Creating Skin Tones
by Allison McCue and Jerry Monds
Student Projects, Pigments&Binders, Photography
After doing a lab where we created egg tempera and watercolor, we wanted to have a chance to try making oil paints. We combined this desire with the information on our photography and light units to come up with the idea for our project. We wanted to see if, using three photographs, we could emulate certain skin tones with oil paints we made. Our three photographs were a Caucasian, a tan Caucasian, and an Asian and the three colors we chose were a red (vermillion), a yellow (yellow ochre/light Italian-selected since it looks like a combination of brown and yellow) and a white (zinc white). At the end of our experiment, we wanted to see how close we had approximated the skin tones by comparing the colors of our painted samples to the color of the photograph using the Colorimeter.
Top Left: Caucasian (2nd from right), Top Right: Asian
Bottom Right: Tan (On the Left)
We took Dr. Gregory’s advice and paint on a green background to avoid overly vibrant and unconvincing skin tones. For the base, we imitated the one demonstrated by Nigel McGilchrist: a square piece of wood (a block like the ones we used in the embossing press would be fine, only smaller like the one that Nigel McGilchrist used to demonstrate), painted white with glossy paint to imitate the gesso and so that the base is highly reflective. This was a good base for the oil paints: as Dr. Gregory pointed out, the shiny white surface acts as a mirror and transmits the richness of the multilayered oil paints, creating a more vibrant image for the eyes.
Materials: One square piece of wood, sand paper, four pigments (Vermillion, Yellow ochre/Indian light, Zinc white, Bohemean green), Porter Interior Latex Primer #1129 White, linseed oil, a large square mirror, a glass muller, one 1.5” paint brush, a razor blade, three photographs, a mixing plate, a glass dropper, mineral spirits, 4 glass jars with lids, 2 small paint brushes, a table spoon, a hair dryer, and a Colorimeter.
Preparing Paints: To make the paints, we added 1 1/2 spoonfuls of pigment (zinc white, vermillion, yellow ochre, Bohemean green earth) with two spoonfuls of linseed oil onto our large mirror.
We mulled the two together and collected the paint in the center using a razor blade. We then used the razor blade to transfer the paint to a glass container.
Finally, we labeled the glass bottles with
their color titles.
Preparing the Board: First, we sanded the piece of wood with a small piece of sandpaper. Then, we painted the wood with Porter Interior Latex Primer #1129 White. We let the wood air dry for 20-30 minutes.
The next day, we painted the board with our Bohemean green earth. To see the effects of the shade of green, we painted one half of the board green with the other half, light green (Bohemean green earth mixed with three spoonfuls of zinc white paint). We let the board dry, again, for seven days (in McCrady since we needed a place warmer than the lab to dry the board). We also decided not to repaint the second half of the board dark green since it took too long for the first coat to dry. Whether wet or dry, the oil paints are quite shiny.
Making Skin Tones: We decided to try three times at creating each skin tone. We also cleaned the brush (using mineral spirits) in between each of the three skin tones so that we didn’t contaminate the next attempt. We mixed our colors using a testing plate. One additional note, we painted swatches of each our basic colors so we could use the Colorimeter to test them as well. In order to assure that the Colorimeter testing was unbiased we used one paint sample and painted two swatches, one on the dark green and one on the lighter green section.
Caucasian: We thought the Caucasian was a combination of red and white, therefore we did not include yellow in these trials. 1) The proportion we tried was one small drop of red to 12 drops of white, taken from the small glass dropper. 2) We modified the original ratio by using the same amount of white and half as much red. 3) The third try was the same amount of red as in the second trial with half as much white. This resulted in too little paint and too thin a coating on the board.
Asian: We thought the Asian was a combination of yellow and white; therefore we did not include red in these trials. Our first try on the Asian was too little paint so we decided to omit it. 1) The ratio here was one drop of yellow to six drops of white. 2) Here one drop of yellow to four drops of white. 3) Lastly, one drop of yellow to three drops of white.
Tan Caucasian: We thought the tan Caucasian was a combination of red and yellow; therefore we did not include white in these trials. 1) The ratio for this trial was one drop of red to five drops of yellow. 2) Here the ratio was one drop of red to three drops of yellow. 3) Here, based on our lack of success in the first two trials, we decided to include white. The ratio was one drop of red to five drops of yellow with one drop of white.
