Paris

by Jon

submit your photo


Hall of Fame
View past winners from this year

Please participate in Meta
and help us grow.

Take the 2-minute tour ×
Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It's 100% free, no registration required.

I'm new to photography and what I've seen so far:

  1. ViewNX allow me to change exposure +2/-2 stops in RAW
  2. HDR is about larger dynamic range and merging shadow details from overexposed image with highlight details from underexposed image
  3. RAW contains more bits than JPG and whats visible on monitor that's why it is possible to protect from highlight and shadows..

So seems to me RAW already contains all required HDR information and additional RAW/JPG photos will not produce better result for HDR (only worse because of alignment process).

There is a lot of controversy on this issue. Can anyone give an theoretically proven answer on this question?

share|improve this question
    
Related (and maybe actually an answer in itself): How can I get a good HDR image from a single RAW file? –  mattdm Jun 1 '12 at 14:17

5 Answers 5

HDR from a single RAW images does not add any dynamic-range than is captured. If the scene exceeds the dynamic-range of your camera, then no matter what you do the RAW will contained clipped channels on one or even both ends. Even if you have the best camera and use it at its optimal ISO setting, the most you get today is just above 14 stops of DR.

Taking a bracketed exposure let you capture a much broader range. Considering the range of modern DSLRs, you can easily bracket at with 5 EVs between each shot. That will expand 14 EV of DR to 24 EV.

Of course part of the question is the way to see this and that is where tone mapping come from. There is only HDR Display and it is very expensive, so everyone else must content with seeing something within the confines of normal dynamic-range (8 stops at most). So, yes, you end up seeing something which is not HDR but you still see things which had more range. Without being careful this obviously results in atrocious look HDR has become famous for.

share|improve this answer
    
The D800 is still not actually capable of more than 14 stops. The hardware itself is capable of, according to DXO, is 13.23 stops. No matter how advanced your post-processing software is or how meticulous your technique, if you try to capture a scene with 14.4 stops of DR, the 13.23 stop sensor will blow the highlights. A full 1.2 stops of blown highlights are gone, downscaling to a standardized 8mp "print" image is not going to change that. –  jrista Jun 2 '12 at 5:02
    
DXO's statistics are excessively misleading in this respect, almost criminally so. Claiming an additional 1.2 stops, more than a doubling, of dynamic range capability due to the simple act of downsampling (with a couple "bonus" points thrown in here and there to boost the number even higher than downsampling alone would allow) is simply producing an invalid statistic from a hardware standpoint. (Even mathematically, given the nature of digital bits, the camera hardware could at most, with a 14-bit ADC, be capable of 14.0 stops before blowing highlights...assuming perfection/zero overhead.) –  jrista Jun 2 '12 at 5:05
    
Yes, I discovered that. As a matter that was one of my first question here: photo.stackexchange.com/questions/4762/… –  Itai Jun 2 '12 at 12:59
    
Hmm, interesting. I even commented on that back then, but as far as I know today, ALL attenuation of image data occurs digitally in DSLR's these days. ADC's are linear for any given ISO, and often have fractional gain at most ISO settings, which leads to quantization error (and therefor further loss in DR.) People don't seem to understand the value of a "doubling" either...if a Canon sensor has 11 stops of DR, and a Nikon has over 13 stops, thats more than FOUR TIMES as much DR that the Nikon has...thats an incredible difference (even without surpassing 14 stops.) –  jrista Jun 2 '12 at 16:46

The number of images required to produce an HDR result without clipped highlights or shadows depends entirely on the scene being photographed, not on the dynamic range of your output device. RAW images may contain a greater range than you can display on your monitor, but that doesn't really matter as you will compress the dynamic range anyway as part of the HDR process by tonemapping.

A single RAW image may still not be enough as the dynamic range of your scene could be huge. We tend not to notice this as our brain compensates for it extremely well, but any situation where the light-source is in shot but not illuminating part of the scene will likely prove impossible for the single image approach.

The proof of this lies in the fact that once you clip highlights in RAW the information is gone forever. Likewise when dark areas descend into pure noise there is no way to recover detail. So there will always be scenes where the dark parts are too dark and the bright parts are too bright to get away with using a single image.

share|improve this answer

I'll focus on this statemenL=:

So seems to me RAW already contains all required HDR information and additional RAW/JPG photos will not produce better result for HDR (only worse because of alignment process).

RAW contains either a 12- or 14-bit A/D conversion of the sensor readings. Not quite that simple, but depending on your camera, normal case is 12 bits, better case 14 bits. Think of HDR as a series of n images (where n is however images you want to use to get the detail you need in the highlights and shadows). Ideally, each image's dynamic range overlaps the dynamic range of the next one in the exposure group by a small amount. At this point, I think you'll see that in a 3-exposure range, assuming overlapping by about 1/3 and 12-bit A/D converters, you will still be using some 28 bits of data. Add more exposures and you can expand this up to 32 bits, which is where most HDR technologies stop.

So it is not true that a single RAW file contains all this data. With bracketed exposures, you are capturing extra data for both the low and high end beyond what you could have captured in a single frame. It is, however, true that you have to exercise great care not to move or bump the camera (and hope there are no moving objects in the frame). So there is some image degradation (that is not quantifiable) associated with multiple frame HDR.

share|improve this answer
    
Took me a moment to ponder this answer for it to make sense, but it is certainly the most definite answer to the question, there is simply not enough stored information compared to multiple differently-exposed images. Following the same logic, this makes a correctly exposed image better than tweaking a RAW's exposure in post? –  MattJ Jun 1 '12 at 22:18
2  
Yes. I believe a properly exposed image will always win. Anything else is either done for effect or you will be making a rescue. And, if you don't have time or inclination to spend much per-image time, getting it right in-camera is a must. –  Steve Ross Jun 1 '12 at 23:34

Your camera only has so much dynamic range. So to take the underexposed and over exposed images at the same time as the normal shot allows you to really pull details out of the shadows and recover the highlights much much better than a single RAW shot. The other thing also is that if you create a virtual copy of one RAW file and reduce or increase the exposure of it in post you risk introducing artificial noise which would be detrimental to the final image.

You'll get a lot more 'information' out of three separate photos at different exposures than you would from just one.

share|improve this answer

RAW is better source for HDR since it also contains unprocessed linear responses from the sensor.

JPEGs are usually processed within camera, at least a gamma/contrast curve is applied, so the pixel values no longer correspond to the true irradiances.

One of the purposes of HDR is to capture relative radiance, that is when the cloud is 120 000 times brighter than wall of a building, so the pixel value is 120 000 times bigger.

Gamma and contrast curves can be applied at the end of the HDR processing workflow for better perceptual quality (our sight responds non-linearly to amount of light, contrary to the CCD/CMOS sensor).

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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