Computer-aided ink mixing tool possibilities

[nimrod]

For a while now I've been meaning to try some ink mixing based on the CMYK color model demonstrated by Limner's CMYK Color Charts. It seems silly to me to buy lots of colors of ink when you can just mix them yourself. Anyway, I was at the Ohio Pen Show yesterday and thought I'd try to pick up the inks that Limner was using. Unfortunately I only managed to find the Noodler's Shah's Rose. For the others I ended up buying Diamine Yellow and Turquoise.

 

After mixing drops of ink together for a while to see what the results would look like, it occurred to me that this manual process was a bit silly; I should be able to model this on a computer and have it tell me approximately where I should start to get any particular color I want given whatever inks I have available. I then wondered how complicated this would be considering that it should be possible to represent any given color using no more than 3 or 4 32 bit numbers. Even though "what to do" with the numbers might not be so simple, at least we're just talking about a single output color here and there are no halftone screens or optical illusions, registration problems, dpi limitations, lithography issues, etc.

 

Knowing that color management (the technical term for the calibration of imaging devices such that they all sense and display colors in the same manner) wasn't a simple problem, I knew this wouldn't be a trivial to do, but it still doesn't seem beyond possibility, so I started reading up on color theory to get a sense of how this might be accomplished for ink mixing.

 

The idea goes something like this:

1. Scan in color samples of a cyan, magenta, and yellow ink to calibrate the model with the colors you're actually using as input.
2. Tool displays the results of the scan, which will likely not match despite OS/driver level color management. The user would be able to then modify the cyan, for example, to say "it looks more like this on the monitor".
3. The tool then displays a 2-dimensional color space image (I'm assuming for now that we only show maximum possible intensity and saturation), allowing you to pick any color in that space.
4. The tool then gives you the ratio of inks needed to produce the desired color.

The printing industry already has tools like this, but they're of course extremely expensive and assume a very specific set of very exact primary ink colors, in addition to an extreme level of precision. Some of these systems even allow you to place a colorometer on a surface, push a button, and a dispensing machine will squirt out lithographic ink in the surface's color, assuming that it's within the gamut of the primary colors. But I don't think the average fountain pen ink mixer really needs this level of automation or precision. Even something like "mix 3 parts of this and 10 parts of that to get close to what you want" would be nice.

 

Some of the issues in doing this that have occurred to me:

• Scanners have different color spaces. In Windows, for example, this is compensated for by including a color profile in the scanner driver. The driver uses this profile to convert the scanner's color space to sRGB (see Wikipedia article). That may sound good but this results in a loss of information; sRGB is limited to what a typical monitor can display and may distort the actual input colors.
• The color space of monitors is much more limited than the full range of colors available in nature or the range of colors that you can see. It may not seem like it, but it is, especially with low saturation or intensity colors. Therefore you're totally unable to even select all the colors on a monitor that you could mix with real life dyes and pigments.
• Different colors of "white" light (sunlight, incandescent, fluorescent, white LEDs, etc) make colors appear differently. This causes the "same" color to look different on paper illuminated with incandescent light vs. an LCD monitor with a fluorescent backlight.
• You will never be able to get all possible colors with 3 inks. For any given set of primary colors there's a gamut which is the range of colors can can be "mixed" from those colors. A really fancy tool might be able to calculate the gamut for a larger set of input inks, for example adding a brown ink that's outside of the gamut of CMYK might allow the tool to calculate the colorspace that the new ink allows you to match.
• I'm guessing that it might also be necessary to mix one part of each primary with one part of the other two and scan that in to calibrate the output of mixing the actual inks.
• From a little messing around it sort of looks like there's a nonlinear relationship between the ratio of inks used and the resulting color, or at least my perception of it. I wonder if this is true and if it is whether it differs from ink to ink.

