Still Trying to Convince Myself on RAW

Started Jun 29, 2013 | Photos thread
Michael Fritzen Veteran Member • Posts: 5,477
Re: Gary - Bit OT but --- Still Trying to Convince Myself on RAW

Gary Eickmeier wrote:

Can RAW do HDR without the need for multiple exposures? I suppose the theory is that since it has 12 bits instead of 8 for RAW, then I should be able to go deepr into the noise to retrieve detail in the black areas, and conversely I should be able to go deeper into the whites to retrieve highlight detail.

Hi Gary,

here you are on to something. At least from the "potencial" of the wider data range. However, when there's nothing to detect/record in the low bits, they remain zero and in this situation the RAW data as is wouldn't provide you an advantage. But this can be changed to some amount if the fact "nothing to detect" isn't absolut (no true black) but dependant of the exposure time, i.e. when there is something to detect when you increase the exposure time, then the low bits perhaps aren't zeroed anymore = data to work with = a chance that details start to reveal. Given that 12bits (or 14bits on the A99) are quite "smaller steps" than the 8bits of JPG, and adding the digital nature of the information processing, it's understandable that in the lower bits of a RAW file a weaker (light) information is able already to change the bit from 0 to 1 (smaller step / signal what generates the "1") as if it was the larger step of the low bit of the 8 bit JPG. Again, as long as those low bits remain zero there's no structure, no data to work with. Now transfer this to the high bits where the bright light info is stored. With the "smaller step size" anything exceeding the value of 255 in the JPG file is blown, lost data forever and not recoverable. It's because of the digital nature which knows only "yes" or "no". In RAW there are 4,096 steps (12bits) or 16,384 steps (14bits) so what can go beyond step 16,384 is a proportionally much smaller amount of data. Data kept = data available to work with - at least in theory.

Now it comes to a critical point. In practise the OOC JPG derives from the same RAW data as does the final result of a specific conversion done on a computer which is generally also a JPG file "only". And it needs to be because normal visualization devices represent only that gamut (some specialized ones somewhat more). So no wonder that it may be hard to tell them apart. This is particulary true when one shoots JPG+RAW and tries to get out of the RAW data a better looking image than the camera already created based on exactly the same RAW data. In many normal shooting situation with no specially challenging circumstances considering DR, WB or so the differences in the final JPG file may be so subtile that they can get impossible to identify. In addition, one can bet that the in-camera JPG engine is developped with the specific purpose the generate a "pleasantly looking image" (or as an overly critical may stand: to trick the eye of the observer). It's a bit about "good looking" vs. "quality" - which may be, or not, antagonic.

When shooting situations gets more challenging though considering DR or the camera has diffulties to match the right WB, not only the work afterwards based on the RAW data may provide significant advantages. In reality however the differences start already when shooting because taking the shot in RAW should be considered as a task of "maximimized collection of light information", having in mind that a zeroed bit info is a lost bit info since no workable data was collected. The principle of ETTR (expose to the right) follows this philosophy. And the exposure process adapts to this since there would be no point in having highlights perfectly inside the data range, not a single clipping, when this exposure setting would mean zeroed bits in the shadows. So one goes for a compromise by shifting the exposure to max possible values for the highlights (just short from clipping) in order to gather some information in the shadows. And then one of the tasks of the conversion process afterwards is bringing the exposure levels back to their natural look - however now with a chance for some detail recovery in the shadow areas when the ETTR was able to record some data there. In the conversion process itself this increased amount of information can be written into the JPG file. But finally, when the output JPG is generated it's still an uncertain whether the viewing device is able to reproduce (show) this additional data. It's not all that different of reproducing a high quality music CD on a kitchen's cheap CD player being this the limiting factor (but the data / the "quality" is there). The highest quality ouput device for photos would be the high quality printer.

Hope this long winded story helps you in your evaluation process.

-- hide signature --

Michael Fritzen

Post (hide subjects) Posted by
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow