A lot of people here seem to have the impression that doing RAW-like operations on jpegs in LR/CS3 is exactly the same as performing them on RAW files. This is not the case.
From my understanding, in the jpeg case, all the program is doing is tweaking the curves on the 8-bit data from the jpegs. However, in the RAW case, the program is actually adjusting the sensor data from the RAW file.
Example 1: Changing exposure
JPEG: Edit the curves to remap the tones.
RAW: Adjusts the gain from the sensor data. (In effect, changing the ISO).
Example 2: Adjusting WB
JPEG: Edit the curves to eliminate colour casts.
RAW: Adjusts the white point from the sensor data.
What's the advantage of adjusting sensor data? First of all, you have more bits to play with (12 instead of 8) which means posterisation issues are reduced. Also, every time you adjust the curves, you're losing tonal information. More importantly, adjusting RAW data means you're not adjusting lossy data. You're not adjusting curves on jpeg artifacts.
Now, I agree that for all intents and purposes, you can stick with whatever you're comfortable with (jpeg or raw), and only experience will determine what you can/cannot do in PP. However, there will be times when you come across a difficult/pathological scenario and it's in times like these that I'll always stick with RAW.
Exactly! I think the reason everyone thinks this is because Adobe has never said anything about it. They only mention that ACR4 will work with all files, not just RAW.
If people really want to grasp the benefits of RAW read Fraser's book Real World Camera RAW for CS2.




