The RAW Truth: Does it really increase the dynamic range?

One of the big advantages of using the RAW format is the ability to

change the exposure value after the picture has been taken. Or at

least that�s what some say. I don�t believe it. I think this is

nothing more than could be done with a jpeg in Photoshop using levels

or curves.

Regardless of how you interpret the info from the sensor when you do

your RAW conversion, the fact remains that you used some combination

of shutter speed and aperture � this cannot be changed. If the

pixels were driven into overexposure no amount of interpretation

after the fact will recover that blown out highlight. On the other

side, if you underexposed, yes you can raise the values for the

pixels, but you can do exactly the same with Photoshop.

Have you tested your hypothesis or is this just an untested opinion? RAW files have

other characteristics than just the ability to change the exposure. Basically shooting

Raw (or if you use Nikons, NEF ) records the raw data from the sensor rather than

running it through the onboard processing software/firmware. In theory (I'm not

enough of a geek to have tested this) there may be additional exposure latitude.

There is certainly is more color information recorded.

Have you ever tried to do significant color corrections to a JPEG file? Big adjustments in color or tonality quickly reveals compression artifacts and quality problems. Seems like the quality of the JPEG file is locked into the particular color, contrast and tones used.

If the RAW conversion converts to 8-bits/channel, then there is value in keeping the RAW file as you are losing image detail in the conversion process.

 

If the RAW conversion converts to 16 bits/channel, the process is not losing any dynamic range. The only real value in keeping a RAW file in that case is if you have a mechanism for changing the interpolation that is done due to the CCD pattern. In theory, a very detailed image should not use the same type of interpolation as an image with soft smooth transitions. In the first case, you'd want a conversion process that has a sharpening effect (sharp transitions). In the latter case, this would not be desireable at all (you'd want a smooth interpolation).

 

Ideally the conversion process would analyze the image and determine how detailed it is before applying the required interpolations or alternatively give the user control of the interpolation parameters.

 

The same basic effect could be applied with a sharpening filter, so saving the RAW file would probably be of very limited use (marginally better quality if you have some control of the conversion process) as long as it provides a 16 bit/channel file after conversion.

My test have been in the real world. I have taken images saved as RAW and jpeg, made the best print from both and could discern no visable differences. In addition, using a RAW file(actually NEF) and varying the exposure comp slider while at the same time observing the histogram, I see the histogram shift intact along its axis - exactly as it does when you do a levels adjustment.

which NEF processor did you use?

Nikon Capture.

I'm curious what exact equipment you were using (which Nikon body, which lens, etc...).

 

On our D100 with primes and the pro zooms, the differences between jpeg and NEF are definitely visible within PS.

 

Can you post your results? Please don't images that have been compressed for web display, but maybe crops of the full images.

The difference between a RAW file with exposure compensation and a JPEG file with post processing will mainly be in the noise levels in the deep shadows.

 

With the Canon software this difference is clear.

 

Of course this mainly results from the RAW files using the full 12-bit data and the JPEG files using only 8-bits.

 

It is certainly easier to recover underexposure than over exposure. Once you've driven the white levels to the max, you can't recover lost data.

<<On our D100 with primes and the pro zooms, the differences between jpeg and NEF are definitely visible within PS.>>

 

Sure they are visable "within PS", I'm talking about the print result.

Take a look at my two pictures posted in this thread: <a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=006F4M">http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=006F4M</a>. Shooting RAW gave me much more detail in the highlights--information that was lost when I shot JPEG even though I metered the two pictures identically. Shooting this picture in RAW, then, effectively increased my dynamic range. You can take the JPEG and try to adjust it with curves, but you won't find any more detail in the clouds or the window reflections. That information is gone.

1. RAW files do provide the opportunity to fix underexposure and retain better detail in shadows, compared to simply boosting levels in Photoshop. They also provide data for use by noise reduction algorithms, e.g. in C1. This is due extended intensity recording range compared to JPEG.

 

2. RAW files (on some cameras, such as 1Ds) do provide opportunity to fix blown up highlights, up to 2/3 to 1 stop.

 

3. RAW files do allow quick and easy fixing of white balance in post-processing.

<pre>Try a Polarizer with JPEG files, results will be similar to unpolarized RAW files.

 

As to RAW vs. standard JPEG the former is generally better. The quality of RAW though can vary.

 

Options I know of for Canon files there are these:

Adobe

Phase One

Breeze

Canon

 

Quality can vary depending on which software you use and what camera created the image files.

 

 

-Andrew</pre>

Any gains in RAW (or 16-bit TIFF) files over 8-bit (per channel) files are found in the lower bits, or shadow areas. Simple conversion from 12-14 bit files to 8-bit files simply discard all but the top 8 bits. That said, RAW files have the same tendency to blow highlights as 8-bit files. You can't restore information that isn't there.

 

Raw files are much smaller than TIFF files with the same bit-depth and resolution. On the other hand, TIFF files are compatible with most image-editing software. It makes sense to shoot in RAW mode, then convert to TIFF in post production.

 

You notice the difference between 16-bit and 8-bit files immediately if you use LEVELS to adjust the picture. The histogram for an 8-bit image develops gaps after adjustment, which may appear as banding or posterization in the final print. The histogram for a 16-bit image remains smooth as long as the adjustments do not exceed the existing dynamic range. In effect, you are positioning an 8-bit view within a 12-16 bit window.

<p><i>> RAW files have the same tendency to blow highlights as 8-bit files</i>

 

<p>Every media does, but some cameras (like 1Ds) have large dynamic range and reserve part of it for highlight protection. You can dial negative exposure compensation in RAW converter and details in seemingly blow-up highlights will come back, upto around 2/3 or even full stop.

This might be of interest to you...it's worth reading all the way through...<br>

http://www.luminous-landscape.com/tutorials/expose-right.shtml

<br>

Cheers

I recently bought a Canon EOS300D. Usually I use my Nikon F4 with full range of lenses loaded with K64 / Provia or Velvia. I decided to spend 2 weeks shooting digital with the EOS300D only (with good lens -> 17-40 F4). I shot nearly 820 pictures in 12 days. All the pictures in Raw format with the exception of some (roughly 20) in jpeg. At home, I browse the RAW pictures and noticed some of them are underexposed or overexposed. I then use the RAW utility (called FileViewer for Canon) for altering the exposures. Results are quite incredible, even with exposure correction of 2 stops.

 

Here follows a personnal explanation (I am an engineer -> technical). In raw format, for the Canon EOS300D, the pixels are recorded in a 12 bits resolution. Converting to 8 bits drops 4 bits. The 4 bits that are dropped corresponds to 4 stops (log in base 2). This deduction seems correct since the Canon RAW utility allows to alter the exposure with a range of +- 2 stops (-> 4 stops). Therefore, there is a huge advantage keeping the files in RAW format that allows 12 stops of exposure range. The aforementioned explanation is just personal deduction from what I experienced practically with the EOS300D raw format and from my personnal professional knowledge (computer science engineer). Apparently, the exposure measured with the camera is centered on the 6 bits of the sensor resolution.

 

_michel moreaux