About Nikon D7000

[girod]

Let me start with 2 questions:

1) What does the 0 in D7K's exposimeter meters? 18% or 12.7% or 7.25% gray?

2) Excluding the +EV that the D7K and the raw converter apply, where should we spotmeter Zone VII (brightest/whitest with detail)? +2/+3/+3.7?

[keithsnell]

Let me start with 2 questions:

1) What does the 0 in D7K's exposimeter meters? 18% or 12.7% or 7.25% gray?

2) Excluding the +EV that the D7K and the raw converter apply, where should we spotmeter Zone VII (brightest/whitest with detail)? +2/+3/+3.7?

Number 1 can't be answered (with any practical value) until you define what raw processor you will use. Once you define the raw processor (and your preferred tone curve), it's easy to figure out. Take a picture of an 18% gray card, process it normally through your raw processor with your preferred tone curve, print on a calibrated printer or display on a calibrated screen and compare the tonality to the "real" 18% gray card.  Is it darker, lighter, or "just right?"

The "practical" answer to question #2 is "+2."  People tend to pick the wrong answer when they forget what the definition of Zone VII is.  Keep in mind that Zone VII is defined as the "lightest area that retains visible detail."  It is understood that this "detail" should not be clipped in any of the color channels.  If you are photographing a white object with detailed texture (a white towel or fabric is a good example) and your spot meter reads +2, you should understand that it is "averaging" the brightness of the area that falls under the 3.5mm area surrounding the selected focus point.  Some of that area will be slightly darker than +2EV, and some of that area will be slightly lighter than +2EV.  +/- 0.5 EV is a typical range, probably distributed in a bell curve.  So, even though your spot meter is indicating +2EV, there is also detail under that spot that ranges up to about +2.5EV and that detail needs to be protected if you intend to render that area as "Zone VII."  Capture NX2, renders an area exposed at 2.5EV at 255, or fully saturated in an sRGB or Adobe RGB color space.  So, given that you are using Capture NX2, Zone VII should be exposed at +2EV.  So what about zones VIII and IX?  Remember that these zones are defined perceptually.  They aren't defined as a numerical range, but rather how the viewer perceives the tonality (and detail) of that zone.  Zone VIII is defined as "white areas with slightly (barely perceptible) visible textures" and Zone IX is defined as "glaring white surfaces - Highlights."  So Zone IX falls just below 255 (+2.5EV) and Zone VIII is in the narrow range just below that where you can start to discern detail.  (If you are exposing for Zone VIII, +2.3EV is probably a good target.)

I haven't tested the D7000 yet to see what the "highlight headroom" is, i.e., that undefined region above the point that is rendered at 255, and the physical saturation point of the sensor.  (All manufacturers typically leave a little bit of "margin" here.  For the D3 it was 0.5EV of "headroom" in the green channel; however I haven't tested yet to see what it is for the D7000.  Anecdotally, based on the 800 or so frames we've processed from the D7000 I think the "engineering margin" (I think a more accurate term than "highlight headroom") is about the same, or about 0.5EV).

I probably won't have time to do detailed testing on the D7000 exposures until about three months from now, when things will slow down a bit for me.

I hope this helps,

Keith

[girod]

Thank you very much Keith. I uploaded an image shot at +3.7EV which was taken when I first got my D7K. Please see what you think.

[keithsnell]

Thank you very much Keith. I uploaded an image shot at +3.7EV which was taken when I first got my D7K. Please see what you think.

Hi Jaime,
The recovered image looks fine.  But why would you expose that scene at +3.7EV?  (There is certainly danger of clipping color channels, and no deep shadows that need to be pulled up.)  It would have been interesting to expose the scene with less exposure (maybe +2EV on the lighter parts of the image) and then compare the renditions against each other. 

Keith

[girod]

I was just curious at that time if I could duplicate DPR's review that the highlights headroom of D7K is +3.8EV (JPEG at that and in all picture controls). But as you've said why risk clippings with the D7K when lifting shadows at ISO 100-200 is so good.

