D300 down). I think this would translate into higher signal to noise for a given read noise level. The D7000 has very low read noise, so there are an extra 2 to 3 stops in the shadows available relative to most other Nikon sensors except for the D3X.

But getting back to highlight headroom and overall DR: there should always only be about 3 stops of headroom for 12% meter calibration. For the D7000, there are then about 10-11 stops from 12% reflected light to the noise floor. Yes, we cannot differentiate those black shades very well until they are lifted in processing, but they do represent accurate image signal well above noise, and so, if lifted to a brighter tone in processing, they can be more visible and useful.

In terms of %Reflectance values, you get the doubling series, 100%, 50, 25, 12.5, 6, 3, 1.5, 0.75, 0.4, 0.2, 0.1, 0.05, 0.025, 0.01 (roughly- some values were rounded). You will note that DxO is measuring and plotting down to 0.01% now for signal to noise.

Again, based on 12% reflectance, there are about 3 stops of highlights and about 10-11 stops for shadows (of useful signal above the noise). As we have discussed, we can only see about 6 stops in a print and maybe 8-9? on a monitor, so the useful signals have to be processed (pulled down or lifted) into the viewable range.

The extra stops on the D7000 are below 12% mid gray. You could assign your middle tone lower but to make it show as the same tonal value as 12%, it would have to be lifted. But assigning it lower allows you to divide the DR more evenly so that you could say you have more stops of highlight headroom. As long as you could make the lifted tone look like the real 12% midtone, including the color accuracy, then this is fine.

I think one of the biggest issues with recovering useful signals is that the color response is not linear and therefore, pulling down highlights or lifting shadows can lead to color tone shifts that, in my experience on a D200 relic ;^), are hard to correct when pushed or pulled more than a couple stops. Maybe the D7000 does better in this regard and for BW, this is not really an issue.

So, I have to agree with the users who have stated that spot metering the tone that you want to record as a near white highlight and then opening up 3 stops on a Nikon with 12% meter calibration would maximize the available camera DR, regardless of whether it is a D7K or something else. Then, adjust in post to taste assuming that color is correctable. The D7K has more dark headroom (bottom room?).

I could have something wrong here, so if someone spots an error, please correct my understanding.

David
eNo said:
As some of you may know, one strategy to protect highlights in digital photography is to place them in zone 7. The following article tries to answer the question of whether the D7000's additional DR allows further exposure of the whites into zone 8 or 9 with full recovery of highlight detail. For reference, D7000 capability is compared with that of the D300's:

http://imagesbyeduardo.com/main/?p=2973
--
~~~~~~~~~~~~~~
Seeking the heart and spirit in each image



Gallery and blog: http://imagesbyeduardo.com
Flickr stream: http://www.flickr.com/photos/22061657@N03
Click to expand...
 
eNo said:
Again, I ask, what would you check instead? It just so happens that when we bought our cameras, we paid for this histogram feature that we can use as an indicator. I guess we can choose to ignore it to avoid all the memorization, but it sounds to me like we're left with guessing. Oh, yes, in the old days all anyone needed was a meter, but I just happen to think the new days where we can double-check our work are much nicer.
Click to expand...
I guess there is sense in the spot meter approach sometimes. For instance, when you don't want to look at the histogram - like at events in dark places where bright LCDs are dusturbing, or if you want to look like a confident pro who doesn't have to check the images and play with the settings all the time. It is indeed insensitive to the PC used. But the meter may be fooled by red or blue highlights. One may not recognize the brightest spot, or miss it with the meter. And, there are lenses with aperture issues (got one such Tamron).

Anyway one is doing this, maximizing highlights without blowing them is helluva popular these days. It makes me wonder why Nikon (and others) can't pay a bit more attention to it and provide some automation... I know, they're to busy making movies work ;) I'm aware there are certain problems, like that the RGB sensor is not able to detect very small spots, and that sometimes one would prefer to blow specular highlights up to a certain size - but I guess they are manageable.

There is a related problem that images with histograms aligned to the rhs edge may not look reasonably exposed. In harsh light they look too dark, while in too bright case of low DR scenes. Especially tonemapping high-DR scenes to the viewable range seems tough for in-camera processing these days. But at least for the time being I guess this could be manageable by splitting exposure into separate recording exposure and display exposure. So that the camera could record the image by maximizing the highlights, but display it brighter or darker. In the probable case where the camera would not know the desired display exposure, one should be able to modify it on the spot (if shooting raw) - then the camera would regenerate the preview JPG from the existing raw data with the modified display EC. And, of course, we need a histogram mode showing raw data (the contents of highlights and shadows headrooms) - I think I've seen sth like this on a recent Oly compact.
 
Fred Mueller said:
The whole idea that you can "spot meter zone 8 or zone 7 and then open up your exposure setting by some exact fractional number of stops, and therefore get a perfect exposure is pretty specious, and if you have time to do all of that, hell, just bracket!
Click to expand...
Have you tried it? I mean the zone 8/7 thing? It's not nearly as painful as you seem to think. No more, no less painful than spotting for midtones. Again, really surprised at this sort of response, Fred.