To give you some idea of what a gamut is, here's a diagram of the sRGB color space plotted on top of a "lab color" color space (closer to what colors actually exist in nature) from the wikipedia article on sRGB:

 

http://upload.wikimedia.org/wikipedia/commons/thumb/a/a8/CIExy1931_sRGB.png/325px-CIExy1931_sRGB.png

Wikipedia caption: "CIE 1931 xy chromaticity diagram showing the gamut of the sRGB color space and location of the primaries. The D65 white point is shown in the center. Note that areas outside the triangle cannot be accurately colored, because they are out of the gamut of computer displays."

 

So each vertex on that triangle is a primary color (red, green, and blue in the case of additive color mixing used on video displays). For printers, different video displays, and other media try to imagine a different triangle which will have regions outside of the sRGB triangle shown above. Also try to imagine that "D65" (white) point being skewed based on the type of lighting in the area.

 

Anyway, I'm wondering what people with hopefully more knowledge on the subject think of the idea and if there were such a tool if anyone would use it. I also wonder if such a tool might even drive ink sales, for example if some company came out with a special set of CMY inks designed for mixing. People might pay some money for a set of CMYK inks and a tool to help you mix them. On the other hand it might lessen peoples' willingness to buy one-off colors of ink, though there will always be out-of-gamut colors that wouldn't be easily obtainable via CMYK mixing, and there will always be plenty of people who don't want to mess around with mixing ink.

 

Here are some of the pages I've been reading (for those interested):

 

http://en.wikipedia.org/wiki/CMYK_color_model

http://en.wikipedia.org/wiki/Gamut

http://www.brucelindbloom.com/

http://en.wikipedia.org/wiki/Color_management

http://en.wikipedia.org/wiki/SRGB

http://en.wikipedia.org/wiki/Lab_color_space

http://en.wikipedia.org/wiki/Absolute_color_space

http://en.wikipedia.org/wiki/Adobe_RGB_color_space

http://www.cambridgeincolour.com/tutorials...-perception.htm

http://www.microsoft.com/whdc/device/displ...olor/icmwp.mspx

 

[psfred]

I suspect you have the cart in front of the horse here -- first you need a good absorbance spectrum for each of the inks, then mix in several ratios to get absorbance spectra of the color ranges.

 

From this data you could determine the color space and derive from that model the dye ratios.

 

You must remember that you cannot produce all colors from a set of three fairly pure CMY dyes, they will not reproduce the absorbance spectra of similar colored dyes of different structure. Do some looking around at photographic color reproduction for some hints. You can probably safely ignore anything about UV sensitivity of silver halides distorting color, though as you are not making dyes in situ from exposed silver halide granules.

 

I would also suggest doing reflectance spectra on defined paper as well, as absorbance spectra may not be identical to reflectance spectra.

 

You are quite correct that this is NOT simple.

 

Peter

[dandle]

You're trying to reinvent the (colour) wheel...

 

Good luck with your endeavor.

[Limner]

Tools of this type do indeed exist for other industries. Here is one that was created for use with fabric dyes. It is not extremely expensive; it is free to use online. Try:

Procion MX, turquoise MX-G

Procion MX, red MX-5B

Procion MX, yellow MX-8G

Procion MX, black MX-2R

 

for your CMYK in the color selectors. Follow the instructions; slide the little sliders back and forth to your heart's content.

 

Tools of this type are useful, though I feel they are certainly no substitute for acquiring an understanding of color theory or genuine hands-on color mixing experience--no matter how "silly" such activities may seem.

Edited November 5, 2007 by Limner

[WillAdams]

If you have PhotoShop w/ multi-channel support, you can then ``mix'' ink swatches in it --- just get the colour values for the inks in advance. FreeHand allowed one to redefine how individual colourants in CMYK were defined which allowed one to then mix inks on-screen.

 

There was a DTP tool, Cerilica's Truism which allowed one to do this as well, but it's disappeared.

 

William

 

 

[nimrod]

I suspect you have the cart in front of the horse here -- first you need a good absorbance spectrum for each of the inks, then mix in several ratios to get absorbance spectra of the color ranges.

That's probably because I understand carts and not horses, and am not familiar with carts that require horses, which is why I posted this looking for people who do. Basically I know computers but not chemistry or physics. Anyway it sounds like I or someone would need a photospectrometer, the cheapest of which is more than I'd be willing to spend on something like this.

