How does the color recipe determination in Color2Drop work? Does it work like the Pantone system ?
No. The Pantone system is based on the comparison of the target color to a color in a database or on the reference sample sheets. All items in the database have the information on how to mix them and this info is used in order to approximate the target color with a color error, which cannot be reduced. Color2Drop, instead, uses physical pigment parameters and algorithms without any target color database. It is based, in fact, on parameters extracted from certain paint base measurements (a "paint base" is a paint produced by a company and serves as a base for mixing recipes) and a mathematical model following the theory of Kubelka-Munk, which solves the problem of computing the perceived color starting from a mixing recipe (Drop2Color). Color2Drop on the other hand consists of the calculation of a recipe starting from the target color, but here no closed form solution exists. First, a color ramp is created for each paint base of a brand through the mixture of the paint base and white or gray paint in a known ratio. Then, the reflected power spectra of these samples are measured, and by using a method of statistical optimization, the reflection and absorption coefficients are extracted. Finally, these coefficients are used in the non-linear equations following the Kubelka-Munk theory with rewritten equations. In order to obtain the composition of a recipe that approximates the target color it is necessary to solve the inverse problem algorithmically in a high dimensional space. This needs to be done by including the spectrum of the required illumination that gives the perception of the same color as the target color using the selected light. This calculation is not unique, therefore, in the end several recipes are obtained thanks to the phenomenon of metamerism. Please note that "perception" in this context is used in the colorimetric sense, observing a small color patch alone without other colors around which would change the real color perception somehow.
I have heard that it is better not to use black for mixing colors. How can I obtain color recipes without using black paint in Color2Drop?
Some artists believe that black should not be used for mixing colors. This idea lacks logic, but it is known that artists are more creative than logical. It is enough to look at the composition of the brown colors in many paint bases and one can immediately discover that they often contain black pigments (PBk). For some reason, therefore, we tend to accept the use of black without objection by the supplier while we want to forbid it to ourselves. This is very illogical! Naturally, black (and whichever other pigment) needs to be used where it is necessary and in the right amount, bearing in mind that transparent colors behave differently from opaque ones. In any case, from version 2 of the program it is possible to choose a reduced set of colors to base the calculation of the optimized mixture on. If the black paint gets deselected (or any other color or set of colors) the system will not use it. However, it is important to note that it is not possible to obtain full black color without black pigments as those who invented the CMYK printing system know it well (K stands for Black!).
Sometimes Color2Drop gives me hundreds of drops/parts of paint, normally of the white paint. What do I do with a recipe that I cannot apply?
It is not due to a program error, but to a physical fact. Some light colors require a certain chromaticity that can be obtained by mixing paints with the appropriate properties of the color strength. Furthermore, in order to have the right final color, it is necessary to add a lot of white to overcome the color strength of the mixed bases. For this reason, the program provides recipes according to the increasing overall drop count, being, therefore the first most "economic" recipe. In case white paint is required in a great amount, the remaining part of the recipe needs to be mixed first, and then added, a little at a time, to the quantity of white that you wish to use effectively. Having to mix only two colors is an easy way to get the final color right. From this point of view, it is good to know that wet acrylic colors are usually lighter compared to when they are dry. Therefore, in the final mixing phase it is recommended to check on paper whether the mix is too dark and add more white if needed.
How do I know what percentage of Light Strength I must define in this system?
This parameter serves mainly to see whether in environmental light condition the mixed colors correspond to the ones on the screen. Defining a lesser value than 100% in the system involves the squeezing of the color space along the axis of brightness and therefore, the brightest obtainable white will be darker. You also loose in the saturation of mixable colors. Following this reasoning, it is easy to find the right value by putting a piece of paper or card with titanium white next to the monitor in order to identify the right value:
In the picture, one can see that 80% is the right value in order to perceive the same white on the monitor as the titanium white on the paper. All the other colors will be lighted with the same illumination strength. In the Color Mixing Tools this means that with D65 lighting, the lightest white will not exceed [227, 229, 225] RGB to which titanium white lighted with spectrum D65 corresponds.
example, in the following picture on the left you can see the weakly lighted
sample that gives a perception of a different color between the monitor and the
By adjusting the intensity of lightness to 20% (very low-light environment), the same color is perceived both on the monitor and on the paper.
Note: When mixing colors, it is always good to keep this parameter at 100%!
One of the parameters of Color2Drop is the chromatic error allowed between the target color and the recipe. Why is it set to 2 DE*94 and what happens if I set it to 0?
