Color Matching

Discussion in 'Gallery & Designs' started by Gil, Dec 5, 2004.

  1. Gil

    Gil Active Member

    Hello All,

    I've been investigating color matching as it applies to card modeling. To begin with this is one of those areas which can be summed up as "one big can of worms" being safer to stay away from as it still isn't understood that well.

    The reason for the investigation is to find an acceptable method to easily find an acrylic color mix which matchs a color output by "your" printer. There are professional systems which allow the color gamut to be califbrated but this is far to complex and costly when it comes to the average modeler. The real match in colors is a comparison between a sample of the building material evenly coated with the acrylic color and a print out of a Federal Standard 595 color. What about screen color? Nice, but it can only be used as a approximation of the printed output. The real test comes in comparing that which is painted to that which is printed.

    I'm looking for input..., ?

  2. Ashrunner

    Ashrunner Member


    I too have found a can of worms in regards to colors for models. It really hit home when I looked around on the net for sample of the three main colors used for the jungle camo scheme of the B-52D. I had the FS numbers of the colors, but when I searched for samples of the colors to use, I found several different variations of what should have been a standard color. That is one of the problems in coloring models. I figure the best way for me to do it would be to find a standard I feel is close the the colors I remember and use that set of color "chips" as my standard. So I found something I like and will stay with it. It's the colors on the 595b 1, 2 and 3 pages found here...

    Also, a while back when I got deeply involved in digital photography, I began reading about color management. I realized I needed a way to maintain a constant color profile throughout my photography workflow process. After a lot of reading and finding out I couldn't afford the required hardware for creating my own color profiles, I have settled for premade ICC profiles. For my digital camera, I have it set for the Adobe RGB (1998) profile. My monitor is set to use the same profile, along with the programs I use most for my photography work. Throughout my entire workflow for digital photography, the Adobe RGB (1998) is the ICC profile used. I used that profile since Photoshop has the adjustment program Adobe Gamma which I used to "soft" calibrate my monitor. Logically, Adobe would at the same time, have it set to calibrate for the RGB(1998) standard.

    When I began recoloring my models, I decided I would set my graphic programs that I use for that, to the Adobe RGB (1998) also. That way, I would have a consistant color standard in everything from monitor to program to printer, since the printer can also be set for an ICC profile. I have also found I can set my scanner to use an ICC profile for when it scans in color and have set it to use the same profile. Everything which uses color on my entire system, uses the Adobe standard.

    Since I began doing this, my digital photography is very standardized, and when I print an image, the colors I see in the print, are as close to those I saw, or see on the monitor, as I can tell. Since it works for photography, it also works for colorizing card models. I know when I use the FS 34159 color, or any other color, on a model, it will look the same in the print as it did on my monitor when I used it to fill a certain area of the model I am repainting.

    When the day comes I can afford to purchase a color management profiling system (good ones run around $1500 US), I will and I will then profile my entire system to obtain as accurate color management as possible. 8v) Til then, I'll stick with what I know currently works for me...the Adobe RGB (1998) profile.
  3. Bowdenja

    Bowdenja Active Member

    Hey Guys,

    Being new to this, I have noticed that it is very hard to match colors in aircraft profiles; both on screen and in printed copy. I have tried several printers and there is a big difference in printed copy between them. Everybody's ink is just a little bit different and every brand printer mixes ink differently.
    Can-o-worms or not this just might have to be something we have to live with.

    I'm currently trying to re-color Maty Modelarz MiG 17 and create a VPAF MiG from the Viet Nam War and it is very frustrating not being able to match the profile.

    Anyway maybe the only way around this is trial and error. And I think that when I do match who is to say that someone else's system will print like mine. I have noticed that some of the models I've built that have been featured in building threads are not printed exactly the same color as I have printed.

    Maybe we could do a test build of the same model and everyone post a picture and see how the color(s) vary from one system to another?
  4. Gil

    Gil Active Member

    Very interesting replies.