 

You're trying to reinvent the (colour) wheel...

Well, not really. I just want a free/cheap version of an existing computerized wheel. What I'm thinking of is more like this:

http://www.pantone.com/pages/products/prod...mp;ca=2&s=4

Except free, software-based, and calibrated to output formulations in terms of arbitrary inks rather than pantone hexachrome primaries or something similarly special. Unfortunately it sounds like a flatbed scanner wouldn't suffice as a sensor though.

 

Tools of this type do indeed exist for other industries. Here is one that was created for use with fabric dyes.

Awesome! This is exactly what I was thinking of except that it isn't calibrated for fountain pen inks. In particular there's no way to get a color that dark just by using "lots" of turquoise for example. Also I don't understand the units that it uses. To get a similar shade of green to what I got by mixing one part yellow and one part turquoise I have to set yellow to "10" and turquoise to "1". To get something similar to what I got with 3 yellow and 1 turquoise I have to set it to 0.5 turquoise and 10 yellow. So I don't really know what's going on here or how I'd convert this tool's units to mix ratios.

 

For those who were wondering, here are examples of the crazy expensive professional color matching ink dispensing/mixing tools that I was talking about. Actually the photospectrometers aren't necessarily super expensive but the automated ink dispensers are

http://www.pantone.com/pages/products/prod...mp;ca=2&s=2

used with something like:

http://www.vale-tech.com/product.php?id=aquablend

 

BTW, I didn't know you could get color calibration stuff for your monitor this cheaply now:

http://www.pantone.com/pages/products/prod...mp;ca=2&s=0

Not bad for $89 but unfortunately it only measures emission sources (like monitors rather than light reflected off of paper).

[nimrod]

If you have PhotoShop w/ multi-channel support, you can then ``mix'' ink swatches in it --- just get the colour values for the inks in advance. FreeHand allowed one to redefine how individual colourants in CMYK were defined which allowed one to then mix inks on-screen.

 

It seems like CMYK with halftone screens isn't equivalent to actually mixing inks together though. I'm not sure if this would work. (Unfortunately I don't have photoshop so I can't try it myself.)

[blak000]

I have a few questions concerning your idea. I don't have any experience with mixing inks, so if my reasoning doesn't really stand, feel free to correct me.

 

First of all: isn't is possible that various "side chemicals" might interact between inks, potentially altering the desired light reflectance of the ink and, consequently, color? Inks have a few different solutes in there, and certain compounds might do a chemical double displacement, decomposition, etc.... creating a new compound with different physical properties. Richard Binder stated at one time that he saw certain inks turn into veritable sludge when mixed together. If that's the case, then it's fair to assume that inks can have other unwanted side effects as well. Also, I feel that your color wheel idea would have to be limited within a designated range of "safe" mixing inks. This could only be deduced by manual testing...

 

Secondly: even though an ink looks a certain color to the human eye, it doesn't really give us a clue as to the level of saturation or types of colors that make up the ink. Let's say you have ink A, which is a certain shade of yellow... then you have ink B, which is a blue-green. Ink A was made with a low pigment concentration of yellow ink. Ink B, on the other hand, was made with a highly saturated level of both blue and green pigments. Even though we might expect a color in-between yellow and blue-green... mixing the two might actually leave you with something that's still looks heavily like Ink B. Even worse, what if ink B was made with a single, stock pigment of blue-green? The manufacturer could add tons of that pigment, heavily saturating the ink, without changing the ink's color. How would that affect the outcome of the mixture? This problem would have to be solved largely through manual testing (again). So to make this color wheel, you would have to list the ink color, brand, and specific measurement needed to attain the desired color on the wheel. This could be largely dealt with by limiting yourself to 3-5 primary inks (within the same brands), but this would limit the range of colors you could make, as well.