In
order to dry all of our painted swatches, we left the board in McCrady again
for three days this time and hastened the drying process using a blow dryer. Some of the thinner painted swatches tended
to bubble with the use of the blow dryer, perhaps because the heat was too
concentrated. If time permitted, air-drying would be most recommended.
PHOTOGRAPH
|
COLORIMETER
|
Caucasian
|
56.41, +10.21, +6.48
|
Asian
|
77.72, +4.39, +18.00
|
Tan Caucasian
|
37.50, +18.92, +12.70
|
COLOR
|
DARK GR. BACKGROUND
|
LIGHT GR. BACKGROUND
|
Vermillion
|
41.17, +45.68, +26.91
|
42.97, +42.28, +24.75
|
Yellow ochre
|
57.63, +8.73, +47.62
|
59.13, +9.05, +46.71
|
Zinc white
|
95.17, +.09, +2.08
|
93.61, +.38, +1.76
|
|
SAMPLE |
DARK GR. BACKGROUND |
LIGHT GR. BACKGROUND |
|
Caucasian paint sample 1 |
68.54, +11.00, +7.10 |
67.47, +8.47, +6.24 |
|
Caucasian paint sample 2 |
79.90, +2.56, +1.99 |
79.82, +3.16, +2.84 |
|
Caucasian paint sample 3 |
75.20, +2.75, +.19 |
76.31, +2.43, +1.05 |
|
Asian paint sample 1 |
86.29, -.74, +13.40 |
87.39, -.54, +15.51 |
|
Asian paint sample 2 |
85.11, +1.14, +20.56 |
85.01, +.94, +20.88 |
|
Asian paint sample 3 |
80.49, +.55, +21.08 |
82.38, +1.42, +24.21 |
|
Tan Caucasian paint sample 1 |
48.11, +8.26, +32.14 |
53.17, +13.08, +39.89 |
|
Tan Caucasian paint sample 2 |
45.70, +21.77, +31.70 |
48.20, +22.63, +34.66 |
|
Tan Caucasian paint sample 3 |
51.61, +11.21, +37.40 |
54.10, +11.82, +41.95 |
First, we found that there was not a significant difference in the Colorimeter readings between the light and the dark green backgrounds. We think this might have been the case since the paint samples were too thick for the green to show through. From now on, we will discuss paint swatches with no reference to dark or light background, just to paint sample number.
The samples that matched the pictures best were the Caucasian sample #1, Asian sample #3, and Tan sample #2. The Caucasian sample #1 although significantly lighter than the photograph matched the amount of red within two points while the yellow was even closer, matching within .5 points. Caucasian sample two did not match because it was too light, with a difference of almost fifteen points, also the amount of red differed by approximately eight points and the amount of yellow by about four points. Caucasian #3 was similar to Caucasian #2 with the exception of being five points darker. The best Caucasian, selecting LAB coordinates from all three samples would be to use all three coordinates of sample #1. Asian #3 matches the best because the lightness is only off by three points while the other two attempts are off by about eight points, yet the amounts of red in all three samples was about four points too low. Our conclusion was that we needed to add more red in all of our samples. Each of the three samples matched the amounts of yellow quite well, with the largest difference being six points. The best Asian, selecting LAB coordinates from all three samples would be to use the L from sample #3, the A from sample #3, and the B from sample #2. The tan Caucasian was the most difficult to match, most notably in the higher than necessary amount of yellow in our paint samples. We chose sample #2 because there was only eight points difference in the L coordinate, our sample being too light. As for the A coordinate, sample #2 was three points too high. The difference in the B coordinate for sample #2, as was the case for all three tan Caucasian samples, was almost twenty points too high, therefore too much yellow. The best tan Caucasian, selecting LAB coordinates from all three paint samples would be to use the L, A, and B from sample #2.
Overall, we think the project was quite successful, doing a good job matching colors, but realizing the faults in our estimates after comparing our work to the Colorimeter readings. We think our score on successfully matching colors was a 90/100. Our one flaw was in thinking that the tan Caucasian was more yellow than was actually true.
We would like to acknowledge Dr. Nigel McGilchrist and Dr. Susan Gregory for inspiring this project with their lectures, Jane Kim for providing the Asian photograph, and Dr. John Bordley in providing instructions on creating oil paints.