Question:

Dynamic Range of Digital Cameras

by  |  earlier

0 LIKES UnLike

I was reading recently an article that stated a digital cameras' sensor is more like slide film; meaning, the latitude of digital photography is very low. (There is very little that separates proper exposure and overexposure). I did some calculations with my own camera, a Nikon D50, and found about 7.5 stops range. While I heard black and white film has 14 stops. My question is, what is the best way to replicate the latitude of black and white film? Is HDR the only way to approach film? I've heard good things about the DR of the Fuji S3 and S5 Pro, but was wondering if there were any other cameras with a better DR. Any comments would help. Thanks.

 Tags:

   Report

7 ANSWERS


  1. It sounds like you got half the story and what you got from the magazine seems pretty misleading.  This is an area I've done a fair bit of research into and spoken direct with the manufacturers as to the real figures.  

    And to add to the confusion you should take into account the colour space involved - sRGB is a smaller space than Adobe (1998) RGB for instance, but it produces more pleasing natural tones because the tones are compressed into a relatively SMALL space / gamut, just like most film!

    The 7.5 stop answer above is totally absurd (Injanier) - bit depth is exponential, so saying 8-bits equates to 8 stops is to misunderstand the process... completely!!  

    Bit depth does NOT determine dynamic range, in fact it has no bearing on it whatsoever, it is simply a finer measure to count intermediate tones (producing a higher count of information / more data, which reduces the need for interpolation and that makes intermediate tone prediction more accurate), so it does NOT give a wider range of light capture - under any circumstances.

    In particular, the statement in the magazine re chip dynamic range depends from device to device AND whether you are shooting tiff, jpeg or raw information (which seems like it confused whoever was writing what sounds to be a very badly researched article):

    Colour neg = 4.5 stops

    Tranny = 4 stops

    C41 'wide latitude' b/w neg = can be around 9 stops or more, but the implication is that it needs holding back and printing in to get some of that information printed or the result will be as flat as a pancake or just the same as other film if printed 'as is'.

    Tiff and jpeg = 3.5 stops

    Raw = can be 11 stops or so

    The print window = 3.5 stops***

    The mag was wrongly informed on b/w film (or it wouldn't print 'straight'), or it would perform exactly as other films with regard to the 'print window' - obviously.

    So, digital comprehensively out performs film in this area and the implication of a 3.5 stop print window underlines the practical impact of using wide dynamic range media... more work or you'll not get any benefit from the extra data, you just get more latitude as to where to chop extreme data.

    You also need to take into account that manufacturers often massage these figures - it's like comparing scanners where one has a much higher price tag and quotes the optical resolution and another quotes interpolated resolution.  The flatness of tone curves etc has often been a source of debate among manufacturers in the past and this is a very practical limit as to how very light or very dark tonal information can be used in a real life sense.


  2. The Foveon sensors have 3 layers of sensors per site recording RGB at each pixel. These may have the highest dynamic range in a 35mm format sensor. I haven't seen any actual test results to prove it though - The photos look good.

    Kodak has a new 50MP sensor used by Hasselblad with new technology (4 channel) designed to reduce signal to noise ratio. That should produce a better dynamic range.

    Those are a few options to look at before the post processing solutions.

  3. Better than the S3 and S5? None that I know of... even the older Fuji S2 has higher DR than the Nikon D200, D50, D70 etc.

    Compositing images for HDR seems like the best approach.


  4. JPEGs can display an 8-bit grayscale. Since each bit represents a power of two, its the same as one stop photographically. So your 7.5 stop calculation is about right. RAW files are typically 12 or 14 bits, which is closing in on what film can do. But I think sensors are more linear than film, which tends to flatten out in the extremes, so you lose depth of detail in the highlights and shadows compared to B&W film. Fuji uses sensors with both high-gain and low-gain pixels for extended dynamic range. I can't testify as to how well it works.

  5. Check out HDRI at http://en.wikipedia.org/wiki/High_dynami...  Basically you take three pictures at three different exposure ratings, and when combine can exceed the dynamic range of film.

  6. Indeed this can all be a little confusing..

    While mono film images have a higher dynamic range, the curved [rather than linear digital] response of film tends to compress the dynamic range of the photo so it is viewable by the standard media. For instance, if you had a digital camera with a 14 stop linear response, there would not be a printer or monitor capable of displaying even half the information in the photo.. the contrast and clipping would be far too great to view. Your test with your D50 may be inconclusive [depending how you conducted your test] as you may have been measuring the dynamic range of your monitor ..which is almost certainly less than the image sensor. Its comparable to judging the sound quality of a recording through speakers incapable of producing all the frequency range ..you are limited by the weakest link in the chain.

    The curved response is to some extent simulated by the S5 pro making it ideal for simulating black and white film. This curve means that the bright areas of the image are recorded with a lower sensitivity thus allowing the detail of up to 14 stops [in theory] to be compressed to a picture of something like half that dynamic range. Therefore the curved response of film can be compared [although the results can differ greatly] to the tonemapping process of hdr. A non-tonemapped hdr image could easily reach 14 stops of information but this is useless as all viewing media again is far less. This compression process is something that will become more widespread in the digital world in the coming years both in cameras and software form.. for example there may be a hdr tonemapping algorithm specifically designed to emulate the b&w film curve. For now it seems Fuji are the market leaders in this area.. but there is the limitation of the 6mp actual megapixels from the S5 pro.. even though it outputs files at 12mp.

  7. HDR is good but not a cure all for sure.

    What if your subject is moving? Only film can tackle the dynamic range you are talking about.

Question Stats

Latest activity: earlier.
This question has 7 answers.

BECOME A GUIDE

Share your knowledge and help people by answering questions.