Maybe it would have been better to replace the word "histogram" with "zone system", but I tend to think of one in relation to the other. I now realize that I can't just map the zone system to the histogram in a linear fashion. You learn something new everyday.
Thanks panos_m for the link...dpreview's evaluation of highlight headroom is based on the JPEG tonal curve applied by the camera (even when shooting raw), which is used by the camera in combination with white-balance to depict post-gama clipping on the LCD. The LCD clipping indication is usually over-aggressive, the degree of which is determined by the specific picture style selected (tonal curve) as well as the white-balance.
May I suggest a test?
If the information is lost then it cannot be recovered. If by applying negative EV information is "recovered" then it was not lost in the first place. It seemed lost due to other reason (tone curve? narrow color space?). In the case of highlight recovery tools they try to guess values of the clipped channel (usually green) based on the remaining channels. In this case data are not recovered but invented.
--
I think that 2.5 needs to be carefully defined, and may not be applicable as universally as folks seem to claim. Rules of thumb sometimes have different thumbs. For instance, according to the DPR D90 review, the D90 has "3.9 stops of detail above middle gray" (aka, zone 5).
This is not as bad as it sounds. Even when you start off with RAW, eventually, you have to convert to the view print format. The defacto format seems to be JPG, so at some point you will be dealing with whether data clips in the image as you want to render it (your chosen contrast, saturation, etc.). As backwards as it may seem to some, though I shoot RAW, I like to make sure I'm using the final intended tone curve in-camera so that I know whether I'm starting to clip (or block). Yes, I can use the RAW to mitigate this later, then reapply the tone curve before final conversion, but developing a sense of how much headroom I actually have when I evaluate the histogram in the field is, for my work at least, not such a bad thing.
That's fair, but what you're defining above is a specific workflow, which may suit your needs but is not an appropriate methodology for making assertions about how well the D7000 retains highlights vs. other cameras. Again, highlight retention/headroom is a function of exposure and DR. The D7000 has 14 stops of DR, which can be expressed as having 0 stops of highlight headroom, 14 stops of highlight headroom, or any value in between. The color issue/hue twists can be addressed in the color profile of the raw processor.This is not as bad as it sounds. Even when you start off with RAW, eventually, you have to convert to the view print format. The defacto format seems to be JPG, so at some point you will be dealing with whether data clips in the image as you want to render it (your chosen contrast, saturation, etc.). As backwards as it may seem to some, though I shoot RAW, I like to make sure I'm using the final intended tone curve in-camera so that I know whether I'm starting to clip (or block). Yes, I can use the RAW to mitigate this later, then reapply the tone curve before final conversion, but developing a sense of how much headroom I actually have when I evaluate the histogram in the field is, for my work at least, not such a bad thing.
This is where someone chimes in about UniWB as the only way to truly do this, but showing my client her green skin tones during a headshot session isn't terribly practical for me.
--That's fair, but what you're defining above is a specific workflow, which may suit your needs but is not an appropriate methodology for making assertions about how well the D7000 retains highlights vs. other cameras. Again, highlight retention/headroom is a function of exposure and DR. The D7000 has 14 stops of DR, which can be expressed as having 0 stops of highlight headroom, 14 stops of highlight headroom, or any value in between. The color issue/hue twists can be addressed in the color profile of the raw processor.This is not as bad as it sounds. Even when you start off with RAW, eventually, you have to convert to the view print format. The defacto format seems to be JPG, so at some point you will be dealing with whether data clips in the image as you want to render it (your chosen contrast, saturation, etc.). As backwards as it may seem to some, though I shoot RAW, I like to make sure I'm using the final intended tone curve in-camera so that I know whether I'm starting to clip (or block). Yes, I can use the RAW to mitigate this later, then reapply the tone curve before final conversion, but developing a sense of how much headroom I actually have when I evaluate the histogram in the field is, for my work at least, not such a bad thing.
This is where someone chimes in about UniWB as the only way to truly do this, but showing my client her green skin tones during a headshot session isn't terribly practical for me.
So your lab prints TIFFs? Really? With only 6-7 stops of DR for the best printing papers? Hmm.The defacto format seems to be JPG, Certainly not for me. I want the full tiff - especially since I print my own photographs. It really depends on your needs. I want the full enchilada!
Somewhere else I stated that "Highlight range from zone V (18% grey) to clipping is the same for every digital camera (except Fuji S5?). 2.5 stops". Its better like this: Highlight range from zone V (18% grey) to clipping is the same for every digital capture . 2.5 stops". That is from 18% grey to white is 2.5 stops. The thing is that not every camera meters 18% grey. Most of them meter lower to preserve highlights (even more than 3 stops. My D3 is 3.3 and the GF1 is 2.90. The GF1 is dangerous because it is changing its metering based on the in camera JPG settings). That is why I made the comment about metering.
This is OK if you use always the same processing tool which compress the RAW headroom in a familiar way. Personally I find it difficult to remember every combination of every different Nikon camera I use with every different Picture Control and base my exposure decisions on this. Its way more easy and controllable to just know how the particular camera meters. Then in post processing I have the option to compress the highlight range in any way.
Its not theoretical. Its real and most importantly stays the same every time no matter what raw converter you are going to use.
Yes, and a print isn't 14 or 16 bits, with 7 stops of DR tops. So what exactly is gained by the TIF handling again? Bottom line: whether you convert to JPG, then print, or go "straight" from TIF, the downconvert is happening -- and steeply.
The conceptual problem I have right off the bat here is that for many of us, we the photographer and not the camera are the ones doing the metering. Are you referring here to the auto-metering one would get in Program or Aperture priority mode? Are you then saying that if we choose to spot meter "our way" we are either reducing or enhancing the highlight range? That would seem a rather planted in mid air conclusion, no?
One wonders why any sane person would choose to do it any other way. Most of us do indeed pick "a tool" and a "familiar way" to process our images. It's called a work-flow in which we have indeed pre-determined certain converters, curves and settings that work for us for certain types of shots. But I guess if the premise is that we can arbitrarily change the way we meter from one shot to the next to get this or that other highlight range, why wouldn't it be logical also that we would randomly choose how we process an image from one shot to the next?
Where did you get this info from, based on the 0-255 levels?
If you mean the RAW headroom for the mentioned cameras I measure it from 2 shots, one as the meter suggested and one heavily overexposed.
The meter in Nikon cameras (at least for those I tested) meters the same despite which picture control you choose. So the RAW headroom is always the same for a particular camera. The difference between Neutral and Standard Picture Control (except the color difference) is the way that HL range is compressed through the different tone curves. The clipping point is the same.
But how do you check that the clipping is actually occuring for the RAW file?
If you mean when I shoot, if a particular area in my frame is supposed to be placed just before clipping, I spot meter it and I add exposure a little less than my camera's RAW headroom.
DR figures in DPR reviews show considerable variation among cameras at the highlights end. But this is one thing for which the DPR reviews should actually be useful
3- Shadows with the D7000 sensor (base ISO) contain some more information than shadows with most other sensors.
--
