31 October 2006

Practical color mixing 3: chroma

This is the third post on mixing color. The first was about value and the second was about hue. Now it's time to talk about chroma.

Identifying chroma

As with value and hue, the best way to identify chroma is in terms of relationships. How intense is the color you're looking at compared with the intensity of other colors around it? Chroma can be hard to separate out from value; light colors sometimes look more intense than they are, and dark colors sometimes look less intense. You get better with practice.

Bad chroma!

Many artists—mostly amateurs, but also some professionals—seem to have trouble identifying chroma correctly. They often paint at a higher chroma than what they see, and they often seem unaware that they are doing so. Let me give you an example. I was browsing through art books in a bookstore the other day and found one about the painting techniques of the impressionists. It’s a very well written book, based on lots of research on the individual methods of many 19th century artists. There are a number of demonstrations in which the author copies a section of an impressionist painting, using the methods of the original artist. In every single case, throughout the entire book, the author gets the chroma badly wrong and pretty much everything else right. In particular, almost every color is one or two chroma steps higher than the corresponding color in the original. Impressionists were not known for making dull pictures, but the author felt the need to “improve” the originals by bumping the chroma, even though she was clearly making a serious attempt to use the same or similar pigments and techniques. What’s more, I don’t think she knew she was doing it. I think she believed she was doing precise copies, but failed to see chroma differences right in front of her face. That’s just a guess on my part; some of the pigments used in the typical impressionist palette were fugitive, so she might have been deliberately compensating for their tendency to fade. But if that’s the case, I couldn’t find where she told us that, and she was certainly increasing the chroma even in areas corresponding to those painted with lightfast pigments. So either the reproductions in the book are badly messed up (and no one caught it) or this artist has a remarkable insensitivity to chroma.

I see similar errors on internet forums in which amateur artists post copies of old master works. The chroma is usually too high—often much, much too high. That might have something to do with how the work has been photographed, digitized, and presented on computer monitors, but in case after case, the posted copy appears consistently more chromatic than the original, even when the artist has shown them side by side. The artists usually seem unaware of this difference, and sometimes have trouble seeing it even when it is pointed out to them.

There is, of course, nothing wrong with deliberately pushing chroma for dramatic or decorative effect. I do it myself sometimes. My concern is with artists who do this unthinkingly, either because they just have an unconscious bias toward “brighter” color, or because they think that chromatic colors are always better or prettier. I think part of the problem is that we have been conditioned to think about pictures in terms of photography. Many artists work from photographs, and even those who do not have spent a lot of time looking at photographs. Most color films and developing methods are deliberately designed to push the chroma, and most consumer digital cameras are designed to do so as well. That makes the scene more “colorful,” and many people seem to think that a “good” snapshot is one that has a lot of chroma, regardless of how chromatic the original scene was. Over time, we’ve become accustomed to looking at pictures with lots of color intensity. That’s what we think pictures are supposed to look like. And, I suppose, many of those who buy art think that way as well, so there may be a commercial incentive to paint very “colorful” pictures.

If you are looking to make a painting that seems like it is full of color, the problem is this: if you are looking at an array of high-chroma colors, the visual system habituates. We see chroma (and other aspects of color) in terms of relationships at least as much as we see absolute values. A whole bunch of intense colors looks lurid, but it doesn't give the impression of a really colorful scene. The better impressionist painters, for all of their emphasis on a modern palette of intense colors, understood this. They used chroma carefully, calculating the effect of one color against another. They were able to get high chroma colors to really stand out by juxtaposing them against much duller colors or dull optical mixtures, deliberately creating a strong visual contrast. That’s something the author of that book on impressionist technique didn’t seem to understand, for all of her impressive technical knowledge of impressionist methodologies. She’s looked at hundreds of impressionist paintings and copied dozens of them, yet she fails to see how those artists used chromatic contrast. Her copies, as a result, are much less interesting than the originals.

If you like color, then learn how to identify the chroma you see around you. Learn how to mix and use subtle mixtures of neutral and near-neutral colors. If you like intense colors, learn how to create the visual impression of really high chroma by using contrast, rather than just blasting away with unmixed colors right out of the tube and hoping the viewer likes “colorful” art. Even today, with a viewing public accustomed to artificially enhanced color, there are plenty of really good artists who know how to use color more effectively, and they stand out from those who do not. So let’s put an end to chroma cluelessness.

I’ll close my little editorial rant here with this: Enough with those bright orange skin tones already! Even Caucasians who spend way too much time in tanning booths have skin that’s way less intense than cadmium red mixed with yellow ochre, or any of the other ways to mix luridly awful skin tones. Thank you very much.

We’ll now return to your regularly-scheduled paint mixing post.

Working with low-chroma color

Almost all paints are high in chroma right out of the tube. Since most of the world is low in chroma, the majority of a realistic painting will consist of neutrals and near-neutrals. So a realist painter is going to have to spend a lot of mixing time reducing the chroma of paint.

How do you do that? At one level, it's easy because, most of the time, whenever you mix one blob of paint with another blob of paint, you end up with something lower in chroma. By that, I mean that the result is less intense than the brighter of the two colors. Often, it's less intense than either of them. So if you want to reduce the chroma of a paint color, pick another paint that's lower in chroma smoosh them together. The problem, of course, is that if you also want to have control over hue and value, you’re going to have to pick your mixtures carefully.

Reducing chroma with mixing complements

A common method of chroma reduction is to mix in a complementary or near-complementary paint. For example, if you want to reduce the chroma of a bright pthalo green, mix in a violet-red (i.e., a magenta). Complements will usually reduce both chroma and value. You also sometimes get hue shifts when mixing colors from the other side of the color mixing wheel—the mixture follows a curved path around the wheel rather than a straight path toward the center. Because of the peculiarities of individual pigments, there is no way to predict these hue shifts without just mixing two paints together and seeing what happens. You can pull the mixture back to the original hue, however, by adding a third paint that is complementary to the color you’ve mixed. Once you have the hue and chroma right, you’ll need to check to see if the value has now been brought too low. Of course, if you add white to increase the value, the chroma will be decreased. So the method of adding the complement can result in frustration as you try to chase the color of a mixture to the right combination of value, hue, and chroma. For low-chroma colors, I find this process much less frustrating when I start with fairly low-chroma paints, such as earth colors and a few others. That way, I do less chasing of color and less mixing overall. It’s also easier if you mix each color to the correct value first, then mixing them together for the desired hue and chroma.

Yellows and violets, although they are on opposite sides of the color mixing wheel, don’t work well as complements (the hues shift severely rather than mixing toward neutrals). To dull down a yellow, it is easiest to mix in a duller yellow or orange, such as raw umber or raw sienna. You can also mix in a neutral gray of the same value (see below). To get a dull violet, the easiest course is often to ignore violet pigments and mix your own violet. Because mixing reduces chroma, the right combination of red and blue will give you a violet of the desired degree of dullness. A good dark dull violet can be made with ultramarine blue and burnt sienna, for example.