    I use Adobe Illustrator and Photoshop and will setup to their 1998 standard. That's a good start. Thanks Ashrunner. It's also a good reason to use the print manufacturers ink cartridges. One of the reasons for buying the i560 was it's low ink useage. I've been using Rhinotek ink cartridges and I really can't tell any difference between Canon's and their cartridges. Might be dumb luck also.

    Back to what I was thinking about though, how to match a paint to the printed output. Let's assume that a print color output is acceptable. A sample of this is then scanned along with an acrylic color painted on a piece of the same cardstock. The two color values are then compared using a paint package which will yield R,G,B difference values. Using a package like Liquitex Pixel2Paint allows evaluation of the color and the ability to select a color value which matches or nearly matches the value of the scanned print output. This program provides the "mix" values to make the new color from those supplied by Liquitex (other brand named acrylics that are less expensive can be used). The same process is repeated and is used to "dial" in the paint mix.

    Haven't tried this yet but it is closed loop so it should work. Once the color mix is established it can be used as a standard for a particular brand of acrylic paint allowing others to come fairly close to the required color without having to go through all the repetitions of the above process (maybe only one step to check fidelity). This assumes that the system has been standardized to a color management system like the Adobe system described by Ashrunner.

    I'll be giving this a try over the next week or so. You can find a three day free trial of the color match/mix program from Liquitex under software at:

    It retails at Michael's for $19.95. Michael's advertises in junk mail circulars and offers a coupon for 40% off regularly priced items on a weekly basis making this fairly affordable.

  5. lunarhighway

    lunarhighway Member


    This is an interesting discussion indeed :)

    Recently found an interesting little trick to get the colours right on a plane i'm designing without to much guessing or a lot of technical fuss on the other hand.

    Basically i figured as there are so many factors that could have an effect on the colours it would be a good idea to at least start with the right ones. :p

    I wanted to paint a standard luftwaffe splinter sheme but all the color drawings on the web look different, so what i did was look up the RLM numbers of the paints and than googled up a model paint site that has all the rlm numbers listed with a neat little color sample next to it. I simply imported these in my software and used the same RGB values to paint the plane.

    When printed it looks acurate, and it's "suposed" to be acurate, so that more than enough for me :D
  6. Gil

    Gil Active Member


    That's a good technique. Were you able to capture the RGB values for future reference? I was thinking of a central repository of some sort so a particular color scheme and/or model could be catalogued. An RGB value, if authentic, will travel as well as any color scheme on the internet and with the ICC 98 standard we should come fairly close to the actual color.

    It still leaves the problem of edge and touch-up paint.

    I was looking at crosses of Liqutex with Plaid acrylic paints which are lower cost. The Liquitex Pixel2Paint software has a disclaimer which states that this might not give the desired results but I have a feeling that it will work anyway. Ran out of time and didn't get time to get the default color set for mixing which consists of the following colors:

    Cadmium Red Medium
    Cadmium Yellow Medium Hue
    Lt. Emerald Green
    Cobalt Blue Hue
    Deep Majenta
    Ivory Black
    Titanium White

    Decided that the best way to select the set is to go to Michael's select the above colors from the Liquitex section, and use them to compare to the Plaid colors to obtain the closest match. Sorry to be so cheap but at a delta of $2.50 per color it's worth the trouble. Once it's done I'll report back on the Plaid equivalents.

    Tell you how this comes out in a few days.


    P.S. Plaid's Silver Anniversary makes a great aluminum on paper!
  7. lunarhighway

    lunarhighway Member

    hi Gil

    Sounds like a good idea to make some kind of virtual color reference "sheet" adjusted to the needs of cardmodelers. of course this can only work for colors that are standardised. But i guess most military paints are/where, so it's just a matter of finding (acurate) colour chips and RGB-ing them.

    as for print - paint conversions. have you considdered using YMCK values?

