4

Several years ago, I asked Why are blown highlights particularly bad in digital photography?, and if you look there, you can see some fairly convincing answers as to why this is.

In an answer to a much newer question, it's noted that "highlight recovery on most modern DSLRs is superb", and I've seen similar statements elsewhere.

Is this, in fact, the case? If the above experiment were to be repeated, would digital fare better?

If so, is this because of:

  1. Improved sensor technology?
  2. Improved features like highlight-protection at a hardware level?
  3. Better RAW conversion algorithms?
  4. Or, something else?

Note that I'm not talking about expose-to-the-right, which is really just a fancy way of saying that there's more information when there's more light (which is kind of obvious when put that way). This is about the issue where the amount of light reaches the limit of exposure, and the falloffs around those areas as the exposure of a part of the frame approaches that limit.

  • While I'm not sure about formal studies, I have numerous examples of anecdotal evidence of photos that were exposed such that the JPEG preview was near white that were able to extract to proper exposure. I know JRista has some similar style demonstrations. Presumably the increases in DR is a significant factor towards it, but I'm not sure what actually changed along the way. – AJ Henderson Jul 21 '14 at 19:49
  • I guess an interesting test would be to push it to the point where the exposure actually clips to the point that RAW recovery can't do anything for you and see how that looks. I'd hazard they are probably internally exposing to the left for JPEGs to protect the highlights since they now have fewer noise issues, but I'd guess once you finally do hit clipping, it may still be rather harsh, but that is complete speculation on my part. – AJ Henderson Jul 21 '14 at 19:55
  • Your previous question is from 3 years ago. I'm sure you know this, but that's only 1-2 generations old. I think the question would be better if this were highlighted and the question asked if the current generation is better than the previous generation -- because that's really the only difference between this and the other question, right? – Dan Wolfgang Jul 22 '14 at 0:10
  • @DanWolfgang That's still a long time in technology, but, sure — you can presume that I have some skepticism of the premise in asking this question. But on the other hand, that's assuming linear improvements; maybe there's been a breakthrough, or we just happened to be at an inflection point. – mattdm Jul 22 '14 at 0:21
  • In many typical cases e.g. a bright day outside, one should be able to estimate what the brighness should have been. If the Sun is shining and the ground has a certain brightness while a wall is overexposed, then from an estimate of the angle at which the Sun is shining (e.g. from shadows plus location and time of day) it shouldn't be too difficult to come up with a ballpark estimate of the brightness. – Count Iblis Jul 22 '14 at 1:26
2

On modern DSLRs, the dynamic range increased, sometimes significantly. But there is a theoretical limit to "highlight recovery": If the (digital) value read from the sensor is all "1111" (e. g. the highest possible number), there is nothing that can be recovered, because all pixels which have the max. value are effectively "the same".

A film on the other hand has no real "1111" (highest possible value), so there you can nearly always recover "something", because there are always little differences.

The third thing that comes into this equation is the displayable Range of the file format. If you use an 8-bit JPG, you are quite limited. But usually you can't see the difference, because your monitor also uses only 8-bit. So if you have a 16bit file, you have plenty of values outside of the range that an 8-bit monitor can display.

So, if any, you can use the nowadays greater dynamic range to "compress" the highlights into the space that can be displayed within a jpeg, thus having a "better" highlight recovery.

  • 4
    8 and 16 bit don't necessarily work that way — the black level "0" and "100%" white can be the same, just with bigger steps in between. – mattdm Jul 21 '14 at 20:30
  • But very often, you can recover some highlights from Images which look overexposed in the preview - at least if you use the RAW-Image (which has also a greater dynamic, depending on the camera between 12 and 16 bit). So it is also possible to use the additional steps for "a greater distance" between pure black and pure white. Also HDR works that way... – Vertigo Jul 21 '14 at 20:34
  • 2
    So let me be more clear. :) Are the bits being used that way, in a way that they weren't five years ago? – mattdm Jul 21 '14 at 20:45
  • 1
    That is something, that only a manufacturer can answer definitively, but in my opinion, yes. Due to the greater dynamic range of today's sensors, you have just a better measurement of things which are very dark or very bright. – Vertigo Jul 21 '14 at 21:45
  • Doesn't film have "effectively transparent/all black emulsion" to offer as an analog 0x0000/0xFFFF? – rackandboneman Aug 20 at 16:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.