Other than yellow and violet, it is very helpful to experiment with, and memorize, pairs of complementary colors. As a general rule, if you want to dull down an intense color, choose a dull complement. Blues have mixing complements in the range of warm yellows, oranges, and middle reds. Middle and cool greens have mixing complements in the range from middle reds to violets. Warm greens have mixing complements among the violets. Some of my favorite mixing complements include raw sienna/ultramarine blue, viridian/pyrol ruby, Prussian blue/Venetian red, and ultramarine blue/raw umber. I expect that most artists develop a set of strongly preferred mixing complements.

Reducing chroma with optical color mixing

If you put a bunch of small dabs of different colored high chroma paints next to each other, then step back far enough, they will blend optically and look like one color. The perceived color will resemble what you would get if you mixed all of those colors together—i.e., it will be lower in chroma than the colors that go into it. If you take this approach to extremes, you get pointillism, which I don’t personally find to be very attractive or effective. Used with more subtlety, however, optical mixing can be one of the best ways to make neutrals, because by controlling the structure of paint blobs, you can create effects that are much more visually interesting than a flat region of neutral color. The eye sees the optical blend, but is also aware of the color variation. You can partially blend colors together, layer them on top of each other while allowing different amounts of lower colors to show through, create interesting textures, or use any of a number of techniques for optical color blending.

Reducing chroma with white

Almost all colors lose chroma when mixed with white (as I’ve already noted, a few dark cool transparent pigments show an initial increase in chroma when mixed with a little white, then drop chroma as they are lightened further). Of course, they also get lighter, but dull light colors are often exactly what you want. For example, imagine that you are painting a piece of blue cloth. The part of the cloth that is highest in chroma will be the “midtones”—the part of the cloth that is illuminated, but is near the form shadow boundary (the terminator) and turning away from the light. As the form turns toward the light it gets lighter in value and also less chromatic. You can create the same effect with paint by mixing a blue color (cobalt blue, say) with more and more white as it turns toward the light. The paint becomes lighter and less chromatic, just as the blue cloth does. You may need to adjust the mixture to make the chroma, value, and hue changes exactly model what you are seeing in front of you, but just mixing with the appropriate amount of white gets you into the right ball park.

Sometimes, you’ll find that white reduces chroma faster than you want it to. The mixture becomes “chalky.” I’ll discuss how to deal with that problem below, when I talk about high chroma color mixing.

A color that has been lightened by mixing with white is called a tint. Tints are high in value and low in chroma. They are also called “pastels.”

Reducing chroma with grey

Instead of using mixing complements, it is possible to reduce the chroma of a mixture without having much effect on value or hue. To do that, use a neutral gray of the same value. The chroma will go down without significantly affecting the other parameters of color. Black mixed with white does not make a neutral gray—it’s much too cool. A 50/50 mixture of ivory black and raw umber, however, is very close to neutral. Adjust the value of this mixture by adding whatever amount of white is required. You may want to mix up a string of neutral grays in advance and use them to easily adjust chroma. You can also buy a set of neutral gray oil paints from Studio Products, graded according to Munsell chroma intervals.

So here’s a good way to work with low-chroma colors without driving yourself nuts. First, start with low-chroma colors such as earths. They will still be too high in chroma for a lot of purposes, but they are a lot closer than, say, a cadmium red light. Second, pre-mix strings of colors you’re likely to use. Each string is one hue, ranging in value from the lowest you will need to the highest. Also mix a string of neutral grays in the same value range. Small chroma adjustments are fairly easy with nudges of small amounts of complementary colors. When you need to pull the chroma down significantly, first mix the right hue and value from combinations of paint from your strings and, if needed, nudges of other colors on your palette. Then adjust the chroma downward by adding some of your neutral gray at the same value as the mixture you’re working with.

Reducing chroma with glazing

If you paint one color thinly over another color, you get an optical mixture. Blue glazed over yellow produces a green, for example. You can use this effect to reduce chroma, since an optical mixture is darker and duller than the colors that make it up. Michelangelo, for example, sometimes made a dark dull blue by glazing ultramarine over black.

The browns and the brown-ish

It’s worth briefly discussing brown colors. Brown doesn’t appear in the visual spectrum or in the named colors on the outside of a color mixing wheel. Colors labeled “brown” are yellows and oranges that are fairly dark and low in chroma. Brown, therefore, isn’t a color per se: it’s a zone within the overall color mixing wheel. There are plenty of earth colors that start out in the zone of brown, and a few non-earths as well. To make a bright yellow or orange more brownish, it needs to be dulled down and darkened (mixing with white won’t do it; you get a pastel tint). If you mix a yellow or orange with a gray of equal value, you will usually get something you could call a brown. If you mix with black, you will almost always get a brown. Some yellows, when reduced in chroma with grey or black, shift their hue toward green. Just a touch of grey or black makes a green gold; more grey or black makes a greenish umber-like color. Bright yellow-reds (oranges), if dulled down with grey or black, make more clearly brown colors, without those green tones. Bright reds, if dulled down with grey or black, make a maroon.

A color that has been darkened by mixing with black is called a shade. So a mixture of cadmium red and black would be a shade of cadmium red. Because even a little bit of black reduces chroma drastically, some artists banish black from their palette, claiming that it is a “dead” color. That’s silly, because there are times when you want dark colors that are very low in chroma. The real world has such colors in it, after all, and many great Old Master paintings were made with the liberal use of black. Nevertheless, you should be aware of how powerful black is as a reducer of chroma and use it with care.

Working with high-chroma color

For almost all colors, chroma is highest with paint straight out of the tube. Some pigments that are dark and transparent (such as Prussian blue or pthalo green) reach maximum chroma with the addition of a little white. Generally, however, if you want high really high chroma, the way to get it is to have a tube of paint with exactly the right hue and value, and just paint with that. Sometimes, you can make that work. At other times, you don’t have quite the right color, or the right color just doesn’t exist in pigment form.

The highest chroma mixtures are blends of high-chroma paints of similar hue. If your cerulean blue is just a bit too greenish, don’t mix it with red, mix it with a high chroma middle blue such as cobalt blue. You want to draw a line through the color mixing wheel that stays as close to the outside of the wheel as possible. So if you like to work with a lot of high chroma colors, then it’s to your advantage to have many tubes of high-chroma paint. That way, when you want a particular color, you will usually have two colors that are similar enough to that color that you can mix them without taking too much of a hit to the chroma.

Warm colors are at their highest chroma when they are fairly light (high in value). Cool colors are at their highest chroma when they are relatively dark. At maximum chroma, warm colors are more chromatic than cool colors. But that’s OK, because warm colors have higher maximum chroma in the real world as well. If you want cool colors to compete for the viewer’s attention with warm colors, however, you’ll need to keep that difference in mind.

