RAW processed profiles in PS CS

[Julio]

Vlad and/or NIteam:

I don't mind doing some extra work creating my profiles but I also don't want to waste a lot of time for little gain. Assuming I already have a profile correctly generated from the NI target of my camera for a particular ISO, could you rate the following possible modifications in RAW image processing with PS CS on how much they affect the usefulness of the noise profile?

Please rank using:
1=Of no consequence
2=Slight affect
3=Some affect
4=large affect
5=time to generate a new profile with this setting!

A) Changing color space (PS CS let's you process the RAW into Prophoto, Adobe RGB, SRGB or Colormatch)
B) Adjusting exposure compensation by...
+1 EV
+2 EV
+3 EV
+4 EV
-1 EV
-2 EV
C) Interpolating to a larger image size
D) Interpolating to a smaller image size
E) Increasing Contrast
D) Reducing Contrast
F) Increasing Saturation
G) Reducing Saturation
H) Using CS built-in luminance noise reduction
I) Using CS built-in Chroma noise-reduction

Sorry to put this in a "test" format (I hate tests! ) but I think answers in this format will help me and many others who seem to ask similar questions here.

Thanks,

Julio

[NITeam]

Julio,

I don't work much with Adobe RAW converter so take my answers as an opinion (repeating your order and marking of questions):

A) 3
B)
3
4
5
5
3
4
C) 4
D) 4
E) 3-4
D) 3-4
F) 3
G) 3
H) 4-5
I) 4-5

You may want to test these before doing real profiling. Try to profile two images with one parameter different. Then compare the noise levels in the profile viewer (Ctrl-I) - if the noise level are more than 20% different then consider this parameter important.

Vlad

P.S. I seem to be a bit pessimistic in my rates for all cases. Make you adjustment on this too.

[Julio]

Thanks for the reply, Vlad.

I finally got some free time to do some profiling and while I was at it I found some unexpedted result.

I have a 10D and use Photoshop CS Raw converter (I shoot RAW 99% time) In CS you have the option to process the RAW into any one of 4 color spaces: SRGB, Colormatch RGB, Adobe RGB and Prophoto RGB. For several reasons I've settled on processing into Prophoto RGB.

I decided to do an ISO 1600 profile in all 4 color spaces with everything at default and no +/- EV compensation. What I wasn't expecting was the large difference in profiles using the profile viewer with SRGB having the noisiest profile (9.84 total) and Prophoto the cleanest (6.64)

Here are the full results:

• SRGB, Colormatch, ARGB, Prophoto
  P= 9.84, 8.25, 7.91, 6.64
  
  Y= 3.59, 3.70, 3.39, 3.58
  Cr= 2.79, 2.69, 1.96, 1.35
  Cb= 1.79, 1.96, 1.67, 1.57
  
  H= 7.17, 6.19, 6.16, 5.43
  M= 5.74, 4.66, 4.27, 3.33
  L= 2.99, 2.39, 2.15, 1.59
  VL= 1.88, 1.50, 1.33, 0.97

I'm not sure why Prophoto produces a cleaner image but I'm it glad it does

[NITeam]

Thank you for sharing your results, Julio. Interesting indeed. I am sure these will be useful for many forum members here.

Vlad

[Julio]

I posted the color space and noise results at the Rob Galbraith forum and I got this reply from Thomas Knoll (Photoshop developer):

For a constant amount of VISUAL noise, a wider gamut color space will have a lower NUMERICAL noise. All you have changed is the "ruler units" you are using to measure the noise, not the actual amount of noise. When you convert each image to the printer profile, the numerical differences disappear, and the noise levels in the final print are the exactly the same. Sorry, but there is no free lunch here.

So... never mind