[girod]

Keith,

What do you think of this suggestion from DavidNC: http://forums.dpreview.com/forums/read.asp?forum=1034&message=38482981

[keithsnell]

It is important to note that the DPReview measurement of highlight headroom is NOT measuring from a meter reading of 0EV on the camera's exposure scale.  To quote:  "Our Dynamic Range measurement system involves shooting a calibrated Stouffer Step Wedge (13 stops total range) which is backlit using a daylight balanced lamp (98 CRI). A single shot of this produces a gray scale wedge from the camera's clipped white point down to black (example below). Each step of the scale is equivalent to 1/3 EV (a third of a stop), we select one step as 'middle gray' (defined as 50% luminance) and measure outwards to define the dynamic range. Hence there are 'two sides' to our results, the amount of shadow range (below middle gray) and the amount of highlight range (above middle gray)."  

An exposure equivalent to 0EV on the camera's metering scale will fall above the point defined as 50% luminance in the DPR test.  You have to remember that Nikon bumps up the midtones (and compresses the highlights) in all of their tone curves for the current generation of cameras (since the D3 onward).  The range measured from 50% luminance will always be greater than the range measured from 0EV with Nikon's current tone curves.  Since your meter doesn't have a "50%" luminance mark, you by necessity need to measure your "real world" dynamic range based on the range from the 0EV reading on the exposure scale.  DPR's measurements help us compare sensors and tone curve implementations, but are not meant to be an exposure guide.

Hopefully this makes sense?

Keith

[keithsnell]

Keith,

What do you think of this suggestion from DavidNC: http://forums.dpreview.com/forums/read.asp?forum=1034&message=38482981

+3.0 is better than +3.8EV.   But there is a difference between the way the file is rendered (with +2.5EV being rendered at 255 or 100%) and the rendered DR plus the "engineering margin" to get to the clipping point of +3.0EV.  That "margin" is there for a reason.  Colors are not guaranteed to be linear beyond that point, banding can occur in certain cameras when approaching +3.0EV, etc.  The 0.5EV "margin" between +2.5EV and +3.0EV should be reserved as "engineering margin" and photographers should understand that eating into that margin can cause more issues than just excepting 0.5EV less dynamic range in the shadows.  There is NO output device that can effectively render 14 stops of dynamic range (without boosting the shadows).  Eating into the "engineering margin" at the highlight end is counter productive with today's sensors/tone curves.

And don't forget that "saturation point" is much different than metering for a zone where all detail is expected to be retained.  "Saturation" is Zone X, not zone VII.  I would still meter at +2.0EV for Zone VII (that's the "middle" of zone VII, not the top) and +2.3 EV for Zone VIII.

[girod]

Aaaaaaaah.......the things that we learn from you, Keith.....all we have to do is ask. "Maraming Salamat Po"  - Tagalog for thank you (salamat) very much (maraming), po is with great respect. DPR should hire you to put a real and practical photographic perspective on those technical reviews usable towards a better implementation of optimal image capture and rendition.

I uploaded another image (taken at the same time as the previous image #1480) to show some color rendition. What do you think?

[keithsnell]

Thank you Jaime,

I posted a comment on that image, but to continue our discussion here, I think the only colors that are accurate in that image are the ones that are in the shadow of the building.  The other areas of the image show either a color shift (skies that are too cyan) or loss of detail that would have been there at a lower exposure.

It is important to note that sometimes colors (especially skin tones) won't "look right" even if they aren't clipped.  Rendering skin tones too "hot" can result in unnatural looking skin tones that just don't seem real to the viewer.  Even if all the color information is still intact in the raw file, it might be "compressed" so far to the right that it just doesn't have the natural tonality that we expect.  The tone curves that Nikon uses in their current cameras attempt to "protect" some of the highlight detail by compressing the highlights as they approach clipping.  Too much compression robs these areas of the image of tonal gradations, texture and detail (and in some cases can skew the color balance).

Hitting the mark with respect to a "good" exposure is as much (or more) about "perception" as it is about the numbers.

[girod]

Aaaah.....good that you emphasized it, Keith - photographic "perception", the thing I have to consciously develop with persistence because I don't have it. I was able to correct the cyan sky with the selective "color control point" by adjusting the blue channel slider. I went up the rooftop and looked at the scenery/subject - you're right, the only color that I rendered right was the brown-reddish marble wall (+ the windows) of the building in the center; the white wall is actually a very light yellow/beige!