Simultaneous contrast

Because of the way the visual system works, a color is perceived as more chromatic when it is placed next to its visual complement. The impressionists made frequent use of this principle. A bright magenta looks more intense when it is surrounded by a dull green. The complements identified in the Munsell color system more accurately reflect human color vision than those in the antiquated three primary color wheel. Here are the visual complements from the Munsell color wheel:
  • Yellow <—> purple blue
  • Green yellow <—> purple
  • Green <—> red purple
  • Blue green <—> red
  • Blue <—> yellow red
Remember that these are visual complements, not mixing complements.

Masstone and undertone

We like to pretend that pigments, paints, and colors are all the same thing. In reality, pigments have many characteristics separate from their “color.” And pigments behave differently in different binding media. One of the most important aspects of pigment variation is in masstone and undertone. Masstone is a pigment’s color when it is applied thickly. Undertone is a pigment’s color when it is applied in a thin layer. Lots of pigments display a huge difference between masstone and undertone, and that often has a lot to do with chroma. Most commonly, masstone is duller than undertone. The difference is most marked with relatively transparent pigments, but many opaque pigments show significant differences as well. For example, in the self portrait I did earlier this year, the background consists of yellow ochre glazed thinly over white paint. Yellow ochre is generally described as a dull pigment, and in masstone that’s true. But the background of the painting is quite intense. The hue is also much more orange than that of yellow ochre in masstone. We get so used to application of paint in thicknesses that make use of masstone that we often forget how paints behave when applied in very thin layers.

Maintaining chroma at high values

With many pigments, it’s hard to get high chroma at high values. Because white lightens the value of paint and also reduces chroma, mixtures with a lot of white become pastel tints. If that’s not what you want, then you may consider the mixture to be “too chalky.” How do you avoid this chalky effect when you’re trying to make light colors that have relatively high chroma?

One way is to start with very high chroma paint. The chroma reducing effect of white is then balanced with a strong base chroma. It can also be a good idea to avoid titanium white, which tints very strongly and can have a greater effect on chroma than other whites. Try zinc white or, if you’re using oil paint, lead white. Studio Products sells an oil paint called “optical white.” It’s zinc white ground with hollow nanosphere particles (really). They claim that this white reduces chroma less than other whites. I haven’t tried it (yet), but because their other products are so excellent I think it’s worth consideration.

Maintaining chroma by glazing

One effective way to maintain chroma is by glazing. If you apply paint very thinly over white, you can get a higher chroma than you could by mixing that paint to the same value using white. And a transparent paint that is applied a little more thickly can be more chromatic at low values than you might be able to obtain with a mixture of the same hue.

24 October 2006

Jan van Eyck

was credited by the 16th century artist and biographer Vassari as the inventor of oil painting. That's not the case, but in the early 15th century he was one of the pioneers of the modern use of oil painting as a primary painting medium.

Apparently, he had a sense of humor. This painting, which may be a self-portrait, was made with a frame carved as part of the panel. At the bottom, it's inscribed (in Greek letters in the Flemish language) with what appears to be a pun. It can be read either as "As I can," or "As Eyck can."

Getting back to drawing the figure

Since the Summer, Kiri and I haven't been able to attend figure drawing and painting classes at the New England Realist Art Center. She got too pregnant, then she had Brendan, who does make things more difficult.

We have been attending sessions on Monday nights of the Worcester figure drawing group at Worcester State College (actually, we've been alternating weeks while the other one watches the kid). An open figure drawing session is very different. First, there's no instruction. Second, it's a lot cheaper. Third, poses run from one minute to fifteen minutes. At NERAC, poses run for four or five three hour sessions. So it's been a bit of an adjustment.

20 October 2006

Archival permanence

Over time, all paintings deteriorate. Badly made paintings deteriorate quickly, sometimes within a year or two of completion. A painting made with a high level of craftsmanship can last for many years before noticeable changes occur.

For most of us, it isn't worth going to extreme lengths to make our paintings as permanent as they can possibly be. You could, for example, choose to paint on high-tech aluminum honeycomb panels. These are light, long-lasting, and much better supports for painting than most of those used by artists, because they don't significantly expand or contract with changes in temperature and humidity. They also cost hundreds or thousands of dollars. If you know that you are a visionary artist who will be producing work of breathtaking magnificence that will be of incredible historic significance, you owe it to future generations to eat only cheap prepackaged noodle dishes at each meal so that you can afford to paint on the most permanent and expensive supports (until you work starts to sell for many thousands of dollars—then, go ahead and treat yourself to a nice juicy tofu burger).

For the rest of us, not so much. Most paintings by even fairly good artists won't be saved for much more than a generation. The best way to preserve your paintings is to make them really, really good (or really, really popular, which 20th century artists demonstrated to have no correlation with good). A painting that people like a lot will be hung on a wall in a room that has a reasonably constant temperature and no wild swings in humidity. Almost any painting will survive for a long time under those conditions. And if people really like it, it might hang in a museum or get restored by a conservator if it starts to show signs of wear and tear. If a painting isn't that great, then even if it's made with excellent crafstmanship and highly archival materials it's likey to be kept in the attic, basement, or garage for years at a time. Even well-made paintings won't last long under those circumstances, and when they start to fall apart, no one will pay for a conservator to fix them. So the most archival quality a painting can have is to be so well-liked that the owner (and the owner's heirs) could never imagine putting it in a moldy basement.

(Of course, if you are a very famous celebrity such as Sir Paul McCartney, your incredibly bad vanity paintings will be treasured and preserved for centuries. Go figure.)