 

Thirdly: if you intend to make this program for widespread use, I see it running into problems. Not everyone's monitor displays colors in the same way. A plain blue on your screen might look a few shades lighter on someone else's. If that person then mixes inks, expecting to get the exact shade he/she sees... well, that person might get a little surprise. Professional artists and designers don't run into this problem as much, because they often take steps to calibrate their monitors (they sell devices for such purposes that often cost several hundred dollars).

 

Just some thoughts I had. Aside from those issues, your idea sounds great. I'm very interested as to how this will turn out.

Edited November 5, 2007 by blak000

[encephalartos]

I have a few questions concerning your idea. I don't have any experience with mixing inks, so if my reasoning doesn't really stand, feel free to correct me.

 

First of all: isn't is possible that various "side chemicals" might interact between inks, ...

 

Secondly: even though an ink looks a certain color to the human eye, it doesn't really give us a clue as to the level of saturation or types of colors that make up the ink. Let's say you have ink A, which is a certain shade of yellow... then you have ink B, which is a blue-green. Ink A was made with a low pigment concentration of yellow ink. Ink B, on the other hand, was made with a highly saturated level of both blue and green pigments. Even though we might expect a color in-between yellow and blue-green... mixing the two might actually leave you with something that's still looks heavily like Ink B. Even worse, what if ink B was made with a single, stock pigment of blue-green? The manufacturer could add tons of that pigment, heavily saturating the ink, without changing the ink's color. How would that affect the outcome of the mixture? This problem would have to be solved largely through manual testing (again). So to make this color wheel, you would have to list the ink color, brand, and specific measurement needed to attain the desired color on the wheel. This could be largely dealt with by limiting yourself to 3-5 primary inks (within the same brands), but this would limit the range of colors you could make, as well.

 

Thirdly: if you intend to make this program for widespread use, I see it running into problems. Not everyone's monitor displays colors in the same way. ...

 

Just some thoughts I had. Aside from those issues, your idea sounds great. I'm very interested as to how this will turn out.

 

I really strongly agree witht he second paragraph. This is based on experience mixing paints. such as watercolors or acrylics. Not all pigments

or dyes have equal strength in mixes. If you are dealing with artist's quality paints, then most brands actually disclose what pigments or dyes

are used in the color, and how permanent the vendor says the color is (lightfastness & other deterioration).

 

With fountain pen ink, there isn't any such disclosure of what the ingredients were, so you don't know what dyes they used to mix up a black, or example.

You'd need to try the paper towel chromatography to see what went into the ink color. Even with that known, different manufacturer's paints behave differently. WIth paint, the relative strength of colors in mixes is something you learn by experimentation, or can find in books. Manual experimentation is not such a bad thing.

 

Examples of how different paint colors behave are Prussian Blue, which is a green blue that is extremely strong in mixes. If you were using it to mix a green, you would add TINY amounts of the blue to the yellow or yellow brown you were using. Terre Verte (green earth), on the other hand, is notoriously weak in mixes.

 

So, I just mix inks with manual testing. The latest unexpected thing I found was mixing Noodlers Rachmaninoff bullet proof pink with some

Noodlers Bulletproof black to get a dark purple. It turns out the pink is much stronger than expected. I started with the pink, and keep adding

more black. So far, its a brownish burgundy that needs more black or blue to be dark purple. I got this idea from seeing the primary components

of MB Violett: magenta, and black, when removing a stain...

Edited November 6, 2007 by encephalartos

[dandle]

Certainly an ambitious project for a fellow who is unwilling to mix more than a few droplets of ink together by hand...but what a fabulous way to take all the fun out of mucking about with colours, and the happy accidents that creates...! viva your digital revolution!

 

 

 

 

[captnemo]

If I'm not mistaken, I think the equations also require fudge factors and there might be some lookup tables and interpolation involved as well to account for human visual perception, which is not "flat". Of course, that's no biggie in a computer program, once you have the info you need.

[blak000]

I really strongly agree witht he second paragraph. This is based on experience mixing paints. such as watercolors or acrylics. Not all pigments

or dyes have equal strength in mixes. If you are dealing with artist's quality paints, then most brands actually disclose what pigments or dyes

are used in the color, and how permanent the vendor says the color is (lightfastness & other deterioration).

