For some time now I’ve been preoccupied with value.
The more I work with it, the more I become convinced that a well observed balance of tones (values) is the key to creating a feeling of light, and by extension form. Frank Reilly reckoned that tones (well, being from the US he called them values too) are 80% of a good picture. Who am I to disagree?
But I’ve been wrestling with a basic problem in painting for as long as I’ve been interested in tone: How to deal with the fact that the available value range from light to dark is much more narrow in paint than it is in nature?
Today’s exercises have finally convinced me that I’ve got the answer to that problem – or at least, one answer. This post is going to get a bit technical, and there’ll be numbers involved. For that I apologise, but there’s no avoiding it.
Regular visitors will be used to me harping on about tone by now. The series of still life drawings were started to try to gain some insight into this problem. After a few of those, I came to the conclusion that compressing the tonal range whilst preserving the ratios between the tones was an approach that worked quite well. The blue and red cast paintings showed me that it was possible to compress the tonal range in a painting dramatically, yet still create form and light. Today I worked on a few little exercises which finally proved to me, beyond all doubt, the validity of this approach.
First, some background on the theory behind these exercises:
This collection of painted wooden cubes, styrofoam spheres and MDF dog tags is what I call my Munsell neutral value set.
They’ve been created for one reason – to help me get a better appreciation of tone. I’ll be using them in a series of exercises over the coming months, all designed to investigate how light behaves when it hits a form, and how best to translate that into a painting. I’ll move onto colour later, I’m one of those boring methodical people that likes to follow a logically ordered path, so I’m dealing with tone first. If I can crack this, and Mr. Reilly is right about his 80%, then colour should be substantially easier to handle.
I won’t go into any detail here about Munsell, since it’s been done very well elsewhere, but if it’s new to you, the Wikipedia Munsell page looks pretty comprehensive. Frank Reilly developed the Munsell system into a practical tool for painting. I first came across Munsell through Graydon Parrish, a superlative contemporary academic realist painter from the US. He’s developed and expanded on Reilly’s method and has evolved his own approach to colour based on it. He was himself taught by one of Reilly’s students, so he knows what he’s about when it comes to this stuff.
Enough background. I recently got hold of the Munsell Student book. This is an overview of the Munsell system and it’s application to painting. It includes colour chips for some of the Munsell colours, among them a Munsell neutral value scale. The cubes, spheres and tags in the photo above have been painted to match the values of this scale.
The Three Cubes Exercise
I’m calling this exercise the three cubes exercise because it involves three cubes. No point complicating matters.I’ve taken one white cube, one black one, and one a mid grey (a value 5 in Munsell). The first stage was to paint each one in isolation, starting with the value 5 one. This should be the easiest, since I know I can hit all the values I see within the available range of my oil paint.
Like all these exercises today, this cube was painted sitting on a grey cloth (local about a Munsell 5.5 I think) in my shadow box. All the paintings have been done in natural light from the window. The light did change a little through the day, but that didn’t really matter since it’s the relationships and not the actual values which count here.
One of the things I had to decide today was how I was going to use the value tags to help me judge the tones. Since I want to get as close as I can to the values I see with this grey cube, I used them by holding them up in front of the cube and comparing the values.
For the purposes of this exercise, I was careful to hold the tags at the same angle to the light as the panelI’m painting on. This is my value 8 chip, so the plane of the value 5 cube facing the light appears as a value 8, pretty much. Actually, I think I adjusted it somewhat at the end, but this was the starting point.
Obviously, if I angle the tag towards the light, it will appear to be a lighter value, and vice versa. That will throw things off. I want to get the values I see onto my panel accurately for his cube, so by keeping the light on my tag and my panel the same I should be able to get close.
This approach is going to give me trouble later, especially with the white cube, but we’ll get to that shortly.
This is the same process for the value of the top plane. This time I’ve got a value 6 or thereabouts, which is one step up from the ‘local’ tone of the cube.