Nevetheless, I think it's a smart to construct paintings with quality materials and good craftsmanship, if only so that customers won't complain until after you are dead. Here are some guidelines for oil painting. If you don't follow them perfectly, it won't cause your painting to explode. But the closer you adhere to them, the more likely your painting will be to last a long time under optimal conditions, or survive brief periods under poor conditions. If you want a painting to last a long time under poor conditions, oil paint is a very bad choice of medium.
  • Rigid supports are better than fabric supports. Fabric is flexible, and every time it flexes (as it will do when temperature or humidity changes) the bond between the support and the paint is affected. Over time, that's very bad for a painting. Copper, steel, and aluminum panels are excellent supports for painting (although they can be heavy). Wood is OK only if it has been seasoned for a year or two after being cut and planed to size. Hardboard is probably OK if there is a good barrier between the panel and the paint. Tempered hardboard is stronger than untempered and that makes it better (despite what some sources say) even though there is a slight amount of oil in the surface of tempered hardboard. Medium density fiberboard is OK only if it is very well sealed on all sides against moisture.
  • It may be that polyester will turn out to be the most archival fabric, because it is more dimensionally stable than organic fabrics like linen and cotton. We don’t know yet.
  • Oil grounds are good to paint on. Lead grounds are the best oil grounds, because lead is a very flexible pigment. Acrylic primer ("gesso") is probably a decent ground to paint on (we'll know for sure in 100 years) but murder on brushes. Traditional gesso is probably an OK ground on a rigid support (the hide glue in gesso is very strong, which is good, but likes to absorb water, which is bad).
  • Use permanent pigments. Alizarin crimson is not permanent, especially in mixtures and when applied very thinly. Impermanent pigments will fade or become dull over time.
  • If you use linen, cotton, or hemp as a support, don't put paint or oil primer directly onto it. The oil will rot the fabric. You need a barrier, such as hide glue or acrylic primer, between the paint and the fabric. Make sure the barrier covers the sides as well as the front of the canvas.
  • Don't apply a lot of thick paint. Thick, heavy layers of impasto are much less permanent than thin layers (about the thickness of a layer of house paint). Several thin layers (allowed to dry in between) are much more permanent than one thick layer. A few expressive blobs of impasto here and there are not going to cause problems, but large areas of thick paint are bad.
  • Linseed forms the strongest paint film of the drying oils. Walnut is less strong. Safflower and poppy are weaker still. Because the same stuff that make the paint film strong also yellows, linseed will yellow more than other oils. But go for a walk through a museum with paintings three or four hundred years old. You probably don't find yourself thinking, “Wow! those paintings now suck because they've yellowed.” (Ignore Brown School paintings from the 17th and 18th centuries that were deliberately painted with an overall dull yellowish tone.) You can barely notice the yellowing, and those paintings were almost all done in linseed. Whites are a little warm, blues turn slightly greenish. That's how bad the yellowng gets on a well-made painting. It's barely noticeable, although some paint manufacturers will try to scare you into buying special “non-yellowing” paints made with oils that are less strong. Personally, I only use paints made with linseed and, to a lesser degree, walnut. I avoid paints made with poppy and safflower. If you do use safflower oil, be aware that the kind you can get in a grocery store is almost certainly not the kind that dries properly when mixed with oil paint.
  • There are a number of good reasons to avoid student grade paint, but archival permanence is not one of them. Student grade paint from a good company will be as archival as their artist-grade paint.
  • It is best to not add anything to your paint—no mediums, no solvents, no nothing. If you do add stuff to the paint, add only a little bit (less than 20% of paint volume). If you add solvents, don't make the paint watery or washy, just add enough to make the paint more manageable. If you apply a layer of medium to the surface of a dried layer of paint before you paint over it, make it a very thin layer.
  • It is best not to add metallic driers to make the paint dry more quickly. If you do add them, I think that lead napthenate is best. Add a tiny amount (like one drop from a toothpick) to a penny-sized blob of paint on your palette. Add driers only to the slow-drying pigments on your palette.
  • In my opinion, it has not yet been demonstrated whether alkyd painting mediums (Liquin, Galkyd, Neo-Meglip, and so on) are sufficiently permanent. They are probably fine for single layer, direct painting. I’ve heard a couple of complaints about delamination in multi-layered paintings that may be due to use of alkyds. Some alkyd mediums can also yellow quite a bit. Personally, I don’t see any reason to paint with anything that smells like that.
  • If you add solvents and oils to your paint, and you work in layers, it’s best to follow the fat over lean rule. That just means that no layer should have less oil in it than the layer beneath it. So be careful about how you use mediums and avoid painting large areas of lean paints (without much oil in them) like manganese violet over large areas of fat paints (with a lot of oil in them) like ivory black. The fat over lean rule is especially important if you paint in thick layers. In thin layers, it’s still a good idea, but less crucial.
  • Varnish the painting after it is dry. By dry, I mean three months to a year after completion, depending on how thick the paint is.
Few painters (including me) work according to these guidelines all the time, and yet their paintings don’t generally fall apart rapidly. Oil painting is fairly forgiving, so long as you respect your materials and stay within a reasonable zone of craftstmanship. So long as you do that, there isn’t any reason to worry about archival permanence unless the voices in your head are very insistent that you are going to be the next Michelangelo.

Personally, I doubt that the art conservation robots in the Louvre in the year 2306 will curse my name because I used sub-standard methods requiring them to spend an extra 324.663 seconds fixing one of my paintings. But that would be really cool.

Update 10/23/06: One other point regarding how to construct paintings that will last. If you paint in multiple layers, make sure that each layer adheres to the one below it. A paint layer that is smooth and shiny is not a good surface for painting over, because the next layer of paint has no mechanical tooth to adhere to. You may want to scuff up the surface with a green kitchen scrubee pad or, if you prefer, wet sand. If you use a medium that contains a balsam such as Venice turpentine or Canada balsam, the paint will adhere better to the previous layer.

15 October 2006

Le Café Marly

One of the restaurants at the Louvre.

14 October 2006

Practical Color Mixing 2: Hue

In the previous post in this series on color mixing, I talked about how to mix the right value. Here I’ll talk about hue.

Identifying hue

Before you can mix the right hue, you need to figure out what hue you want to mix. That’s often kind of hard, especially with the dull, low-chroma colors that predominate in most of the visual world. Look around you. What color is the wall? A yellow-green? Or is it more of a middle yellow? How about the cable leading to your monitor? Is it black, or some very dark greyish color? If so, is it a warm dark grey or a cool dark grey? What about the shadow falling on the floor from your desk? You get the picture.

When I’m standing around waiting for something I find myself trying to identify the color of various things around me. And not just the color of the thing (the “local color”) but the color of the shadow side, the light side, and so on. I think about value and chroma as well as hue, but the hue is often the hardest to figure out. As with any other attribute of color, it’s best to think in terms of comparison of one color with others around it. Once I think I know what the color is, I consider how I would mix it. That seems like a pretty geeky thing to do (and it is) but it’s a useful exercise. We think we know what color things are, but while it’s easy to say that the sky is blue, it’s a bit more of a challenge to determine that the part of the sky beyond those clouds is a green-blue, while the sky in between the clouds is slightly purple.


So once you know what hue you want, how do you get it? Well, if the hue is pretty close to a paint that’s already on your palette, you might be able to just tint it in one direction or another. Say you want a violet blue and you have ultramarine on your palette. Without having to think about color wheels or anything complex like that, you could simply add just a bit of a more purple color, such as dioxazine violet. That may well get you where you want to go, simply and easily. In doing this, the thing to realize is that any given paint can only go in two directions from where the hue is right now—either clockwise or counterclockwise on the color circle. Ultramarine blue can be made more violet or more green. That’s it. Cadmium orange can be made more yellow or more red. If you’re just trying to nudge the hue around a little bit, all you have to decide is which direction to go and select a color next door on the color wheel to move it in that direction.

Using a mixing color wheel

The nudging strategy is great for small adjustments, but it starts to fall apart when you need to make a hue that isn’t close to one of the paints you already have. At that point, it’s useful to go back to the concept of a color mixing wheel. As I’ve pointed out previously, a color mixing wheel does not provide a precise guide to what you will get with any two pigment mixtures. Individual pigments are simply too idiosyncratic in their mixing properties to allow any kind of absolute prediction of how they will behave when mixed. But a color mixing wheel will help you to get into the approximate ballpark, at which point you will be close enough to use the nudging aproach described above.