 

With fountain pen ink, there isn't any such disclosure of what the ingredients were, so you don't know what dyes they used to mix up a black, or example.

You'd need to try the paper towel chromatography to see what went into the ink color. Even with that known, different manufacturer's paints behave differently. WIth paint, the relative strength of colors in mixes is something you learn by experimentation, or can find in books. Manual experimentation is not such a bad thing.

 

Examples of how different paint colors behave are Prussian Blue, which is a green blue that is extremely strong in mixes. If you were using it to mix a green, you would add TINY amounts of the blue to the yellow or yellow brown you were using. Terre Verte (green earth), on the other hand, is notoriously weak in mixes.

 

So, I just mix inks with manual testing. The latest unexpected thing I found was mixing Noodlers Rachmaninoff bullet proof pink with some

Noodlers Bulletproof black to get a dark purple. It turns out the pink is much stronger than expected. I started with the pink, and keep adding

more black. So far, its a brownish burgundy that needs more black or blue to be dark purple. I got this idea from seeing the primary components

of MB Violett: magenta, and black, when removing a stain...

 

Umm... I don't understand what the disagreement is over. Seems like you raised the exact same point that I did. :hmm1:

Like me, you feel that the ink might be made up of variable pigments types, that might lead to some strange, unexpected color. You also suggest that manual testing might be the best way to do this. In fact, you went on to provide one concrete example of this anomaly occurring: with the Noodler's Rachmanioff and Noodler's Bulletproof Black.

 

I'm not trying to bash on you in any way... just a bit confused. Perhaps you could point out where I may have misinterpreted your post? As far as I can see, this is exactly one of the things I was talking about.

[nimrod]

I've read about the sludge issue with mixing "incompatible" inks, but from what little I've heard this is usually obvious right away rather than something that occurs slowly after you put the ink in a pen. Of course this is just what I've heard and I don't know how much experience it's based on but it has to be based on more experience than I have

 

Also I had guessed that certain inks/dyes were probably "stronger" than others which is what I was trying to say in the following extremely cryptic sentence: "From a little messing around it sort of looks like there's a nonlinear relationship between the ratio of inks used and the resulting color, or at least my perception of it. I wonder if this is true and if it is whether it differs from ink to ink."

 

So from what you guys have said it sounds like trying to predict the outcome of mixing any two totally arbitrary inks (even ones that you have good absorbtion spectrum data on) isn't possible. Any digital tool would probably need to be limited to trying to predict the outcome of mixing certain known "well behaved" and well understood inks that have been carefully studied by someone with a photospectrometer.

 

Regarding monitor color calibration issues: I mentioned that in the original post. Basically the idea in most operating systems is that you have "color profiles" for input devices (scanners for example) that convert the input devices color space to a standard device-independent color space called "sRGB". You then have things like monitor drivers and printer drivers that contain profiles on how to convert sRGB to the color spaces of the monitor or whatever. Of course this doesn't entirely work because there are different methods of converting between color spaces. Basically the out of gamut colors can either be "clipped" or reinterpolated to spread them over the new range of possible colors. It's also possible to either adjust or not adjust for the white point of each color space.

 

Anyway, this is really more of an issue when going from camera -> sRGB -> monitor, and camera -> sRGB -> printer because you have several places to lose information and pick up distortion: camera -> sRGB conversion, sRGB -> monitor conversion, monitor -> sRGB "conversion" (meaning someone adjusts the image to look right on their monitor which for various reasons might not be representing sRGB accurately), and finally sRGB -> printer. At each of those boundaries you're converting from one color space to another, and you will drop or distort information at each one even if everything is calibrated exactly.

 

In this case what we'd be doing is going from monitor <- sRGB -> ink formulation. Any tool like this would have to assume that the monitor is accurately calibrated to represent sRGB, which not-by-coincidence just happens to match the color gamut of most monitors. The fact that Pantone sells a device with software for $90 that does nothing but calibrate your monitor to properly represent sRGB is a testament to the fact that most monitors aren't actually correctly representing sRGB for whatever reason, be that due to incandescent light in the room or peoples' crazy brightness/contrast settings.