I did this five times, Once for each plane of the cube,once for the background cloth, and once for the cast shadow. What I ended up with is four separate values, which should be all I need to paint this cube as I see it.
Since the value of the cloth and the local value of the cube are very similar, the value of the cast shadow will be very close to the value of the plane of the cube in shadow. I’m thinking about Loomis and the truths of the form principle whilst I’m doing this. This one, whilst obvious, is certainly relevant here:
“The lightest areas of the form will be within those planes lying most nearly at right angles to the light. The half-tone planes will be those obliquely situated to the direction of the light. The shadow planes will be those planes lying in or beyond the direction of the light so that the light of the original source cannot reach them. The cast shadows are the results of the light having been intercepted, and the shape of such intercepting forms will be projected onto other planes.
That’s what I’ve got here, light plane, half tone (the top of the cube), the shadow plane and the cast shadow. And here’s my Munsell values, nice and simple. I’d like to add another truth, which doubtless Loomis thought was too obvious to include,but here it is: The value of a cast shadow is determined by the value of the object it is cast onto, not the one which casts it. That’ll become more relevant later.
Next job is to mix up some of each value I’m going to need. Thankfully, I only need four so it won’t take long.
I should point out here that these are Munsell neutral greys, which means adding something to take away the bluish cast that black and white alone will give to the grey. The Munsell student book recommends using burnt sienna for this, which is what I used. It does the job well. The black and white were ivory black and titanium white.
I’m finding that the more often I mix up values to match my tags, the better I get at it. At first I found it very hard and it took forever just to match a single value. But I’m pretty sure that one of the many advantages of doing these exercises will be a more sensitive and accurate eye when it comes to judging values in nature. I’ll be very bloody disappointed if that doesn’t happen.
So here’s the results of the first exercise. I’ve adjusted some of the tones, reducing the steps in the light and half tone planes mostly. That part was just done by eye.
So far, so good. The painting of the cube turned out pretty close to what I saw, and seemed to me to be pretty consistently following the form principle. The cast shadow being the same value as the shadow plane of the cube helps the eye to read the cloth and cube as pretty much the same value (in local terms) and helps the halftone and light planes to work I think.
Part two – the Black Cube
This cube has been painted black, a value 1 in Munsell. Well, it would be if you could get black (that is, no light) in paint, which you can’t. It’s actually about a 1.5 I think. That presents me with my first problem -I can’t hit the black of the shadow plane on this cube with paint.
Unfortunately I haven’t managed to get such a good picture of this one, but I’ve got 2, as before, for the cast shadow. The shadow plane of the cube is ivory black, as close as I can get to value 1. The top plane is a 2 and the light plane is a 4.
The cloth has for some reason gone up to a 7.5 in this painting, I can’t remember why I did that now, the light got stronger I think. Somehow this cube doesn’t work as well for me as the first one. There must be something off about the relationships I think. But overall, I think it works as a black cube on a grey cloth.
Part three – the White Cube
I was expecting this one to be hard, particularly using my method of holding up the value tags at the same angle to the light as my panel. If I hold up my lightest tag, the white one, it’s always going to appear darker than the light plane of the white cube unless I hold it at the same angle to the light as that plane of the cube.
What to do? I can hold all my tags up at the angle of the light plane, but that will throw out my darks. I already know, from painting the black cube, that my black facing the light appears as a value 4. But the cast shadow I know is a value 2. This simple little white cube is out of my available range. I could drop all the other values in the picture to compensate for my white appearing as a light grey, but I’ve tried that before and it just makes for a darker picture. It doesn’t make for a more convincing picture, at least, not in my experience.
This really comes to the heart of what I wanted to figure out with these exercises. I know that I need to compress the tones in order to get this painting to work as a white cube, and so far I’ve been doing that by eye. But I’ve often been less than convinced by the results. In principle, I know that I want to preserve the ratios between the tones, so I need to work out how much I compress the steps of the value range by to get a convincing picture of this cube.