    you can convert RGB values to YMCK (in the paint mixer in photoshop you can see both) , and you'll get the values the printer would use for Yellow Magenta Cyan and blacK... in theory using these colors in paint would give you the right mix only this scale assumes your paper is white and the paint transparant. so you'd have to find a way yo calculate the amount of white needed... i'm a bit unsure as to how that should be done. but i've noticed that altough there are 4 values even pure black wich is strangely a mix of black and colour and is 300%... that's as far as i got before i wanted to think about something else :p

    i once tried to match some blue-gray colour for the belly of a plane, and i just couldn't get it right untill i found there was also some red in there , making that subtle difference, so finding an easy way to understand the components of printed colours will be very handy in these cases i think, especially when using basic colours wich should be identical in any type of paint .

    anyway don't know if any of this is any use, just my addition to the creative thinking pool :idea:
  8. Gil

    Gil Active Member


    Yes, I thought about using the CMYK system. The problem is that paint doesn't come in printer's ink cans and there's no control over the dot size. The reason Black is included in the CMYK system is to overcome the muddy brown that mixing CMY together yields..., theoretically it should yield black but because of limitations in the pigmenting agents it does not, hence the addition of black.

    Your blue-gray example is the reason I went to Pixel2Paint. It gives red as a component for blue-gray. Saves a lot of experimenting and time. I have a hunch that this might work fairly well allowing a small batch of edge and touch-up paint to be prepared for a particular model build.

    Stay tuned.

  9. wunwinglow

    wunwinglow Active Member

    CMC2 might be a start, you can download it for free from this link(!)

    It expires after a while, but just uninstall and reinstall, or check to see if it has been up-issued. I have to say I find just changing the colour from cmyk to rgb in Coreldraw (11) gives as good a result. Also, you will find areas of each colour spectrum that do not translate from one system to the other. You can just twiddle the various settings, hit the enter key and the corresponding colour code pops out, with a swatch. And a whole lot more besides.

    Also, the type of paper you use, and the printer settings, will give different coloured output from identical files, even from the same PC and printer. I would be inclined to print out several sets of colour swatches using your usual set-up and materials, note which look the best on the paper and stick with them, even if they look odd on screen.

    Colour calibration? Can of worms? Anyone read 'Dune' lately?


    PS the program includes a colour checker set of swatches for you to print out.
  10. Gil

    Gil Active Member


    Been wondering where you've been...,

    Nice program, too bad they make you renew it every 24 days. The swatch printout is also useful for comparing colors and finding out how "off" your color is.

    I also found the following site in the theme of this discussion and which, with your background, you either know already or will be happy to bookmark it.

    Stay tuned as this investigation continues...,

  11. wunwinglow

    wunwinglow Active Member

    Hi Gil, sorry I've been otherwise occupied recently! My copy of the Munsell program originally said 30 days to renew, but I just checked again and now it says 142 days!

    Thanks for the link to the skinners site, I did know of it but hadn't visited for a while, so thanks for the reminder. Good site! There is even a monitor contrast check on one of the pages. Click on the 'Colors' sidebar link and look at the bottom of the page.

  12. Gil

    Gil Active Member

    Looking Around the Net at UnMatched Color


    The following was prepared to show the degree of variance of supposedly identical colors on the internet.

    I sampled one of the net's more popular modeling sites and chose the USAF SEA Color Scheme during the Vietnam conflict. Columns 1 & 2 are sampled directly from the offering site. Column 3 is the Federal Standard numbers in R,G,B values with column 4 the other SEA variant color set (guessing right now that two were used as there's a confliction of sources). Visually the variance is fairly obvious. I suspect that the samples were modified by the site so they'd show up better on a wider range of monitors (they sell paint). JPEG encoding shouldn't have any effect on these colors as the areas are large enough for the JPEG algorithm to center on the 24 bit R,G,B value.

    The point is that the net is an unreliable source of color information if you're thinking of using information on it as a color source (not that this is any real revelation). It also shows that the FS colors don't look that great either and I've looked at them over several color management standard settings (although column 4 looks good).

    This news update is in anticipation of developing the actual acrylic paint color set for a particular color scheme.

  13. wunwinglow

    wunwinglow Active Member

    Gil, would it help if you had scans of the original FS standard swatches? I have a copy somewhere about here....