 

It might be possible to have the user calibrate the monitor -> sRGB conversion, for the tool's purposes at least, by adjusting some sample cyan, magenta, and yellow swatches in the tool to match what they see when they use a brush or q-tip to apply their unmixed inks to white paper, essentially saying "the cyan looks like this on my monitor" etc.

 

And of course if someone is REALLY serious they can spend $90 on the gadget to precisely calibrate their monitor's color profile so that it represents sRGB exactly, but the simple answer is that this problem isn't as difficult as color management for digital photographs or scans of ink swatches because there are fewer places where color space conversions need to happen. Since sRGB <-> monitor conversion shouldn't be that lossy, the only really difficult one should be sRGB -> ink formulation.

 

However I wasn't really going for something as exact as pantone. People using this would have to accept some error and be willing to adjust, it's just a question of how close you can get without fancy equipment. I don't think the sRGB -> monitor issue is necessarily as bas as you think, and most likely the variation introduced by different papers, pens of different wetnesses, etc are worse than the sRGB -> monitor distortions. We're talking about fountain pen ink here and not carefully calibrated offset presses

 

 

So anyway, I admit that I don't really know much about this stuff, especially the chemistry and physics aspects. That's why I posted the message, looking for other people who do to comment. I also know the whole idea isn't entirely crazy because Pantone and others have similar software for formulating lithographic ink and paint for your house, though it sounds like trying to do this for any random ink that someone has laying around is crazy even if they do have a photospectrometer sitting around. In fact the inks needed might have to be specially created for the purpose. Then again some guy did create a tool for predicting the outcome of mixing dyes (as Limner pointed out, thanks!) that weren't specially created for use with a software tool.

 

The guy with the dye mixing tool says he'll add data to his tool if you mail him dye samples, so I might mail him some ink and see what his tool does with the data. Based on that I might invest more effort but for the time being it sounds like just experimental ink mixing is probably more practical than trying to use a mathematical model, especially if you don't want to limit your mixing to a really well understood set of 4 inks.

 

On a related note: I'm afraid I might have offended some people by saying that I felt that what I was doing was "silly" when I was mixing drops of ink to see what would happen. I apologize for that, I didn't really mean that literally in fact I now have a new appreciation for just how difficult it is to predict the outcome of ink mixing. (At first I thought it might just require some simple linear interpolation of RGB values but that's obviously not the case at this point

Edited November 6, 2007 by nimrod

[WillAdams]

nimrod,

 

In FreeHand, one can control how each colourant in CMYK is displayed on-screen.

 

This means that one can say, make C look like PANTONE 301, Magenta look like PANTONE 188C, Yellow look like PANTONE 7510C and blacK look like PANTONE 431C, and get a decent representation of how two or more of the inks look when overlaid, incl. when they're tinted.

 

It'd be just as easy to make the inks look like arbitrary fountain pen ink colours and then do builds of 100% of up to four colours to get a rough approximation of what they'd look like when blended --- as others have noted, the interplay of different dies will affect this, but it's a workable representation in most instances.

 

Unfortunately, it's a global application setting, so one needs to make notes about how a file should be changed if one needs to revisit it.

 

William

(who used to set up jobs like that when working at a flexography shop)

[Huffward]

This might seem a banal point but, sure, you can use CYMK to define the actual hue of ink in the bottle (though not its intensity I suspect - though I'm on shaky ground here).

 

But the actual colour on the page depends to a high degree on the pen feed, pen nib, and the paper (colour, surface, absorbancy etc.).

 

I can't see how there can be any standard or mathematical way of saying: "if you use this ink, your writing line will look exactly like THIS". And if you can't do that, there seems little point to the exercise (rather like the perfect egg timer that always hard-boils your egg).

 

Let's not calculate all the fun out of that eternal Holy-Grail quest for the perfect ink.

