Adobe Photoshop Training with Dan Margulis

I have an unusual problem and I'm wondering if color management may provide a solution.

My company sells a custom-made device based on inkjet technology (HP Deskjet 695 mechanisms printing through a Windows 98 or XP frontend) that prints images onto edible items. Because of this special application, we obviously must use food colorings rather than ink.

Our dilemma is that the food colorings are not as vibrant as the inks that the printing mechanism and drivers assume are being used, so the colors come out drab and off-color. This is particularly apparent in the reds due to the fact that regulatory prohibitions on the use of Red Dye #2 prevent the food coloring manufacturers from producing a strong magenta food coloring to match the magenta ink.

A typical bright cherry red color scanned into a digital image will have approximately 90% Magenta and 80-90% Yellow. Because the food coloring magenta is not vibrant (and somewhat brownish) the result of a 90M/80Y color formulation is a brownish orange rather than a bright cherry red. Flesh tones are a jaundiced yellow color, etc. All in all, not very natural or pleasing.

I have attempted to compensate for this "bad red" problem by slightly weakening the strength of the Yellow we use, with some limited success, but I am curious to know if anyone thinks that there may be some possibilities for controlling this problem by altering color printer driver profiles. If so, do you have any suggestions about how we should proceed? I know very little about this field, but my intuition tells me that this may be an area where we could greatly improve the quality of our image reproduction.

Thank you for any advice or suggestions.

Scott Smith

___________________________________________________________________________

I've been awaiting a reply as you have.

The situation sounds similar to what we faced here when we were printing temporary tattoos (on an old single color s-line Heidelberg). There the "inks" were FDA-approved and the colors were nowhere like "regular" cmyk colors were. for Magenta had a lot of yellow in it, black was a greenish brown, cyan was more blueish, and so on.

We just sort of soldiered away and gradually devised a way to get something that looked ok...often we were faced with handling logo colors and other things that were a bit sensitive.

what we decided was critical was the rgb-cmyk conversion (perhaps a big Duh! from others on the list) which meant that we had to instruct the people doing the conversion to do it differently from the way they'd been "taught", including reducing yellow, making sure that black was a 3-or 4-color black, etc. Once we had this figured out it usually worked ok.

Mac Townsend

Adcom Graphics, Digital Imaging

Fairfield, California

www.adcomgraphics.com

___________________________________________________________________________

From: Dan Margulis

Date: Monday, 20 Oct 2003 23:28 EDT

Subject: Printing with food colorings

Scott writes,

If that's so, then the mistake is trying to match the magenta at all.

The reason we use magenta, cyan, and yellow inks as opposed to three other random colors is that these are the direct opponents of red, green, and blue. Therefore, no other combination of inks of different colors can possibly produce these three: if we want pure magenta, we have no choice but to use a magenta ink, whereas a deep purple or a purplish red could be constructed in many different ways.

In this particular case, you already know that you can't get a good magenta color at all. Therefore, there's no particular reason to limit yourself to a method that only works in theory if the magenta ink is good. If you insist on trying to work with an inferior magenta ink then you can't get a good red, as you found out, except by trying to eviscerate the yellow, which is a bad idea.

The solution is to admit that your magentas are going to be lousy and concentrate on not having your reds be that way as well. You should go for a much redder ink and for a bluer cyan. This way, you'll be able to hit most blues and reds fairly well at the expense of magentas and cyans. Blues and reds are far more important colors.

If you're familiar with the Ink Setups in Photoshop's Custom CMYK, you can make a setting that will compensate for this reasonably well. Or, you can have somebody make a profile for you.

Dan Margulis

___________________________________________________________________________

I want to see this - measuring umpteen patches printed on birthday cake icing. And you'll need a profile for each substrate, um , flavor.

Henry Davis

___________________________________________________________________________

I was wondering about that. Is there an Eye-One or DTP41 model for soft and squishy substrates?

Mac Townsend

Adcom Graphics, Digital Imaging

Fairfield, California

www.adcomgraphics.com

___________________________________________________________________________

Mac,

If there is an easy solution, I would like to hear about it.

My guess is that if you really want to measure cake icing it would cost you a bundle. You would probably have to get an integrating sphere and do one or a few colors at a time to get the LAB measurements. Then feed those measurement into a profiling application. Now if you did a thousand cakes a year then there might be business model to consider.

The only good news in this scenario would be licking the spoon afterwards.

Jim Rich

___________________________________________________________________________

I didn't post this question (but I did reply that I've had somewhat similar problems with temporary tattoo inks--which are not really cmyk either)

The guy who did post it (apparently) sells a modified Epson printer that bakeries can use to decorate cakes and such. AFAIK it prints not directly onto the cake with frosting but but with coloring onto an edible (?) substrate that is then slapped (or carefully placed) onto the cake. I've seen the process referred to before, and even thought about bidding on ebay for one of these things a few months ago (always looking for output options!).

I suspect that a solution to this would have important economic advantage to the company selling the equipment. One cannot expect a baker to be versed (or interested) in color and whi red doesn't come out red. But if the company was able to bundle a profile that got it closer to what was in the photo....(and photos are likely a major draw for this package since other stuff can be quickly handled with a tube of frosting.)

Mac Townsend

Adcom Graphics, Digital Imaging

Fairfield, California

www.adcomgraphics.com

___________________________________________________________________________

I once collaborated on a Siggraph presentation on the new Raspberry, Lime, and Blueberry color space, though this was in connection with ray tracing Jello-brand gelatin deserts.

http://www-2.cs.cmu.edu/~ph/jello/jello.html

Considering the secondary role of quality for edible art is not that high, a digital image alongside a Macbeth chart should get in very close to the Lab values required for setting up a profile or set of curves.

Matter of fact, a Macbeth color checker cake, (or tatoo), might be a worthy product and of itself. Everyone will want one. Remember, - you heard it here first!

Mike Russell

http://www.curvemeister.com

http://www.zocalo.net/~mgr

http://geigy.2y.net

___________________________________________________________________________

Mac writes,

Good grief. We're not talking about waterless printing at a 600-line screen here, but about printing where nobody is expecting much quality and where the current process is completely hosed.

If we just replace the brownish magenta ink with something redder, that should make the printing better even if nobody compensates for the hue change. Then, all we have to do is change the M definition and the CM definition and the MY definition within Custom Color to something redder and it will be better yet. Anyone whose eye can distinguish red from magenta should be able to define a redder ink even if they don't know what LAB or xyY value they're looking for.

It's, er, a piece of cake.

Dan Margulis