Steven Quiller sells a useful color mixing wheel. Bruce MacEvoy at Handprint has a somewhat different one that you can print out for free (it's designed for watercolor, but I have found it to be reasonably useful for other media as well).

Say you need to mix a yellowish green, but don’t have anything close to that on your palette. If you look at a color mixing wheel, the two colors on either side of green are blue and yellow. As we all know, you can mix a green from blue and yellow, and if you adjust the proportions correctly, you can pretty easily get a yellowish green. If you have any set of paints that are selected to fall at reasonable intervals across the color wheel (at least a cyan, magenta, and yellow), you can mix any desired hue using two or three paints.

A traditional color wheel is set up so that all of the colors on the outside are as high in chroma as that hue goes (without regard to value). On the inside are less chromatic colors, arranged so that the closer to the center they are, the lower the chroma. The basic mixing procedure goes like this: (1) identify a point within the color mixing wheel that represents the desired hue and chroma; (2) look for one or more lines between two paints that pass through (or near) the color you are trying to match; and (3) consider whether a third paint (typically one on the opposite side of the wheel from the desired color) might be needed to adjust the chroma downward. If the paints have equal tinting strength, you can figure out approximately how much of each paint you will use, based on where the desired color falls on the line between the two paints being mixed. If one paint is stronger, you’ll need to adjust accordingly to account for that. As a general rule, put out some of the weaker paint and add the stronger paint to it. Alternately, put out the paint you will use the largest amount of and add the other paint to it. Add in small increments at a time—I find myself overshooting frequently if I’m not careful.

With oil paint, it's best to mix with a palette knife rather than a brush. Once you're used to it, the knife is faster because you can clean it so quickly, and your paint piles don't become contaminated with other pigments.

Coordinating hue and chroma

Notice that if you draw a straight line between any two colors on the outside the wheel, every point on the line represents a lower chroma than those two colors. So mixing tends to reduce chroma. As a general rule, any mixture is duller than the brighter of the two paints being mixed, and often duller than either one. There are a very few exceptions (some warm pigments become a little more chromatic when mixed with each other and some cool pigments become more chromatic when mixed with a small amount of white), but chroma reduction is the usual effect of paint mixing.

Often, it’s useful to have mixing reduce chroma, because in realist painting you are frequently trying to mix a color that is duller than the tube paints you have available. If you’re trying to mix a flesh tone with bright cadmium colors, for example, any reduction in chroma is welcome (you've probably seen bad amateur portraits with bright orange flesh tones). But there are times when you are trying to mix a high-chroma color, and in that case the chroma reduction from mixing can be frustrating. Because of this effect, it’s often a bad idea to just muck around with paint, hoping to get close to the color you’re looking for. Every paint you add to the mix cuts the chroma down, so after awhile you are just mixing paint into a sort of nondescript grayish color—i.e., you're mixing "mud."

It’s is much better to decide what color you want, choose two or three paints that you will use to get that color, and try to stick with those. Minor nudging with other paints is OK, but if the mixture goes radically in a direction you didn’t expect, don’t keep throwing additional paints in, hoping you’ll eventually get to your desired color. Once you have mud, just scrape it off your palette (or use it as the basis of some nondescript color you need elsewhere) and start over. Take a step back and think again about what you’re trying to accomplish and how you’re going to get there.

As noted above, many pigments fail to follow a straight line on the color wheel when mixed. In particular, some paint mixtures follow a circular mixing line. That means that, while the wheel predicts the correct hue, the chroma is higher than expected (this is particularly common with greens). In that case, go ahead and mix the hue you want, then tone it down (I’ll talk about adjusting chroma in the next post in this series).

Coordinating hue and value

We’ve talked about getting to the right hue and chroma, but what about value? It would be easier to make this color mixing thing work if there were only two parameters to worry about, and many color mixing books kind of pretend that’s the case. When I talked in my last post about getting the value right when mixing, I suggested that the first thing you do when trying to make any particular color was to first mix the colors you are working with to the correct value. You can then mix them together and get the hue and value you are looking for—because the paints are already at the correct value, you don't have to think about that factor any more, greatly simplifying the problem you're trying to solve. Mixing value first usually works, except when you can't get the right chroma and value because the white paint is pulling the chroma down too far. I’ll talk about strategies for dealing with that problem when I discuss chroma in the next post. Under most circumstances, the “mix the value first” rule makes color mixing much easier to control.

Warmth and coolth

The idea of warm and cool colors has many implications for composition that don’t belong in a discussion of color mixing. Warm colors are generally thought to include red, yellow red, and yellow, while cool colors are thought to include blue green, blue, and purple blue. (I’m using Munsell hue terminology here.) In between colors include green, green yellow, purple, and red purple (some people would label green and purple as warm and green yellow and red purple as cool). There are some aspects of the warm/cool division that are useful to include in a discussion of color mixing.

If you reduce the chroma of a warm color, it appears less warm (raw umber is less warm than cadmium orange); if you reduce the chroma of a cool color, it appears less cool. The most chromatic warm colors are much higher in chroma than the most chromatic cool colors. Compare, for example, cadmium yellow light (a high chroma yellow) and pthalo blue (a high chroma blue). The yellow is much higher in chroma than the blue. Not only that, the yellow is also much lighter in value than the blue. Warm pigments can be very light (high in value) at high chromas. Cool pigments are much darker at their highest chroma. Adding a lot of white to a cool pigment, to bring the value up near to that of cadmium yellow, decreases the chroma still further. You can’t have a high chroma, high value cool color—the physics of light don’t allow it. These kinds of differences are why the Munsell color space is shaped like a bumpy, irregular cylinder.

Because of this effect, high chroma warm colors have more overall punch than high chroma cool colors. Many art books tell you that warm colors advance and cool colors recede. That’s wrong, although it has a germ of truth. The eye looks for contrast. In most paintings, chromatic colors have more contrast with their surroundings. Higher value colors also have more contrast. Because warm colors are more chromatic and higher in value, they have more contrast, so they jump forward. If you drop the chroma and value of warm pigments to match those of cool pigments, they become brownish and don’t have any extra punch at all.

The warm/cool contrast can also be useful when you're trying to figure out what hue something has. It can be easier to ask whether a hue is warmer or cooler than a color near it than to try to figure out its hue directly. Any hue can be shifted either clockwise or counter clockwise on the hue circle. You can think of this as shifting warmer or shifting cooler. For example, a yellow can shift toward green (cooler) or toward red (warmer). That is not to say that red is warmer than yellow (different people have different opinions on that issue), but that shifting toward red is shifting away from a cooler color (green is definitely cooler than red), so it’s useful to think of that as “warmer” in this context. Similarly, a purple can be shifted toward red (warm) or toward blue (cool). As you are painting, you can think in terms of these comparisons. Is the hue on the light side of an object warmer or cooler than the hue on the shadow side? That comparison is an easier task than determining the absolute hue of the light side.

