RE: filmscanners: Vuescan: "device RGB"

["Richard N. Moyer" <dickmoyer@mail.earthlink.net>]

There is a lot that doesn't add up - - regarding PCD "space", and VS 
using PhotoCD "space".

1. You will note that you cannot do a profile conversion (profile to 
profile) in Photoshop to a PhotoCD (space) profile, of which I have 
about 8 of them in my ColorSync folder.

2. PhotoCD format - - and their ICC profiles ARE proprietary, and any 
use thereof would require a license.

3. If you examine each PhotoCD ICC profile, you will see a number of 
CLUT listing, and the sources are labeled "secret". That word. Each 
16 bit profile has about 8 listings, all "secret" - - or proprietary. 
This is NOT the case if you examine Kodak's ProPhotoRGB space - for 
example.

4. You can open a PhotoCD - or a PictureCD using Kodak's embedded 
profiles, or can use one of the other Kodak CD profiles in your ICC 
profile bank, but you cannot save to PhotoCD, meaning you cannot 
embed (or format) Kodak's space.

5. If VS uses PhotoCD, what is he (Ed) using for tables, if the CLUTS 
are proprietary? And anyway, what good does it do in this case? I 
really wonder if this is fact that he is "assuming a space", and he 
thinks that space looks like a PhotoCD space.  In an (off-list) 
correspondence with me Ed championed sRGB The sRGB profile consists 
of about 16 data points. That's it. A synthetic space meant for video 
and graphics on the web.

6. The scanner has its own RGB color response, usually evidenced by 
its profile. The profile merely maps the scanner's RGB data in a way 
consistent with the scanner capabilities, and therefore does not try 
to "remap" into a space either smaller or larger, or distorted, from 
the scanner response. Therefore what you get in Photoshop is 
"undistorted" RGB values, as delivered by the scanner. Most modern 
scanners have RGB responses far in excess of the media they scan, 
including E6, or Ektachrome, the media generally acknowledged to have 
the widest color gamut. In fact, the Nikon LS1000 has a gamut,or 
color response that well exceeds Ekta Space; the Imacon is way, way 
out there. Anyway, the scanner profile neither adds or subtracts from 
its intrinsic gamut.

7. If a profile is not used by VS, you have what is called "raw 
data", meaning as is, and not mapped. That is, if Ed isn't 
"remapping" in the background that you don't know about. And, even if 
the scanner profile is applied, that RGB will not budge from the 
untagged values. This is because the profile merely maps out what the 
scanner is doing in the first place.

8. All Kodak PhotoCD profiles exhibit very unusual gamut profiles, in 
either L.a.b., Yxy, or XYZ space. The are all complex. The only thing 
you can say is that they universally fit within Kodak ProPhotoRGB 
space. I don't know why Ed would even consider these spaces - - 
unless for PR reasons.

9. If VS is using some sort of "space" which alters the RGB values, 
and then doesn't "tag" the resulting image - you never know how much 
distortion has been introduced from "real" values. The image comes to 
you as untagged. If you apply a profile in VS, then the image should 
come to you as "Tagged", and you should see that (ICC profile) in 
Photoshop. Without this requirement, or data trail, you are in never 
never land regarding the fidelity of the original image.

You might be bridging into touchy, unsafe waters by forcing out 
clarity on this issue - - Ed has a useful product to many on this 
list. And a good product to many. Better not lift the lid too much to 
look underneath.



>Tony writes ...
>
>>  The raw scan is in an unspecified device space,
>>  scanner RGB. Ed's transform, applied during the
>>  production of the Crop file, munges that against
>>  his characterisation and the result is a scan
>>  with altered data values within Vuescan's working
>>  space (which I previously said I thought was maybe
>>  sRGB, but as has been pointed out it ain't,
>>  it's Kodak's PCD space -
>>  ...
>
>       That's more comforting.  For taking a raw scan, from any scanner, and
>into the variety of color spaces Vuescan offers, I assume Ed is
>assuming (1) a specific scanner may have the potential for delivering
>a wide gamut of color ... (2) a transform from that gamut to any
>internal color space can squash that gamut, and you'll never get it
>back.
>
>       That being said, and altho I trust Ed, I know little about PCD RGB,
>and there seems to be little available regarding comparisons with the
>common working spaces (if comparisons can be made ... some of what
>I've found would imply apples and oranges).  I will assume, until I
>realize otherwise, VS's internal space is sufficient for 1 & 2.
>
>shAf  :o)