  14. George

    George Member

    Scale effect on color

    Hi all,

    I have been investigating the color and it's effect on scale. This is due to that I have experienced the color on several models to be to "much".

    I found this excellent page which confimed my suspicions...the original
    color must be mixed with a certain amount of white to give the proper effect!!!
  15. George

    George Member

  16. wunwinglow

    wunwinglow Active Member


    Straight out of the real FS595, with a brand new Canon 3200F. Make of these what you will!!

    Tim P
  17. Gil

    Gil Active Member

    Color Matching, Dithering & JPEG Encoding

    Hi Tim,

    Chips out of the real thing! I'm impressed!

    I took your samples and looked at them in Photoshop. Found that they're not pure samples but when magnified consist of dithered colors which is consistent with the JPEG algorithm used to encode the swatches.

    Out of curiosity I randomly sampled each taking an average in R,G,B which is shown below. The boxes under the average for each is the R,G,B color rounded to the nearest whole number. This is interesting from the point of view that a pretty good facsimile can be recreated by sampling and averaging. It also means that if you aren't sure about the colors authenticity you can always apply the noise filter to "dither" it up! You'll notice though that the recreated samles are nearly indistiguishable from the originals...,

    Stay tuned...,


    P.S. I downloaded the image that is posted and sampled the averaged swatches..., they're nearly all a single color differing only by a single digit at best even though the original, created in Photoshop was "saved for web" as a jpeg file. In other words Photoshop hardly affected the color of the swatches created from the average of those you sent...,
  18. wunwinglow

    wunwinglow Active Member

    Hi Gil, when I did the original scans, the individual pixels were different colours, just running the cursor over each image gave a 'rolling' rgb read-out, so your average figures are going to be pretty good. I noticed the images picked up dust specks and handling marks ( the kind of bruising you get on matt surfaces) even though these were all but invisible on the originals. My swatches are quite old, but have been kept in the dark and dry. They certainly haven't been left open on the worktop with dripping brushes of mixed Humbrol paint hovering dangerously overhead!

    I can post the original bmp files if you are interested, but I reckon the colours you have look very good. On my monitor, anyway! Have you tried printing them out yet? I'd be intrigued to see the results.

    Best wishes,

    Tim P
  19. Gil

    Gil Active Member


    I printed them out. Interesting results...,

    The colors don't appear to be as dark as they appear on the screen. I'll have to try them directly from a vector format application to achieve monochromatic only output. Printing jpeg formats won't give exact results. One saving grace though is that both swatches are the same! So averaging color samples is a fairly good way of beginning to sample a color even off the web!

    I'm currently wrestling with getting the paint mix to match the FS as scanned. It's a bit of long process of mixing, painting, scanning, correcting and repeating till the colors appear and scan identically. The whole idea being that you'll only need five paints to achieve any color desired. Least that's the goal. Pixel2paint first round achieved really low value (low luminance results). Found Ivory Black, Ultramarine Blue and Payne's Grey are all the same color which is 10 points lighter than Ebony (lamp) Black. Egads, no wonder this is so poorly understood!

    More later as it comes to me...,


    P.S. Printed the tan sample with color management on and "saturated" selected. Extremely close to that on screen!
  20. Ashrunner

    Ashrunner Member

    Nice thing about color management is it actually saves ink in the long run. At least on photographs it does...don't know about model kits, but I would assume one color print is like any other...well, maybe not 8v)

    Anyway, Gil, I found your tests and averaging of colors interesting especially since the colors are the ones I am using on my repaint. I have attached a chart showing the three main colors I use. The values are all from the FS color list from the skinner's site listed a little ways down. Actually, the RGB values came from a similar site with the same color chart listings. I have come up with a total of eight other colors I will need for the repaint. However, the additional colors are aren't in question here.

    When I selected the colors I am using, I based the values on those which closely reflected my memory of a factory fresh painted B-52D. In the sun of the tropics, the colors bleached quickly. That I decided not to try and replicate.

    The colors are come out very close to my memory when using the Adobe (1998) color management process.

Share This Page