12 October 2006

Heraldic contrast

Traditional European heraldry (coats of arms) has a lot of rules regarding how designs can be constructed. One of the fundamental rules is this: no metal on metal or color on color. There are two metals: gold (represented as yellow) and silver (represented as white). All the other hues that can be used are "colors." The rule is that colors can't be placed next to other colors, only metals. Metals can't be placed next to other metals, only colors.

This may seem like an antiquated piece of trivia, useful only to those who are desperate to be the fifth cousin twice removed of the Duke of Cornwall or somesuch, until you look at street signs. In the U.S. (and those parts of Europe and Canada I've visited) almost all street signs follow the heraldic metal on color and color on metal rule. A U.S. stop sign is a metal (white/argent) on a color (red/gules). Highway direction signs also (white/argent on green/vert). Speed limit signs? Black/sable on white/argent. Those few signs that don't follow the heraldic convention, such as construction signs with black text on an orange field, are much less noticeable than the vast majority that do.

What's going on here? Heraldry was originally designed so that painted shields and banners would be clearly visible at a great distance on the battlefield. For that to happen, you need to have a lot of contrast. As it turns out, with the pigments they had available, white and yellow had the best contrast against the other pigments. Thus the avoidance of color contrasted against color or metal contrasted against metal. Even with modern pigments and special reflective surface treatments, the rule pretty much holds up, so sign designers follow it even if they don't know where it came from. Next time you're on the road, try to find signs that break this rule. If you do find one, note that the contrast is poor compared to signs that follow the rule. Same with signs on buildings.

This rule can be useful when you are designing a color scheme and want to highlight a focal area of a painting. If you follow this old heraldic rule, you will have all the contrast you could need.

08 October 2006

Battle scenes

are uncommon in Renaissance panel painting. Here is one exception, part of a series of three huge panels, done mostly in egg tempera, commemorating a minor Italian battle (the Battle of San Romano) by Paolo Uccelo. This one is in the National Gallery in London. I've seen it's sister painting in the Louvre. One impressive aspect is the gilding. All of the armor is done in silver leaf, punched and scribed with three dimensional patterns of mail and plate, then glazed with thin layers of asphaltum ground in oil. Very impressive.

05 October 2006

04 October 2006

Practical Color Mixing 1: Value

OK, let's recap. In my first post on color, I concluded that the standard three-primary color wheel is not useful for learning about using and mixing color. In the second post, I briefly reviewed the Munsell color system as a means for describing color and for identifying visual compliments. In the third post, I talked about the difficulty of developing a simple system that could adequately provide a method for artists to mix color.

What we're left with is coming up with color-mixing strategies that can allow a painter to get the job done without too much frustration. Unfortunately, this is complicated stuff, with lots of exceptions and special considerations. It doesn’t make as neat and easily-explained an idea as a simple color wheel. There’s just kind of a lot of stuff that’s important (or at least useful) to know about. And because all of the stuff interrelates, it’s hard to split it up into easily-digestible topics. But that’s what I’ll try to do. In this post, I’ll discuss value. Later, I’ll talk about controlling hue and chroma. Of course, since they are so closely interrelated, in each discussion of one of the three components I’ll also have to talk about the other two.


If you’re having trouble mixing a color with exactly the right hue, chroma, and value, concentrate on at least getting the value right. The brain weighs value far more strongly than it does hue and chroma. There are any number of paintings out there with wierd hues and no consistent use of chroma, but they work because the values work. (That’s not to say that hue and chroma aren’t important—they are—but to emphasize that value is the most important of the three for a beginning or intermediate painter to concentrate on.)

Value should be considered at two levels in realist painting. First, you need to consider the value structure of the painting as a whole. Do you want most of the painting to be light (a high key painting), most of it to be dark (a low key painting), or for there to be some kind of balance across a wide value rang (a full key painting). What is the range from darkest dark to lightest light? It’s useful to establish that range early, because every object will be rendered in relation to that key.

Overall Value Range

As you think about this, you need to be aware that the value range of paint has only a small portion of the value range of human vision. Consider a painting of a sunset. That sun is much, much brighter than your highest-value white (which you’ll need to tone down in order to get the right hue and chroma). In order to give an impression of that brightness, you will need to make the rest of the scene quite a bit darker than you otherwise might. You are choosing a value scheme that represents the value relationships most important to the composition. That means that you won’t be able to have much contrast in the darks, because their value range is compressed in order to emphasize that very strong light.

On the dark side as well, the blackest black on your palette reflects a lot more light than a really dark shadow does. So there are times when you need to compress the lights in order to show a full range of contrasts in the shadows. Of the painting media, by the way, oil paint has the widest value range, particularly in terms of really dark darks. So it's easier to create believable three-dimensional form with oil paint, and that's one reason why it's so popular.

Therefore, in making decisions about the key of a painting, you need to accept that you are necessarily working with a limited value range, and you need to make intelligent choices about how you use it. There is no such thing as “paint what you see” in this calculation. Artists often manipulate the value range to achieve a specific effect. Rembrandt and Carravagio, for example, both painted low key paintings. But more than that, they made the darks very, very dark, the midtones quite dark, and the lights very light. That creates a marvelous dramatic effect, but there’s nothing realistic about it. It is worth looking at a lot of paintings and thinking about what choices the artists made regarding value range, because those are the same choices youre going to have to make every time you paint. Trying to go for a middle road, in which the lights are fairly light, the midtones are fairly medium, and the darks are fairly dark, is not always the best choice, because it isn’t very interesting and because it’s often not the right way to represent the emotional content of the scene. Don’t think that hue and chroma are the primary determinents of the way your painting feels, because often the real money is where the value is.

Light and Shadow

Within the overall value scheme of your painting, you’ll need to consider value as you work on each object or passage. Often, it’s good to think in terms of two, three, or (maximum) four values in the mass areas each object depicted. On the Munsell value scale from 0 to 10, for example, you might paint the shadow side of a house at value 3 and the light side at value 4. A head might be represented at value 5.5 in the upper lights, value 5 in the darker lights (sometimes called the “halftones”), value 3.5 around the terminator (the shadow edge), and value 4.5 in the reflected lights. You could mix these tones on your palette before you start, lay them down in the appropriate areas, then blend as desired. You could then add highlights at value 6.5 and dark accents at value 3. Doing it in such a pre-planned way can be much easier than figuring it out as you go.

To get these values right, it’s much easier to think about relationships than it is to think about matching the actual value of the object you’re trying to paint. So if the overall key of your painting sets the value of the light side of a cube at value 7, the thing to do is to observe and think about how much darker the shadow side of that cube is. What value in paint best reflects the value relationship you are observing in real life? Is it a 5? 4.5? It takes a lot of practice to get these kinds of relationships right across an object, and then to keep those relationships right across many objects in a single painting (many beginner paintings seem to be keyed differently in different parts of the picture).