Color2Drop is able to calculate recipes also with zero error but this requires very precise paint ratios. This precision collides with various practical aspects: 1) The number of drops/parts resulting from each base can be enormous; 2) The physical dimension of the drops (or parts squeezed from tubes) is different from eye dropper to eye dropper (or it depends on the paint consistency which can be different from tube to tube, according to the physical properties of the pigments) and changes with the evaporation of the solvents; 3) The calculation of the recipes is based on the Color Matching Functions defined by the CIE (standard observers at 2 degrees) which differ from the color sensibility of every individual; 4) The parameters of the system have been obtained through measurements that have their own innate errors; 5) The lighting spectra are standard and certainly differ from those real ones used, at least where the illumination strength is concerned; etc. Considering all these variables, I have observed that the error set to 2 practically provides more valid and useful recipes. If the mixed color is not exactly the desired one, also due to the personal parameters of the eye of the painter, it can be easily adjusted manually because the color suggested by Color2Drop is in fact very close to the target color. From this point of view it is important to understand that an error of 1DE*94 is roughly the threshold for perceiving two fields of adjacents colors as different. Errors up to 3 among non-adjacent colors in a picture are invisible to anyone because the human color perception depends heavily on the surrounding colors!
There are about 19 lighting spectra given in version 2 (and higher) of this system. Which one do I have to use?
If a picture is placed in a room with artificial light, more precisely with a halogen light bulb, the corresponding spectrum will be close to illumination spectrum "A". If the light comes from the indirect sun, a "D65" (or "C") can describe it reasonably well. However, it is useful to know that lighting spectrum "A" dulls the cold colors and emphasizes the warm colors with an approximate 1:3 ratio, making it impossible to be able to mix the blues or even neutral grey. Therefore, if the aim is to have blues or saturated blues in the picture, the best thing is to mix all the colors for lighting spectrum "C" or "D65" and to let the eye of whoever is looking at it normalize the lighting condition. What matters most is that the colors are right relatively among themselves because the human eye has the capacity to eliminate the dominant colors automatically. Clearly, if by using light spectrum "A" the desired colors can be mixed, then when putting the screen of the computer near the final picture lighted by a halogen light, one will have the same perception of color from one part to the other. This does not happen if the colors have been mixed in a light different from the final one.The screen, in fact, has the white point around 6500K approximated by the spectrum "D65", while the halogen light is around 2900K. Therefore, the blue colors are emphasized on the monitor and dulled by the light bulb. The phenomenon, explaining why two colors with different spectral properties seem to be the same in a certain light but not otherwise is called "illuminant metamerism". If somebody is interested in mixing paint using the Munsell space as a guideline, illuminant 'C' must be used because this color space is defined only for this spectrum.
To provide an example, I have calculated the composition of gray around 40% (RGB= [100, 100, 100]) for sunlight "D65" using Color2Drop. Using Finity acrylics, I need to squeeze 2 parts of yellow, red and white with 3 parts of blue. The following picture on the left shows the colors directly from the tubes (not measured precisely, but only estimated), while to the right, you can see the mixed gray:
Placing the paper sample with the mixed gray close to the monitor, you can see from the photo that the gray on the monitor (the color at the top calculated for illumination spectrum "D65") corresponds to the color on the paper:
At night, with a halogen light (therefore, with a spectrum close to spectrum "A") on the table, I redid the photo relocating the sample close to the monitor. As the picture shows, now the color of the sample corresponds to the brown on the screen (to the color at the bottom calculated for light spectrum "A" :
From this image, you can see that halogen light dulls the blues and plays up the reds. In addition, the white of the paper has changed color. This photo illustrates this phenomenon well, thanks to the fact that the monitor emits the same light by day and night. However, our eyes do not notice this effect because they manage to remove the dominant color of the light by themselves without us being aware of it. This is just one of the many wonders of our eyes and brain!
If I have mixed a color for a certain light, how do I know how this color changes when the light changes?
Since the target color is defined in either an RGB or a CIE-Lab form, this system assumes that the RGB reference white point is at 6500K, that is, a light spectrum of about "D65" (sunlight with cold spectrum). However, there is no information on the initial target spectrum because the color is already in a tristimulus form. The mixed color, however, modeling an actual paint, is represented in the system through its complete spectrum with 36 values that cover the entire visible spectrum by 10nm at a time. With this information, it is possible to calculate the RGB value or the CIE-Lab coordinates with any light if we know the spectrum of the illuminant. This system has 19 standard CIE illuminant spectra from version 2 on. Once the color is mixed and the required light is defined, it is possible to select the menu of paint illumination with a click using the arrows or the mouse. Color2Drop will visualize both the color MIX and the ROUND resulting from the current light allowing you to estimate how the final color will change with the light that illuminates the picture relative to the target color seen on the monitor. In this example, the color was mixed with light spectrum "A" (as in the main image); however, upon changing the lighting (that is, moving the color sample to the sunlight) a great difference emerges:
When I start from RGB values of a photo, I obtain colors that are too dark. Why does this happen and what can I do to have the right recipe?