Edited November 6, 2007 by Huffward

[blak000]

But the actual colour on the page depends to a high degree on the pen feed, pen nib, and the paper (colour, surface, absorbancy etc.).

 

I can't see how there can be any standard or mathematical way of saying: "if you use this ink, your writing line will look exactly like THIS".

 

This is a very good point Huffward. Still, maybe Nimrod could settle for a program that can predict a certain color for when the ink is still in the bottle? Even something simply as dry vs. wet writer would create a different color. My Lamy Studio tends to put down a plain looking blue with PR American Blue, but my Lamy 2000 (a wet writer) lays down a much more intense, brighter blue with the same ink. I feel, it would get too complicated if he had to try and weigh all the other variables.

 

Like Willadams states, disregarding those instances where the ink might react funny, the program could be made to at least give a good approximation. I think that's the best you could really hope for in this case. It would take a significant amount of extra work to gain a slight-to-medium improvement in accuracy. I guess, Nimrod, it depends on if that improvement is worth it to you.

 

I think you could still make a pretty good program if you limited it to a single label? I think it might be relatively safe to assume that mixing ONLY within Waterman, or ONLY within Pelikan, wouldn't cause too much of a problem. Personally, I'd go with Waterman if I were you. You get a good stock of base colors, and Waterman has a reputation as being a rather free-flowing, well-behaved ink. You want to center the program around a set of inks that's applicable to a wide range of pens.

 

Who knows, later on you might add in new inks that you've deemed safe in an upgraded version. I think once you've established the base colors, you could just do a simply test run with each new candidate, calibrate the wheel to match the resulting hue, and add in the proper color code. This is, by no means, a small endeavor, but I could see it as becoming a staple prog within the fountain pen community.

 

Update: Just another thing that crossed my mind. You would also have to take in account the type of colors that would appear with varying ratios? As we discussed earlier, different saturation level/pigment content could alter the outcome. Many people like to mix 1:10 or 1:5 ratios to get a desired mix. Of course, if we're sticking with rough approximations, this may not be too necessary.

Edited November 6, 2007 by blak000

[nimrod]

In FreeHand, one can control how each colourant in CMYK is displayed on-screen.

 

That would be fun to play with except that things like FreeHand and Illustrator cost like $500+

[nimrod]

But the actual colour on the page depends to a high degree on the pen feed, pen nib, and the paper (colour, surface, absorbancy etc.).

 

I can't see how there can be any standard or mathematical way of saying: "if you use this ink, your writing line will look exactly like THIS". And if you can't do that, there seems little point to the exercise (rather like the perfect egg timer that always hard-boils your egg).

 

Let's not calculate all the fun out of that eternal Holy-Grail quest for the perfect ink.

 

Yea that's why I was saying that I'm not sure how much monitor calibration should matter. I can get a pretty large range of darkness/lightness just using a q-tip.

 

The sort of tool that I'm thinking of wouldn't really replace manual ink mixing experiments but might help people get close to the hue they want and save a little bit of time. Call me a nerd, but to me thinking about ideas like this IS fun.

[nimrod]

This is a very good point Huffward. Still, maybe Nimrod could settle for a program that can predict a certain color for when the ink is still in the bottle? Even something simply as dry vs. wet writer would create a different color. My Lamy Studio tends to put down a plain looking blue with PR American Blue, but my Lamy 2000 (a wet writer) lays down a much more intense, brighter blue with the same ink. I feel, it would get too complicated if he had to try and weigh all the other variables.

Color in the bottle probably wouldn't be useful. It would probably have to be based on the color of the ink on paper that has been dipped in the ink, to the point where it can't absorb any more. I think someone mentioned using toilet or tissue paper with a photospectrometer probably for this reason. I don't know if toilet paper is really a good standard since no real paper that someone could write on is likely to be THAT absorbent. Maybe a little strip of HP 32# paper soaked in ink and allowed to dry would be good. Anyway there has to be some good material that could be used for this that basically shows the darkest something can ever get.