One useful exercise is to do a series of paintings in a single hue. If you do it in shades of grey, it is called a grissaile (pronounced gree-zai). A 50/50 mixture of black and raw umber is a good base tone for a grissaile that you can mix with different amounts of white to get the desired value. Doing a series of grissaile studies can help develop your ability to judge and paint value. For each stroke of paint you put down, think about whether it should be darker or lighter than the paint surrounding it, and by how much. Over time, you’ll achieve a much greater sensitivity to value.

One of the typical mistakes that beginners make is to focus on hue and chroma at the expense of value. They might be trying to paint a figure, for example, and mix up some “flesh tone” (or they might have a tube labelled “flesh”). They proceed to paint the whole figure that color, then timidly throw in a slightly darker tone for shadows and edges. The figure looks flat and unconvincing. If you are going to concentrate on a particular subject, and you want it to look dimensional, you need a fairly wide value range within that subject. If that means compressing the lights or darks so that background elements have less contrast, that’s OK. If you’re painting a figure, the shadows on that figure should be a significantly darker than the lights. It is only with a wide dynamic range that you can create the illusion of three dimensional form. I remember when I was first taking painting classes, my teacher would look at my figures, sit down with my palette, and make the darks a good two value steps darker than I had made them (and I thought they were pretty dark). It was a painful experience, but it showed me how to create a successful illusion of form.

White and Black

So how do you make paint darker or lighter? You can, of course, simply add black to darken and white to lighten. If you’re only concerned with value, that will always work. But both of those colors will, under many circumstances, distort hue and chroma.

White, of course, is a critical mixing color; it is the dominant pigment in many paintings. As you add white, however, the color tends to drop in chroma and will often shift hue. Hue shifting isn't too hard to deal with; you can add a touch of a warm or cool color (usually warm) to correct the hue. Dulled chroma is harder to fix (that's one reason why white paint isn't used in traditional watercolor technique). One strategy is to avoid titanium white when you don’t need a really bright opaque white. Both flake white and zinc white are less overpowering and have less tendency to kill the chroma in mixing. Zinc, especially, is good for this purpose. Another strategy is to use glazing, rather than mixing, to adjust hue and chroma. But it is the case that some high-chroma colors are very hard to approximate with paint. So when you are deciding on a value scheme for your painting, one important consideration is whether you will need to showcase any very high chroma colors. If that is the case, you may need to adjust the key of the painting so that those intense colors are the right value without having to lighten or darken them much. So, for example, you might use a higher-chroma, slightly less bright light and it will read as very bright in contrast to the relatively darker, duller colors elsewhere.

I think of black as being one of the less important colors on my palette. Some painters never use it, claiming that is is a deadening color, that there is no black in nature, and that excessive use of it makes your painting look like it has a hole in it. That’s hogwash. Take a look at paintings by guys like Leonardo da Vinci, Diego Velazquez, or Carravagio. They relied heavily on black. Can you really say that their paintings would have been vastly better if they had only known that some modern painters think black makes a painting look damaged? “That Leonardo guy, if only he’d known to avoid black, he might have made a name for himself!” Yeah, right. And don't think they didn't know how to darken colors without black, because they certainly did. They chose to use black because it was the color that worked best for what they were trying to accomplish.

That being said, it is certainly true that mixing a color with black will reduce chroma, and that black is best used with care. Sometimes, strong chroma reduction is exactly the effect you are looking for, so that’s when to use black. Black also causes color shifts. Mix black with a bright yellow (such as cadmium yellow light). Do you get dark yellow? No, you get an olive green (which can be quite useful). Under many circumstances, black acts like a very, very dark blue. Black is often best used the way it was usually used in the 15th century, to darken (and reduce the chroma of) earth colors and other low-chroma colors. I don’t usually use black as a dark dark; instead I tend to make a really dark mixture (such as ultramarine and burnt sienna or viridian and pyrol ruby) and use that. In part, that’s because in oil paint pure black takes a long time to dry and is a relatively “fat” color. But it’s also because pure black is a bit dead compared to mixtures.

Controlling Value

So, if we need to be careful with white for lightening and if black is of limited use in making colors darker, how do we manage value? Carefully.

Take a look at a color wheel, particularly the colors on the outside of the wheel (the highest chroma pigments in each hue). You’ll notice that warm colors. like yellow, are quite high in value. By contrast, cool colors like purple are relatively dark. So your value mixing strategy will need to depend, to some degree, on what part of the color wheel you’re working with.

For example, how do you make a dark yellow? Do you look on the other side of the color wheel, find that a violet is the compliment of yellow, and mix that in? That doesn’t work very well, because many warm colors don’t have a true mixing compliment (they don’t mix to a neutral grey). So violet added to yellow produces a severe color shift away from the yellow hue. Fortunately, there are dark yellows already available—browns are basically dark yellows. So, as artists have been doing for many centuries, you don’t try to mix a dark yellow. If you want to depict a gradation from a light yellow to a dark yellow, you blend in a series of brown colors: cadmium yellow to yellow ochre to raw sienna to burnt umber, for example. Of course, if you don’t like earth colors, it’s perfectly possible to mix similar ones. Some artists like to use a very limited palette of only cyan, magenta, and yellow; with those and white, you can mix colors very similar to yellow ochre, raw sienna, and burnt umber. Personally, I find it much easier to simply use the earths.

How do you make a dark purple? Easy. Purple pigments are already dark, so all you have to do is adjust the chroma (if needed). If you need it even darker, you can mix it with dark transparent colors that are near to it on the color wheel (a transparent dark blue like Prussian blue and a transparent dark red like pyrol ruby, for example). You could even add a touch of black. How do you make a light purple? You’re probably going to have to use some white. If you need a light, high-chroma purple, you may have some trouble, because the white is going to knock the chroma down. So you may need to start with the highest-chroma purple you can find (such as dioxazine violet), so the final result remains relatively chromatic. As noted earlier, you might also want to use zinc, or a zinc-flake mix (a big glob of zinc by itself dries a bit brittle and that can affect the stability of the paint film). If you need a light, low-chroma purple, on the other hand, then mixing with white, then making slight adjustments to hue by mixing in small amounts of other colors, will probably work just fine.

It is also the case that, for many cooler colors like blues and violets, mixing with a small amount of white will increase the chroma. For example, in oil paint, ultramarine blue with a bit of white added is more chromatic than plain ultramarine. But adding a lot of white decreases the chroma. Warmer colors are tend to be at their maximum chroma straight out of the tube.

The easiest overall mixing situation is when you are trying to reduce chroma at the same time you are adjusting value. If the color has a mixing compliment, then, typically, that color will reduce value and chroma at the same time.

Mix the Value First

When mixing two colors, it’s often a good strategy to first get both of those colors to the intended final value. Then, when you mix them together, it’s much easier to judge hue and chroma. If you have a clear idea of what colors you’ll be working with in a painting session, it can be useful to mix up a series of paint strings. One string is a single hue/chroma along a series of values. So, for example, a useful flesh tone is a neutral mixture of cadmium green and cadmium red. You might create that neutral brownish color on your palette, then mix in different amounts of white to make a series of gradations from very light brown to the base cad green/cad red mixture. You could then make darker tones along the same string by adding different amounts of raw umber. Now you have a string of one hue and chroma, but different values. Another string might be based on yellow ochre as a base tone, mixed with different amounts of white to make lighter tones and different amounts of raw sienna and burnt umber for darker tones. As you paint, if you need a hue that is in between the two strings, it’s easy to mix paint from each string at the same value to get the right hue.