The first thing to look at is that the white point of the monitor must be set to 6500K. Then, you have to remember that the monitor emits light while the painting reflects it. Therefore, if the light is not strong enough, the perceived color of the picture will be darker than the color seen on the monitor. A very similar thing happens when one paints outdoors, for example under the sun, and then returns to the studio realizing that the picture seems too dark.
The solution is very simple thanks to the fact that Color2Drop provides consistent recipes among themselves. With any photo-retouch program, it is enough to modify the gamma curve by brightening the image. After a couple of attempts, one should be able to find the right quantity of brightness in order to have the desired color perception. Clearly, this is a subjective issue that can be resolved on an individual basis. Based on the spectral measurements that I have run there is no darkening effect caused by the system.
On the left photo, the color sample is lighted by very low intensity sunlight. On the right photo, however, the light is stronger and the sample color matches the color of the monitor.
With version 2 of the system, it is possible to regulate the light strength in order to make it correspond to real light. In the picture to the left, we are at 35% of the ideal light strength. The system is capable to mix the colors correctly and (also) consider this parameter as well if the resulting color is within the color gamut of the paint bases. However, the color space will be flattened in this way: the white of an image on the monitor cannot be lighter than the paper seen near the screen in the photo! However, it is not recommended to reduce the light strength (compared to 100%) to mix colors.
The real solution to the problem, in fact, is to find the sufficiently strong light to avoid creating this effect. In order to see if the light is the right one, it is enough to place a white or a titanium white card next to the monitor. If the hue of the card does not correspond to the neutral white on the monitor, it means that the light has a different color temperature compared to that of the monitor. If, vice versa, the value is different (lightness), it means that the light intensity is not correct and needs to be adjusted. The two photos above are excellent examples. These photos have been taken with the right exposition value with reference to the monitor, while the light intensity changes the color value of the sample placed slightly behind the screen. The difference of the brightness of the sample between the two images is obvious.
When I mix a color by hand, it seems identical to the color of an object or an image. When I photograph the object or the original image near the mixed color, why do I see a difference between the two colors that I otherwise cannot see in reality?
It is because of a phenomenon called "observer metamerism". Practically, everyone's eyes convert the full spectrum of colors into electrochemical signals through a type of receptors called cones. Tristimulus color theory, which explains our color perception, stems from the hypothesis (Young in 1802, following others' studies from 1777!) that the human eye has three different types of color sensitive receptors (cones). Cones are extremely sensitive to light and can distinguish among different wavelengths. The three type of cones (there are highly color sensitive ladies and no men with four types of cones too, but this doesn't change the meaning of what follows) respond to different wavelengths of light emphasizing the short wavelengths (blue), medium (green) and long (red). The sensibility curve is individual and varies from person to person. Normally we do not notice this difference, except in pathological cases like color-blindness, because we are unable to see through other people's eyes and because our eyes/brain compensate for the average light as they do for the average hue. These sensibility curves were first noticed in the late 1920s. The XYZ color space (based on direct measurements of the human eye response) was derived from a series of experiments conducted between 1928 (Wright) and 1931(Guild) involving 17 people (a very small number of human observers of nearly a century ago which still is the main reference today!). It was also one of the first mathematically defined color spaces created by the CIE (an international committee for the establishment of color standards) also known as CIE XYZ color space. The operation of many digital cameras (and also the Color Mixing Tools) is based on these curves. When two colors, that seem identical to us, have been obtained from different colorants, their spectra differ in reality. This is why color perception varies from person to person, that is from person to camera as the various spectra through different sensibility curves cause various tristimulus perception. On the other hand, the various spectra may give identical tristimulus values. This is why Color2Drop manages to give various recipes for the same target color despite the spectral difference from recipe to recipe. These various recipes provide the same color perception perceived with sensibility curves equal to the standard observers'. Upon changing the observer, the recipes may create a different color perception. The only way to have the same color perception between two different samples for everyone and for cameras is to have the same spectrum for both samples. Obviously, same perception in this case means that everyone will see that the two samples have the same color, but the individual perception will vary from person to person due to the individual filtering by our eyes resulting in a subjective conversion into tristimulus coordinates.
