Multichannel colour change

[David Hatton]

Briefly, the workflow for 3 colour gum negatives in James says 'open image in rgb, auto levels, multichannel, split channels etc..'

When I convert to multichannel the on screen colour changes. There seems to be too much of a magenta cast.
Does anybody have any idea why this should occur and how do I prevent it? I'm assuming this colour bias will show in the final print.
Anyone have any ideas ?
Thanks
David

 

[Bob Carnie]

I open in RGB go to the Channels Pallette hit the little triangle , split channels and the red , green and blue channel pop up in greyscale, invert , and flip horizontal.
send to printer.

Briefly, the workflow for 3 colour gum negatives in James says 'open image in rgb, auto levels, multichannel, split channels etc..'

When I convert to multichannel the on screen colour changes. There seems to be too much of a magenta cast.
Does anybody have any idea why this should occur and how do I prevent it? I'm assuming this colour bias will show in the final print.
Anyone have any ideas ?
Thanks
David

 

[R Shaffer]

My workflow ( pretty much ) follows Bobs. I don't change the mode to multichannel. I follow Sam Wang's tutorial @ unblinkingeye.com

3CG

 

[PVia]

Same here, 'cept I invert before splitting channels...

 

[David Hatton]

Hi Guys,
Yes that's what In do but I decided to use the multichannel option and hence the question
I'll rephrase it.
Why does the colour balance change when I change from RGB to multichannel?
Any ideas?
David

 

[gmikol]

Keep in mind that when converting to either CMYK (4-color) or multi-channel (3-color), that photoshop is using the default profile assigned for that color space. For CMYK, it's typically "US Web Coated SWOP v2", used for high-volume, continuous-roll offset presses, but for Multichannel, there is typically no default profile assigned, so it appears that Photoshop uses pure synthetic CMY primaries in rendering the image, which aren't balanced or color-controlled, which is why the image colors appear to shift so dramatically.

If you stop and think about it, if PS is using pure synthetic CMY primaries in doing the multichannel conversion, then the method that David describes produces the exact same results as the method that Bob and Rob describe. I compared the negative seps visually on-screen and its true.

Hope this helps.

--Greg

 

[David Hatton]

If you stop and think about it, if PS is using pure synthetic CMY primaries in doing the multichannel conversion, then the method that David describes produces the exact same results as the method that Bob and Rob describe. I compared the negative seps visually on-screen and its true.

--Greg

But its not is it? If you convert RGB to multichannel the colour balance changes. To me this indicates that one or more of the seps has changed density. If you change back to RGB again the colour imbalance is corrected. Doesn't this mean that photoshop 'remembers' the changes its made and corrects it. How and why does this occur - and why?
I'm confused as you can tell...

 

[gmikol]

But its not is it? If you convert RGB to multichannel the colour balance changes. To me this indicates that one or more of the seps has changed density. If you change back to RGB again the colour imbalance is corrected. Doesn't this mean that photoshop 'remembers' the changes its made and corrects it. How and why does this occur - and why?
I'm confused as you can tell...

David--

I appreciate the fact that you're confused, and it seems like you're getting hung up on how PS is presenting the data to you vs. the actual data contained in the image.

First off, don't take my word for it, try this for yourself. Start with a nice, colorful image (easier to see the difference in the 3 channels).

Take 1 copy of it, keep it as RGB, invert it, and split the channels.

Take a second copy, convert to Multichannel, invert it, and split the channels.

Now compare the inverted R channel (from RGB) to the inverted cyan channel (from Multi.)
Compare G channel to magenta channel.
Compare B channel to yellow channel.
Each pair of images should be identical.


Now for understanding why the image appearance changes in Photoshop...I hope you can follow me on this.

If you've set up PS correctly, there is a working space (like AdobeRGB, ProPhoto, etc.) assigned for RGB. Because of this profile, PS understands how to translate the R,G,B color numbers stored in the file into colors on the screen.

Now, let's assume you convert this image into a CMYK color space. There's also a default profile for this color space. Now, the default profile (typically, "US Web Coated SWOP v2), has a smaller gamut in some areas than even sRGB, so when you convert from your RGB working space to the CMYK space, PS uses the profile information to map the colors in the RGB space that are outside the CMYK gamut, to colors that are in-gamut (via the different "Rendering Intent" choices). This actually changes the information in the image, and a deep, pure blue in the original may become muddy and slightly purple, for example. If you were to then convert that image back to RGB, those slightly duller, muted colors would remain, because the information in the image was actually changed (think of it as "clipping", like if you overexposed on a digital camera).

You're operating in a 3rd case. There is no default profile associated with multichannel mode. As a result, PS uses a purely mathematical transformation to get from RGB values to CMY values, but it has no profile information to tell it how those CMY values relate to actual colors, so it uses some default set of parameters, which turns out to be muddy and magenta-looking. Now, because there is no gamut boundary defined (no profile), there was no clipping involved. So no information was lost or changed. This is why converting back to RGB in this case yields the original image, because you're back to RGB values, and PS knows how to display those properly.

This last case I describe is similar, in a sense, to viewing a color-managed image in a non-color managed environment. In that environment, the image viewing program makes some assumption about the color space the image exists in. It displays the image according to that assumption, which may or may not be correct. This is why an image may appear duller or super-saturated compared to how it looks in PS. The display assumption is not correct, but the image appears different without a profile, even though no information in the image is being changed.

I sure wrote enough...I hope you find it useful.

--Greg

 

[David Hatton]

Yes, I've got it! I feel quite stupid actually now. Conversion to multichannel drops all colour management and presents the 'raw' image to the screen. Easy!
Many thanks
David