Working with a set of pre-mixed values is called a set palette. Some artists take it to extremes, always mixing up a pre-set group of value strings before staring to paint. If you do this habitually, it helps to mix a lot of each paint value in advance and put them into tubes—that way, you don’t spend half of each painting session mixing strings of paint. I don’t do that, but I do typically mix two or three strings at the beginning of a painting session and work from them. You can buy sets of pre-mixed neutral greys at different values from several companies; the greyscale set from Studio Products even comes graded according to Munsell values. I’ll come back to the usefulness of neutral greys later on when I talk about managing chroma.

03 October 2006

Madonna with the Child and Two Angels

by Fra Fillipo Lippi, tempera on panel, 95 x 62 cm (37 x 24" ). This gorgeous and delicate painting was done in 1465. Lippi was an interesting character, a Friar who had an affair with a nun. They were both allowed to resign from their respective orders. She bore him a son (Fillipino, who also became a painter) and a daughter (who was not artistically inclined, so far as I am aware).

For all that, his style is beautiful and, I think, possessed of a profound sense of the divine. He probably taught the more famous Sandro Botticelli, whose style is clearly influenced by him.

01 October 2006

A hottie

This painting of a lovely lady is attributed to Raffaello Sanzio (Raphael). La Fornarina, oil on panel, 1518-19.

The style of this painting is not as delicate as much of Raphael's other work. For that reason, there is speculation that this painting might have been done by a follower of Raphael, or perhaps it is a copy by a skilled assistant in his workshop of a lost original.

She's a babe nonetheless.


Among oil painters, there seems to be a common misconception that glazing is some kind of mystical technique that only a few can master. The basic process is, however, very simple. Glazing is putting one layer of paint over another so that you can see the underlayer through the upper layer of paint. Glazing is a form of indirect painting, which just means that you are painting with more than one layer, allowing previous layers to dry before you add more paint on top.

Glazing can be used for a number of purposes. As I noted my post comparing the glazing methods of Italian and Netherlandish Renaissance painters, glazing can be used to create optical color mixtures (a blue glazed over a yellow makes a green) or to create modeling effects (thicker layers of transparent paint are darker, so you can adjust value by adjusting the thickness of the paint). Some artists glaze over a whole painting to unify the overall tone. Others will glaze specific parts of the painting. One method is to do an initial monotone underpainting (in shades of grey, for example) then apply color over it. This simplifies the process of painting by first tackling pure value, then working out color. Some modern portrait painters will do an initial painting of flesh in shades of green (they incorrectly call this a "verdaccio"). They then glaze with reds and oranges (complimentaries and near-complimentaries to green), providing the flesh tones with a sense of vitality that is difficult to achieve with direct painting. Glazing can also be useful for maintaining chroma in light colors. Mixing with a lot of white will seriously reduce the chroma of most colors, resulting in a look often described as "chalky." If you glaze the same color over white, however, you can achieve an optical effect that is high in value, with more chroma that you could get by mixing that color with white.

Because a glaze darkens what it covers (unless its a scumble—see below), it is best to do the underpainting lighter than the intended final effect. If you are going to glaze, it's important for the underpainting to have as smooth a surface as possible. That's because irregularities will trap excess amounts of paint in the glaze layer, creating wierd little spots of darker paint. So, before the paint dries, it's a good idea to go over it very lightly with a soft dry brush, looking for lumps and gently brushing them down. After the underpainting has dried thoroughly, you may want to wet sand to create as smooth a surface as possible.

In selecting paint colors to glaze with, it is useful to distinguish among opaque colors (like cadmium yellow), semi-transparent colors (like ultramarine blue), and transparent colors (like alizarin crimson). While any of these colors can be used for glazing, transparent and semi-transparent colors are darker when they are put on more thickly. Opaque colors can be used for glazing, but only when they are applied in a thin layer. A thick layer of an opaque color is not a glaze, because you can't see the underpainting through it.

Many oil painters think that the best way to glaze is to dilute the paint with an oil-resin medium to a watery or syrupy consistency (this is what a lot of art instruction manuals tell you to do). The paint becomes less opaque because the pigment particles are separated by a larger than normal amount of transparent vehicle. This type of glaze is called a dilution glaze. In my (deeply humble) opinion, it's the wrong way to glaze. It's bad technique for (at least) three reasons: (1) all of that extra resin and oil will darken and yellow over time, ruining the effect; (2) dilution glazes tend to create a sort of "tinted photograph" effect that doesn't have the solidity a painter is usually trying to depict; and (3) the documentation I've found on historical glazing techniques suggests that only small amounts of resin are detected in glazing layers in Renaissance Netherlandish paintings, which I consider to be the gold standard in glazing for both beauty and longevity.

A better method is called a reduction glaze. This approach involves adjusting the transparency of the paint by adjusting the thickness of the paint layer. While you can do a reduction glaze with nothing but pure oil paint, it helps to first lubricate the surface by applying a very thin layer of a slippery medium. My preferred glazing medium is a 50/50 mixture of black oil (linseed cooked with lead) and Venice turpentine (if you don't like to use substances containing lead, linseed oil will work almost as well). Studio Products also sells an excellent glazing medium. Put a drop of medium on the surface, rubbing it in with your fingers to spread it as far as possible. This way, you can cover a large area with just a few drops of medium. In addition to putting some on the surface, you can also put just a tiny bit of medium in your paint, but I don't usually find that necessary.

Mix up the color you want to glaze with. Apply it thickly and evenly to the desired area with a brush. It will look like a horribe mess at this stage. You will now reduce the thickness of the glaze to the desired opacity and value. Do this by dabbing with a soft brush, smearing with your fingers, rubbing with a cloth or sponge, or whatever works to adjust the glaze to achieve the desired effect. With a little practice, a reduction glaze is really pretty easy. You can get nice gradations in color and value by creating a gradation from thin to thick. Or you can create gradations from one color to another. Once you have the glaze spread to the right thickness, you can, if you like, paint into it with other colors. For example, you can apply light highlights into a wet glaze and then blend it in. If desired, you can let your glaze dry and then add one or more additional layers of glazing. For example, you can get really intense, chromatic darks by glazing with multiple layers of transparent paint.

When mixing colors for a glaze, it is sometimes helpful to add a small amount of white to your mixtures. This provides a greater sense of solidity. If you glaze with very light colors containing a lot of white, it is usually called a scumble. Titanium white, being very opaque, can be a bad choice for scumbling. Flake white and zinc white are much easier to create transparency effects with. A very white, hazy glaze is called a velatura ("veil"). A velatura can be a great way to depict transparent smoke, haze, or fog.