The solution, at least the one I used today, was to work out the difference in steps of the Munsell scale between each tone block. Once I knew the total range of the steps I could see, I could translate that to the narrower range I can hit with paint. The tags were invaluable here.
First, I needed to find how many steps the top plane (my half tone) was down from the light plane, for which I’ll be using pure white.
To do this, I angled my tags towards the light. Now my white tag matches the value of the light plane of the cube, and I use another chip, at the same angle to the light, to see how many steps of the scale I need to go down to get to the tone of the top plane.
The grey tag there is a value 7, so I’m seeing three steps down from the light to the halftone.
Here, I’m comparing the light plane with the cast shadow. I’m holding up my darkest tag, the black one, but angled towards the light it’s still showing lighter than the cast shadow. That’s what I mean by going out of my range. I got round that by finding the number of steps from the top plane down to the shadow plane of the cube (it was 5). Then I went back to my tags, but this time held them at the same angle to the light as my panel, and found the number of steps between the shadow plane of the cube and the cast shadow. I got 5 steps.
Hopefully this gives some idea of what I’ve done. The top plane is 3 steps down from the light plane. The shadow plane is 5 steps down from the top plane. The cast shadow is 5 steps down from the shadow plane.
All together, that gives me fourteen steps, 13 to 0. I can’t get a value 13 in paint, or a 0, but now I can choose the range I’m going to work with, and work out where each value will fall in that range. I’m starting with the cast shadow at 2, up to 10 for the light plane of the cube.
I normally glaze over when I see value charts like this, but I can’t see another way to show this. I’ve translated my value 13 white to a Munsell 10 (actually more like a 9 in reality, white paint). The top plane falls at around 8, the shadow plane at 5, and I’ve ended the range at the cast shadow, which I’ve kept accurate at 2.
To wrap up, here’s the three cubes together.
If you’ve managed to get this far through this post, now you know how I worked out a way to keep the ratios between the tones the same, whilst compressing the value range. I hope it was worth it.
Did it work? Well, the painting of the white cube does look like a white cube on a grey cloth, so I’d say yes, it did.
Most likely there are simpler ways to arrive at the same conclusion, and I’m quite sure that many generations of painters have been doing this by eye and getting along fine. But I really wanted to tie this down today, to see once and for all if my idea of compressing the value range but keeping the ratios the same would work with a subject outside of my available range.
The next stage of this exercise is a painting of all three of these cubes together, in the same painting. That brings up the range problem at both ends of the scale. I’ve already done it, but I’m too tired to include it in this post, which is already over long. Hopefully I’ll get it up on the site tomorrow. For now,I’m happy that I’ve proved to myself that compressing the value range is a valid approach.
To my eyes, the white cube looks like a white cube despite the fact that the steps between the values in my painting are smaller than they were on the actual cube. I think it works because the relationships between the values have been preserved.
But still, the grey cube, the only one here for which I can exactly match the values I see in paint, works the best of the three. It will take a few repeats of this exercise to discover whether that’s simply because I didn’t need to compress the value relationships, or because I didn’t compress them quite right on the black and the white cube.
I do think I’ve got a lot more practice to do with this, but at least I’ve proved to myself that the basic principle is sound, and should translate directly into better and more convincing paintings eventually.
This is the first post in a series of six exploring Munsell and value studies:
The Munsell Value Studies are posted in six parts:
Part 1: Munsell Value Studies
Part 2: Three Cubes with Munsell Values
Part 3: More Munsell Value Studies
Part 4: Real World Objects with Munsell Values
Part 5: Cubes and Spheres in Ten Munsell Local Values
Part 6: Still Life Paintings with Munsell Values
The Keys to Colour - Free 6 step email course
Learn how to:
- mix any colour accurately
- see the value of colours
- lighten or darken a colour without messing it up
- paint with subtle, natural colour
once you analyzed the value step relationships and then decided on your new range (from 2ish to 10ish) how did you end up with the actual new value ranges? unless im misunderstanding you, it doesnt seem like your proportions are the same. can you explain please?