 

It might also be possible for someone to tell the program "here's the shade of lightness that my pen writes with this ink on my paper" to help it predict what range of lightness in other colors might realistically be seen (rather than what it looks like with the paper totally soaked).

 

I think you could still make a pretty good program if you limited it to a single label? I think it might be relatively safe to assume that mixing ONLY within Waterman, or ONLY within Pelikan, wouldn't cause too much of a problem. Personally, I'd go with Waterman if I were you. You get a good stock of base colors, and Waterman has a reputation as being a rather free-flowing, well-behaved ink. You want to center the program around a set of inks that's applicable to a wide range of pens.

Yea, basically I'd have to pick some inks to start with. Unfortunately I don't have a photospectrometer to play with though so this may keep me from ever getting anywhere. I might try mixing up some swatches diluted with different amounts of water and mixed in different ratios, then measure the RGB values with a scanner and try to create some sort of function that can interpolate between them. Not sure if this would work though.

 

One question in my mind that goes beyond just this tool is what are the best CMYK inks to use with or without some tool? Recently I took a piece of paper and painted some swatches of color ink jet refill ink on it with q-tips, then painted swatches of some similar fountain pen inks for comparison. I need to scan this in and upload it for people to look at.

 

Basically my assumption is that ink jet refill inks are ideal CMYK inks (or as ideal as technology currently allows for CMYK) since they're made specifically for a CMYK process in ink jet printers. Theoretically then whatever fountain pen inks come closest to the hue and saturation of these ink jet inks ought to be best, but as people have pointed out you might have two fountain pen inks that look close to magenta on paper, but when you dilute them 50:50 with water they become radically different shades because one has more "strength" (my term, not sure if "dye concentration" is a better one) than the other.

 

So I'm not sure what kind of experiments are really necessary to determine which inks are best for CMYK mixing, with or without a tool. Limner was using Noodlers Navajo Turquoise, Shah's Rose, and Yellow. Some limited testing that I did on the paper that I need to scan in shows that:

 

Navajo Turquoise is significantly more saturated than Diamine Turquoise and to me looks really close to ink jet cyan. I suspect that Navajo Turquoise is the best cyan.

Shah's Rose seems really close in hue but is not quite as saturated as ink jet magenta. I wonder if it's possible to do better.

Noodler's Widowmaker is about as saturated as ink jet magenta, but is a little too red.

Diamine Yellow and Noodler's Yellow are about the same in saturation and not that close to ink jet yellow, which actually looks kind of orangish yellow.

Sailor Yellow is more orangish (or maybe just more saturated yellow??) and is closer but still not there.

I think that Noodler's Apache Sunset might actually be closest to ink jet yellow but I don't have a sample to test right now.

 

And of course ink jet CMYK inks are unlikely to be equal. CMYK processes may not be that standard between manufacturers and the refill inks I am using for my controls could be (bleep) for all I know.

 

For the test I used 32# HP laser paper, q-tips, and I rubbed the q-tips back and forth until the colors wouldn't get any darker.

 

Update: Just another thing that crossed my mind. You would also have to take in account the type of colors that would appear with varying ratios? As we discussed earlier, different saturation level/pigment content could alter the outcome. Many people like to mix 1:10 or 1:5 ratios to get a desired mix. Of course, if we're sticking with rough approximations, this may not be too necessary.

 

If the tool can't tell you what ratios of ink to mix to get close to a specified color then it essentially doesn't do anything. Calculating the ratio of inks to mix is the whole point of the tool.

[WillAdams]

Well, FreeHand _has_ been around for a _lot_ of versions (11), and it's had this feature since at least FHv8 (which I just picked up a copy of for Windows for $1.99 from the local Salvation Army). It shows up at reasonably low prices at used software sites and places like Amazon's marketplace quite often as well.

 

You can download a demo of FreeHand MX to try it out.

 

It's well worth whatever you pay for it, even full price for the latest version, even though not up-dated to Unicode and OpenType &c.

 

Best. (vector) Drawing. Program. Ever.

 

William