Digital Exposure Latitude

Rick58

Been spending a lot of time on here!
Joined
Jun 23, 2012
Messages
4,227
Reaction score
1,473
Location
Reading, Pa
Can others edit my Photos
Photos OK to edit
Can anyone tell me what the exposure latitude is for digital processors?
I'm assuming it would vary between between processors so I'm particularly interested in the FX CMOS processor found in the Df
 
DxOMark.com states that in RAW mode the sensor's dynamic range at lowest ISO is 13.1 EV. BY 1600 ISO, the dynamic range drops to 11.92 EV, and begins to decline fairly rapidly thereafter, which is pretty standard.
 
Digital exposure latitude is about more than just the dynamic range of the image sensor.

Half of all the luminosity data in a digital image is in the brightest stop (EV) of exposure. That is the basis for the ETTR - Expose-To-The-Right - concept of controlling digital photo exposure.

To keep it simple assume an image sensor has a dynamic range of just 6 EV (stops), and that the scene photographed has 4096 levels (12-bits) of luminosity.
Note: Most DSLR's output 12-bit or 14-bit digital luminosity numbers from the Analog to Digital (A/D) converter that is between the analog image sensor and the camera's digital image processing computer chip.

With 1/2 of the luminosity data being in the brightest EV of exposure, 2048 levels of luminosity are in that stop of exposure.
The next brightest EV of exposure then has 1/2 of the 2048 or 1024 levels of luminosity, and so on so it looks looe this:
3rd EV - 1/2 of 1024 - or 512 levels
4th - 256 levels
5th - 128 levels
6th - 64 levels

So at some point in the darker regions of the total dynamic range the image sensor is capable of, we have little control of exposure.
We have no control at all once we pass that point that is 1/3 of a stop, because that is the smallest exposure adjustment DSLR cameras have available.

So we expose for the highlights, and let the shadows take care of themselves.

Film is exactly opposite - Expose for the shadows and let the highlights take care of themselves.

http://www.adobe.com/digitalimag/pdfs/linear_gamma.pdf
http://wwwimages.adobe.com/www.adob...ly/prophotographer/pdfs/pscs3_renderprint.pdf
 
Wow, you guys are way too deep for me.
Is'nt there a standard that if I expose for neutral gray, X number of stops of over exposure will produce pure white (255) with no detail while X number of stops under will produce pitch black (0), again lacking all detail?
 
Dynamic range is simply the number of stops between black and white. Mike's information appears accurate, but note that 12- and 14-bit images apply to RAW, not JPEG, which is limited to 8 bits, just 256 steps of brightness from black to white. To take full advantage of a DSLR's dynamic range for a given negative, you need to shoot RAW, and that's just a fact of numbers. A rule of thumb is that you can safely expose 2.5 stops open from a highlight and maintain detail. That doesn't apply to every sensor. Some sensors can go to 3+ stops over and still maintain detail in highlights. You can find the dynamic range of your camera by calibrating it with a Sekonic L-758 light meter and Sekonic exposure target or ColorChecker Passport. Any DSLR has at least 5 stops of dynamic range, so you could open up 2.5 stops from a highlight and preserve detail. Some have about 8 or so. I don't think any have more than 10 in real world applications, at least not 35 mm.

Dynamic range figures that appear in reviews or tests are often exaggerated or achieved in ways that don't apply to the real world. Some tests boast 18 stops, but it's not always clear how they arrive at that and you can't actually get 18 stops out of a RAW file. To get 18 stops you would have to be able to open up nine stops from a highlight and preserve detail. Try it. There's simply no 35 mm DSLR that can produce an 18-stop RAW file. I don't think 13 stops is realistic either. That's 6.5 stops up from a highlight. When editing a RAW file, Lightroom will let you input only 5 stops, and there's so much noise in any remaining detail there'd be no point in using it.
 
Thanks for that. That clears it up nicely ;)
 
Part of the issue is that while they can say that the camera has a DR of 13-11 stops, much of that range is unusable for 'good photography'. In realistic terms, it's more like 5-6 stops.

This is my take, not sure about scientific accuracy. Exposure with a digital sensor is pretty linear at the top/bright end. The curve goes up pretty consistently and then just hits the end (pure white). But on the low end, it's a lot more like an asymptote, where the curve approaches zero but doesn't get there for a longer while.
In other words, you have to be careful of blowing highlights, because they will OK, right up until they are blown....but you can almost always pull some detail out of the shadows....it won't look good (full of noise etc) but you can still discern details, so technically it's still within the dynamic range.
 
DXO Mark is an independent testing lab that use industry standard RAW image testing techniques and protocols which=h they use to quantify image sensor performance.

Highlight detail recovery is dependent on the how many, if any, of the 3 color channels - red, green, and blue (RGB) are not at their maximum value of 255.

If only 1 of the 3 color channels is 'blown out' (at 255) the most highlight detail at that point in the image can be recovered.
If 2 of the 3 color channels are blown out only a little highlight detail can be recovered.
If all 3 color channels are blown, "forget about it" -there is no detail left to recover.
 
I think 2 stops from neutral grey is generally safe. But there is also a huge variable in "how" your camera meters/displays the histogram. The histograms on my D800 lie badly, the histograms on my D4 are much more accurate.

I don't ETTR because I usually don't have the time/luxury of fine tuning exposures. I generally prefer to bias towards underexposure in order to protect the highlights.
 
Thanks for that Paul. That's exactly the information I was looking for. I wanted to see where the altitude fell compared to B&W film and you answered the very well. I think I'll attempt a simular test with the Df
 

Most reactions

New Topics

Back
Top