The same reasoning is valid when we try to compare the mixing recipes with a print. The inks used for printing have different spectra compared to the colorants used in the art world! In this case, we have an additional phenomenon as well: the print is created by remapping the RGB color space of the monitor/file into the color gamut of the inks thus increasing the color difference significantly.
I have finished the white included in the Color2Drop recipe. May I use the white of another brand that I have?
White is a paint base just as any other paint base. Even if it is placed in the achromatic zone in the color spaces, it has its color strength and a very strong chromatic influence on the trajectories of the mixed colors. The most opaque and commom white today is the Titanium white. Normally, the suppliers have a white that uses titanium dioxide because of its particular physical properties, but the crystal structure of dioxide and its treatment may vary. Moreover, it is possible that the ratio between the amount of pigment and the solvent differs from brand to brand. Therefore, it is not possible to substitute a white with another white directly. In these cases, you can try to mix the part of the recipe without the white and add your white afterwards adjusting the quantity by hand and correcting the potential chromatic variations.
Why does Color2Drop give me various color mixing recipes and not just a single one that already has the optimal number of drops/parts?
There are several reasons
for this. An important practical aspect is that each color drop has a slightly
different dimension depending on the speed with which the color comes out of
the eye dropper, how much the eye dropper is tilted, how much the hand is
shaking etc. (or, similarly, there are many reasons why the amount of paint
squeezed out of a tube can vary, plus it is hard to keep it constant from tube
to tube). Therefore, it is best to use many drops/parts of each paint to get
the desired result and avoid making a mistake due to the different quantities.
Taking more drops of each paint, their average quantity has less variation
compared to single drops (this is well known in statistics applied to physical
phenomena), therefore, the result will be more accurate. For this reason, the
preferred recipes are the ones with the most similar amount of paint bases as
opposed to the ones with a significant difference. For example, if a recipe has
2 parts of a base and 3 of another one, it is possible to mix accurately 4 parts
of the first and 6 of the second, assuming that 4 drops/parts have an overall
amount which is more constant than just one drop. If in order to obtain the
same final color I have another recipe with 1 drop of a base and 5 of another,
in the end I will have to mix 4 drops of the first and 20 of the second paint
base in order to have the same accuracy as the previous recipe. In other words,
in the second case I will have to mix more paint, a total of 24 drops, compared
to the first case where I was able to use only 10 drops.
It seems to me that the color of the paint changes when I add a final transparent coating or fixative. Does it have to do with perception or is there an actual chromatic variation?
Light crosses the layer of transparent paint film (that acts as a light filter) and the spectrum of the perceived light contains both the film and the paint beneath (with the light bouncing back and forth in the transparent layer depending on the refraction indexes). The combination of the two is perceived as just one color by the eye. This means that there is a real difference in the reflectance spectrum and, therefore, in the color perception. This is one of the reasons why this system does not provide accurate recipes for very dark colors (such as some dark greens of Pen Color). At this point, with the low lightness of the paint, the spectrum is more attributable to the property of the film surface reflection than to the pigments, exactly what happens to glass with something black behind it when one can see the reflection but not the color black.
Is it possible to include the parameters of transparent paint lines into the Color2Drop / Drop2Color system?
No. Transparent paint provides a color perception that is born from the interaction of light filtered from the color film and the support (paint background surface) that appears partially visible. Color2Drop / Drop2Color do not have any information on supports (canvas, paper, cardboard or else) or on the layering order of paint, but they take into account only the chromatic responses of the paint bases assuming a complete opacity. In reality, I could have included the equations in order to consider the light reflection from the supports because the mathematical model is well known, but I would have created an additional mathematical game without useful effects for painters. In this case, we would also have to compute the paint film thickness in the recipes in order to obtain the target color. The problem is that a fraction of a second can change the thickness of the film when spraying with an airbrush (and paint film thickness cannot be controlled with a brush on a nanometer scale) making the use of this information impossible. Note, however, that when a transparent color is mixed with an opaque one, an opaque color develops. For this reason, various transparent colors have been included in this system, but opaque ones still need to be used in each mixture. Otherwise, the color in general is not going to be the same as the target color and the computed recipes will not be valid.
Where can I find programs to read the RGB value on my screen?
There are many programs of this kind. You only need to type in the words "color picker" into a search engine. A simple one I use is Color Mania.
Where can I find in-depth information on the world of colors on the net?
There are many sites where you may find useful information. Here is a brief list of them that I have found interesting:
Color Mixing Tools >