~~~~~~~~~~~~~~
Seeking the heart and spirit in each image



Gallery and blog: http://imagesbyeduardo.com
Flickr stream: http://www.flickr.com/photos/22061657@N03
 
This is all fine theory, and in the last couple of days, I've been really thinking about what all this means practically , and in the end, I must say, "not an awful lot." DPR says there's more like 3.8/9 of range in the highlights, and others point out this depends on where one places the midtones (zone V), which is in the end an arbitrary choice (I often don't place it precisely on what would be 18 or even 12% gray). So all this is sort of up to my usage and preference of how you want to handle a given scene's range. I do know this, though: I can place whites in Zone 7 or push them over with Zone 8, after which a -1EV adjustment in post gets me pretty much the same level of detail I would have gotten if I had bound them to Zone 7. I'll let others decide for themselves whether this proves that there's a little more highlight headroom than the 2.5 claimed throughout this thread.

In the end for me, it's the practical outcome that matters. I don't take theoretical pictures.
--
~~~~~~~~~~~~~~
Seeking the heart and spirit in each image



Gallery and blog: http://imagesbyeduardo.com
Flickr stream: http://www.flickr.com/photos/22061657@N03
 
I found that D7k has about 2/3 more room from midtone to clipping than D80, to me that's useful, I use spot and manual for all my shooting, except for flash.

Now, the D80 can do a pretty good job as well, my last shots with it last month in France ;).




HyperCypher said:
Of course! You had a little bit less DR to work with but do you really think this is going to change a whole lot with the D7000 - I doubt it. I have owned a D40, then a D7000, and now a D700. In most of that time I also shot my son's D300, often. For me there was practically NO apparent difference in the "feel" for what you had to do to get a decent exposure, or preserve highlights.

All of these cameras are perfectly capable of blowing the heck out of highlights. In fact, the most reasonable comment I can make is how exactly the same I shoot each one of them with regard to retaining highlights, and for all of them by far the best method is to shoot and then trust the histogram, especially the RGB histogram.

The whole idea that you can "spot meter zone 8 or zone 7 and then open up your exposure setting by some exact fractional number of stops, and therefore get a perfect exposure is pretty specious, and if you have time to do all of that, hell, just bracket!

sorry to throw a skunk into the party.....

Fred
Click to expand...
--
Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)
 
As D7000 has lower shadow noise, u can place the highlights even lower zones than zone 7 and push the shadows later (without much visible noise). This will result in better highlights. This is how highlights r preserved in digital photography, as shown here (the outdoor scene): http://www.dpreview.com/reviews/nikond7000/page17.asp
eNo said:
As some of you may know, one strategy to protect highlights in digital photography is to place them in zone 7 . The following article tries to answer the question of whether the D7000's additional DR allows further exposure of the whites into zone 8 or 9 with full recovery of highlight detail. For reference, D7000 capability is compared with that of the D300's:

http://imagesbyeduardo.com/main/?p=2973
--
~~~~~~~~~~~~~~
Seeking the heart and spirit in each image



Gallery and blog: http://imagesbyeduardo.com
Flickr stream: http://www.flickr.com/photos/22061657@N03
Click to expand...
 
And here lies the answer how better shadow noise performance results in better highlights. Once the light level reaches 100%, it is blown, no matter what digital camera it is. But digital cameras differ in shadow noise performance, for example, assume camera A has acceptable noise down to 0.4% reflectance, and for camera B it is down to 0.2%. It means that to protect highlights, u can underexpose camera B by one more stop than camera A, and recover the shadows later. But if u underexpose camera A by same amount as camera B, camera A will give unacceptable result once the shadows r recovered. So u can not underexpose as much with camera A. And as u can see, the upper limit is 100%, but the lower limit can be further lowered by a camera with better shadow noise performance. It is not like that 0.01% is the lowest limit, it can b 0.005%,0.0025%, 0.00125%, 0.000625% or any lower value. But the highest possible value can never b higher than 100%.
DavidNC said:
DavidNC wrote:

In terms of %Reflectance values, you get the doubling series, 100%, 50, 25, 12.5, 6, 3, 1.5, 0.75, 0.4, 0.2, 0.1, 0.05, 0.025, 0.01 (roughly- some values were rounded). You will note that DxO is measuring and plotting down to 0.01% now for signal to noise.
Click to expand...
 
I have a question:

Is the d7000 documented over-exposure in high contrast scenes lens dependant?

I found out testing the camera using the same aperture, same shutter speed , focal range and iso that during bright high contrast scenes that the camera blew the highlights ( raw and jpeg) with the kit lens, however, it exposured properly with the DX 35mm f 1.8...........

I am wondering if anyone else has experienced the same issue.
 
From the same page: "In real life this means that, due to the exceptionally low shadow noise in the D7000 output, the shadows in a scene can be lifted in RAW conversion to create a high dynamic range image without sacrificing highlight detail or getting excessive noise in the shadows ." In digital photography, better shadow noise = better DR (= better highlights).
KLO82 said:
As D7000 has lower shadow noise, u can place the highlights even lower zones than zone 7 and push the shadows later (without much visible noise). This will result in better highlights. This is how highlights r preserved in digital photography, as shown here (the outdoor scene): http://www.dpreview.com/reviews/nikond7000/page17.asp
Click to expand...
 
Mario Mejia said:
Is the d7000 documented over-exposure in high contrast scenes lens dependant?

I found out testing the camera using the same aperture, same shutter speed , focal range and iso that during bright high contrast scenes that the camera blew the highlights ( raw and jpeg) with the kit lens, however, it exposured properly with the DX 35mm f 1.8...........
Click to expand...
I don't have the D7000, but with the D90 this certainly may be the case. Lenses may have less than perfectly calibrated apertures, may have different transmissivity of glass, and may mess with metering.
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top