Did you purchase your Munsell neutral value set as is or did you put it together yourself? If you put it together yourself, how did you make sure you were getting the values correct?
Hi Bob, the proportions should be reasonably close, although we’re dealing with whole numbers of course not fractions so it’s never going to be perfect. At this point, I was quite early on in my explorations in value and was still making many mistakes, as you can see!
At the end of the day, there’s only so much you can pick up from reading a webpage. I hope that people will take ideas form what I write here and make their own explorations. The answers you find yourself are always going to be the most useful ones for you. I hope that makes sense.
Hi Chris,
I got the Munsell student book and copied the values from the scale in that book with oils. Once I had some practice copying them and could get them accurate, I made little MDF chips so I had a more durable version.
I’ve since come to believe that the accuracy of the value steps isn’t really important though. Value is relative, and our perception of it is changed by the surrounding values among other things.
Using a scale is incredibly helpful, but it doesn’t need to be made up of perceptually accurate steps to be useful for judging values.
Hi, Paul. I´m from Argentina. I apologise for my english, I know it´s not well.
I have a doubt. Munsell´s scale is form by 11 (from 0 to 10) values, but you only use 10. It´s the same scale? When you use a number 5 value, it is from 11 values or 10?
Thanks for teaching us.
Hi Nanac, You’re right, the Munsel scale goes from 0 – 10. When I wrote this first post I hadn’t realised that. The next post in the series corrects it 🙂
But when I use a value 5 in this post, it’s the Munsell value 5 (I was using the value scale from the student book at the time I wrote this) so that much is correct.
Hi Paul, a long time reader of your wonderful articles. You might find the following exercise in reilly/munsell color space similiar to yours, though it is digital.
http://www.broadhurst-family.co.uk/lefteye/MainPages/reillymunsell.htm
The author also has a little known treasure, a color watcher tool that helps to bridge the digital/analog divide, providing dynamic objective reference points in studying tones and colors.
http://www.broadhurst-family.co.uk/lefteye/MainPages/TheColourWatcherTool.htm
Thanks for that Sammy-Lee, that is a very interesting page. It talks mostly about chroma of shadows, and what it says about that I agree with – except that I found when plotting the hue, chroma and value changes across a form from shadow to light by eye, using Munsell chips, that the straight line was more of a curve.
But basically we’re saying the same thing. the lower chroma shadows are much more realistic because that’s how shadows appear to us.
It makes sense when you think about it. Colour is created by light being reflected off a surface. So the less light, the less colour. Obviously, there is less light in a shadow so the ‘colour amount’ – the chroma – will be lower.
Thanks again. I’ll have a look around that site a little more.
Hi Paul.
Can I have a larger picture of your materials? Those styrofoam spheres look so smooth. Is it because of the distance or is it actually like that?
Hi Cedric,
I don’t think I have a larger picture handy I’m afraid. But they’re actually quite textured when you see them in real life. You’re right, it’s the distance that makes them look so smooth!
Thanks Paul. I was initially hesitant to use styrofoam spheres for fear that the texture might make it harder to study values. Looks like you didn’t have a problem with it though.
They work fine. Especially once you’ve primed them, a lot of the texture disappears then. I used acrylic-based primer for oil paints, which is quite think and dries fast.
Thanks Paul for that , but I need to know if you know is this the ways they teach students about colors in the traditional art schools like Angel Academy of Art ?
No, not as far as I know.
Hi Paul…
This is a wonderful post! I have gotten so much from your structured approach to painting. I tend to jump in and begin, but, invariably, I end up being held up on some level before I reach the finish line. Value problems are major and most beginners neglect to even consider the problem of value. Having recognized the problem after taking an online workshop, I am now discovering how complex the problems really are. You are such an asset to those of us struggling to improve our work.
I’m a beginner, and still struggling with learning the difference between hue, value, and saturation. I think those are correct names for the three concepts – the different terminology doesn’t help! But my point: thank you. It looks like these focused posts will help me inch my way to understanding, and your essay on drills is reassurance that I’m heading in the right direction.