What are the numbers for Dynamic Range of a high end digital camera

You can find DxO SNR charts for some cameras at http://www.luminous-landscape.com

There's not an easy answer. The limits are staturation of the sensor on one end (highlights), and noise (shadows) on the other. The question is, how much noise can you take before you say that there's not enough signal?

In general, digital cameras have slightly more dynamic range than most slide films, but aren't close to negative films. Somewhere between 5 and 6 stops. Since there are only a few makers of the sensor chips, this isn't going to vary widely from camera to camera.

Why this isn't measured and reported on in reviews is beyond me. I think it's a pretty important specification - it would be one of my top considerations if I were in the market for a digital camera.

I had some trouble getting this straight in my head. Actually I'm not sure if I even have it right now.

I used to distinguish dynamic range by its practical effect on pictures. Slide film would blow out the sky, but negative film would keep detail even in the clouds. The effect is as if it had compressed the scene brightness, and it's hugely difficult to pull anything meaningful from the shadows of a underexposed slide.

With digital and the incredible ability to jack the shadows to emulate negative film, I think we have to distinguish between the effect of a compressed brightness range and the sensor actually recording more usable detail. Or is that the difference? Is it only appearance we're after?

Let me put it another way. I take two pictures, one with negative film and another with digital exposed for highlights. Then I use Shadow/Highlight and noise reduction in PS to make the digital JPEG look like the negative. Do I suddenly have more range?

Now let's say I used a RAW instead, and used two files converted to expose for darks and lights. Do I have more range now?

DI

The best digital (some 22mp med format backs) have 11 stops of usable latitude - way better than neg film. Here I'm talking usable not compressed highlights or muddy shadows. My 20D when shooting RAW has about 8 stops - again this is usable dynamic range that can be extracted with software. Neg film requires a lot of work to extract this kind of range and overexposed light areas get very grainy and compressed. Shooting Jpeg limits you to around 6 stops which just about beats tranny.

I'd be very surprised to see any affordable digital camera or back with anything remotely approach an 11-stop range, manufacturers' claims not withstanding; not in this decade, anyway.

Fuji claims a wider than average dynamic range for its S3, mostly in the highlights which tend to blow out. So far tests show that the S3 has the potential to live up to this claim, altho' it appears to be somewhat hindered by Fuji's implementation. Maybe a firmware upgrade can correct some problems.

For now it's still much easier and more cost effective to get the widest dynamic range from negative film. Even badly underexposed negative film can be massaged by scanning and digital image editing to yield shadow detail that would be almost impossible to wring out through conventional printing.

Hi Perfect:

Good question --- comes down to the difference between exposure-latitude and dynamic-range. If, for example, a sensor has an exposure-latitude of 1 under and 1 over that means you could intentionally underexpose or overexpose by 1 stop and still produce an acceptable image. A typical color negative film has about 2.5 stops of exposure-latitude in each direction a typical slide film has less than one stop in each direction. In general, a typical DSLR sensor is about the same as slide film.

I have seen dynamic-range reported up to 13 stops in dpreview. I think this refers to a differentiation of pixels at the extremes. For example a pixel that is 6 stops over exposed might have a brightness of 253 and one that is 6.5 stops overexposed might have a brightness of 254. A difference for sure, but not one that is printable or linearly related to other brightness values.

To bring out shadow or highlight detail you should adjust exposure lighting to easily fall within the exposure-latitude of the media you are using by using flashes, reflectors, filters, double exposures, etc.

Unfortunately I am as guilty as some other people are in using these two terms interchangeably.

Best, Fred

Thanks all,

6 stops should be plenty assuming I have my head in gear when making images. 8 stops would be useful and 11 stops would be too much for the vast majority of my subjects.

I prefer to wait for the right light and since I do mostly landscapes a little planning usually provides that. i.e. late evening or occasionally early mornings.

R. N Clark has a few very interesting pages with test results and a ton of links on this subject. It seems like digital has gotten an undeserved bad reputation in this respect:

http://clarkvision.com/imagedetail/

I'd guess about 9 stops from darkest texture to lightest texture on my Canon 1Dsmk2. This was shot with a Leica 19mm (ISO 400 f/5.6, 1/25s handheld). Often a good indicator of dynamic range is how it handles mixed lighting; in this case, it's perfectly acceptable IMO. I didn't have a spot meter on hand to check, but I'd estimate 9 stops. A CN film like NPS would likely add two more stops. Just my quick off-the-hip based-on-experience opinion, not based on any rigorous testing. But as a digital camera I think it's pretty good. (This was shot raw and converted in C1 Pro 3.6, Etc LoSat profile.)

San Francisco City Hall

What camera did you shoot this with? Digital or film? It's not clear from your post.

Canon 1Dsmk2. It's a digital camera.

Right, but with a Leica 19mm?

Yeah, I use a Cameraquest R-EOS adapter, and the lens has been slightly modified to clear the mirror...

In response to "I cannot find any quoted figures for the Dynamic Range of any digital camera" :

The SMaL cameras that I have been experimenting with have a claim on SMaL's website of 120db. The flatfoto2 documentation reduces that to 85db. Note that SMaL uses a patented process that is capable of varying voltage pixel by pixel. I have several technical threads in digital cameras forum that have explored this but for the most part have been rejected due to cheap plastic cameras containing SMaL imagers. I personally am convinced that they do capture a wider dynamic range than the best and most expensive digital cameras when using a single exposure. The biggest issue is a current lack of ability to display higher ranges on computer monitors or worse paper prints. A human eye that is taking continuous exposures is sensitive to dynamics as high as a million to one (or 20 stops or 120db). So for absolute realism in future photographic images displayed on future DLP devices two more orders of magnitude will be necessary. SMaL was small, Cypress is not so small, only time will dictate true value of Keith Fife's patents which began during his days at MIT.

I always think that its important to remember that a print, hanging on your wall at home, has, at most, 5 stops of dynamic range. i.e. from blackest black to whitest white is around 5 stops for a glossy print such as a cibachrome. This is based on measuring with a spot meter which more closely matches what we as humans see rather than a densitometer which measures light shining at a specific angle and reflected at a specific angle. i.e. it ignores the diffuse reflections which we see whereas a spot meter is measuring the diffused print reflection that we actually see.

The point here is that, whilst a camera may be able to record 10, 15 or 20 stops of range, it is impossible to reproduce that range in a print and trying to do so will will result in a very compressed dynamic range. Whether that gives the result you want is entirely a subjective consideration. Yes, you can capture more information than a print is capable of reproducing, but the work involved in post processing to achieve a reasonable print from it may be large.

It's also important to remember that the human eye is constantly adjusting aperture as it scans a subject so what we see as we look at bright clouds and then deep shadows in a subject is not the same as a camera will reproduce. If a print could reproduce 20 stops of range our eyes couldn't view it all without some becoming totally black or blown out depending on what our eye's aperture was set to.

http://www.dpreview.com/reviews/print.asp?review=sonydscr1&page=17

350D has a good sensor (ISO100 = 8.1EV, ISO1600 = 7.6EV) and even better noise performance.

Boky