My RAW Workflow

Since I "Edit" 100% of my final photos in one way or another anyway, RAW workflow doesn't add much hassle to things. At the very least I color grade the images. (Either making them "accurate" or not accurate... it's a creative choice) I use Photoshop "Adobe Camera RAW" loader.

Some people are critical of the quality of the CS2 RAW loader . I find it good enough for my uses. And the convenience of doing everything in one place outweighs any minor quality differences there might be.

JPEG / 8 bit doesn't have the bandwidth for slamming the color around too much so I usually work in 16 bit from the RAW files. "16 bits" really isn't as "accurate" as you might think simply because of how the linear output of the sensor is mapped up to your gamma corrected working color space / display. But it does carry a little extra data that would otherwise get thrown out. It's nice to hang on to that as long as possible. One good thing about working in larger color spaces like ProPhoto RGB is that less of the data is likely to be "rounded out" of even the 16 bits. Keep reading and it might start to make some sense... or none at all.

Here is break down of my process.

0. When shooting, "Expose to the right"
1. Copy RAWs to the computer on a card reader
2. Open RAW in CS2 RAW loader as 16bits/channel, ProPhoto RGB color space (it is the widest gamut color space that ships with Photoshop)
a. While in ACR (Adobe Camera RAW) I do the basic white balance and if required (as is usually the case when exposing to the right) exposure compensation.
3. Image is now in CS2
4. Edit image as required (Still in 16bits/channel, ProPhoto RGB)
5. After image editing is done I direct the output images as required.
a. Save a Photoshop format image with all data if required for later editing
b. CONVERT to sRGB THEN 8 bits per channel for web / modeles proofs
c. Convert to AdobeRGB or CMYK as per printers requirements (for print) You can leave it at 16 bits if you want. Or not. It depends on the printers pref.

The important thing to remember when working in a more advanced color space like ProPhoto is to make sure you re-map it to something usable by the target audience. The ProPhoto color space wont render correctly on 99% of most image viewers because they have no concept of color spaces. Also, 16Bits/Channel is complete overkill for almost all displays. You only need it as a working space to help reduce rounding error. And, depending on your camera, exposure, etc. the signal to noise is WAY worse the precision of 16bits anyway. You are sampling "noise" for no reason. You keep the extra bit depth around as long as possible mostly for the benefits to image processing. (less rounding error) And the slight improvement in capturing actual precision from the RAW file it gives. (There is some signal buried in that noise after all) It's often ok to throw out the extra bits once you are done with your editing. But not until you are done. It's a one way street. There is not much point in promoting 8 bit images to 16 bit because the precision is already gone.

Basically, I try to do all editing in 16bits/channel, ProPhotoRGB and only convert to the smaller gamut / bit depth as the very last step.

There you have it. All my secrets.