I spent the last couple days, in particular too many hours yesterday, working on a new painting, Five Concentric Wireframe Cubes. I also took a lot of pictures along the way, as I drafted out guidelines on a three foot by two foot canvas and then systematically mixed and laid colors into that drawing.
I tend to share bits and pieces of process on my twitter feed, @joshmillard, but I thought I’d round up these process photos as a blog post.
I’m pleased with the result, and also pretty sore from all the careful freehand edge-making at various 60 degree increments on a large canvas. Working larger like this introduces new challenges compared to the many square-foot paintings I’ve done in the last couple months. New complications, but also exciting possibilities.
One implication of a larger canvas is that the machine-cut stencil process that fits smaller paintings (e.g. Continuity) really well is more difficult to use well. I have done a couple of larger stencil-driven pieces so far this year — Sierpinski Carpet, Concentric Squares which required four coordinated stencil plates; Conservation of Area which required six — and they came out well, but it was a great deal of extra work to involve the notionally time-saving stencil element. I’m still working out the balance, there.
But difficulty of coordinating multiple stencil plates aside, FCWC isn’t a great fit for that anyway: the painting is made up of a lot of fields of directly adjacent color, so there’s no negative space gaps for stencil material to define.
It’s also a (relatively) simple design; compared to the two pieces linked above, it has no thin, closely-packed lines, but rather fewer, larger areas made out of (to think of it one way) one or more contiguous unit-inch equilateral triangles. And so drafting it out with rule and pencil was as good a solution as any.
The painting started as a sketch, one of several I made the other day on this theme of concentric wireframe cubes. This one felt most interesting to tackle.
The sketch is on isometric graph paper, with triangles at approximately 0.25 inches on a side; the finished painting scales this up by a factor of four, taking a 5.5 inch tall sketch and turning it into a 22 inch tall figure. This clean conversion factor made the process of drafting it out on the canvas a little easier than it otherwise might have been; I’ve tied some other large-scale work to similar convenient multiplying factors, as with the wall paintings I made last year using unit-inch triangles for the Menger sponge and making the smallest lines of the Cantor set 0.125 inches wide.
Drafting the guidelines for the painting took a couple of hours, working at the dining room table with a metal yard stick, a smaller ruler, some 30/60/90 triangles, and a yellow colored pencil. The result, above, is difficult to photograph, though I promise it’s there if you look really close.
See? The whole thing was like that, pinky swear.
My drafting on this was not exceptional. Part of that is that stretched canvas is by nature not rigid and so straight lines can end up being a little less straight (or less parallel, anyway; concave, flexible surfaces get a little non-Euclidean under even the best circumstances) than you might hope. I had similar problems painting on my office wall; the wall was not really flat. It’s an old house.
But part of it is that I was not nearly as careful and deliberate about setting up an isometric grid as I could have been. I did my best to set up some fairly true initial end points on the drawing, essentially a large hexagonal perimeter, and to recalculate from end points periodically instead of just making relative measurements from one line to the next, but, still, errors accumulate. You can see from the detail shot that my various intersections ended up off by a bit in spots particularly toward the bottom left region, sometimes by as much as an eighth of an inch.
Not great. But good enough for the work, since it’s large enough that the initial impression of it will be from a distance where those errors disappear in strong lines and colors. And I could nudge things a little bit locally when putting the actual paint on.
After the drafting, it gets straightforward, in theory at least: put in one color after another. I did the painting freehand, just carefully applying paint to edges with a sable #12 bright brush, which is about 3/8ths of an inch wide and holds a very crisp edge, and then filling color in and brushing everything again at a consistent diagonal to get a flatter, more organized look to the paint.
I did the drafting Monday afternoon; I began laying colors in that evening, and got this far before calling it for the night to spend time with my wife.
And then back to work yesterday morning.
One implication of a large canvas: reach. In principle, canvas is canvas, and a paintbrush should land the same way anywhere. In practice, I don’t have the steadiest of hands, and being able to prop my arm or wrist against something helps my lines go on straight.
I’ve spent my whole life drawing in pen and pencil and tend to work small (see for example my tiny Bird Presidents drawings) and to draw from the wrist or with my fingers while the heel of my thumb is planted heavily on the paper. With painting I’m trying to fight that a little and use my arm and the length of the brush to add expressive freedom to my marks, but for a painting like this that is so fundamentally built around precision that’s not really an option.
And so finding a way to brace my painting arm, either on the canvas (easy when there’s not much paint on it yet) or on the brick shithouse of an easel I built earlier this year, or on my other arm itself braced on something, is key to getting my lines straight. The easiest solution in this case, where the top of the canvas is a couple feet above the main easel support, is to just flip the thing over.
So at each stage of this I ended up inverting the canvas so I could get those top edges on cleanly. You can also see above the edges-first part of the process, getting those outer lines on in green before proceeding to the less fiddly fill-in work.
Conceptually FCWC is three sets of five colors, facing the three respective directions in this isometric rendering. This is all of the first, top-facing set done. I chose to use unmixed colors out of the tube for these, just pure vibrant pigments, with the other two sets being mixed lighter and darker respectively to create the sense of (however impossibly well-behaved and shadowless) volume of the final image.
The five colors I used here: cadmium red deep hue, cadmium orange hue, cadmium yellow hue, chrome yellow hue, and cadmium green. (The “hue” variations are synthetic alternatives to actual cadmium- or chrome-based paints; they’re less expensive and also less toxic. The cad green I own, a recent acquisition, is the odd man out there.)
The sketch went red to blue and my initial thought was to echo that in the painting, but once I started thinking about colors the cad red/orange/yellow progression really appealed to me. I like those colors a lot in general, and they step along nicely, but cad yellow is a very warm yellow, and so jumping to green from there would have been jarring. Chrome yellow is a good midpoint, a cooler, bright yellow; finishing up in cadmium green felt just about right at that point.
Changing the colors from the sketch made for some headachey double-checking on some of the complicated interior bits of the painting; if I were going to repeat this process in the future I think I’d plan out my colors at the sketch stage more deliberately. But I didn’t know I’d be painting this when I made that sketch, so, welp.
I mixed each of my base colors, one at a time as I proceeded with the painting, with some titanium-based transparent white to lighten and slightly desaturate them for the left-facing portions of each wireframe. The mixing process there was very straightforward, more so than the final dark set, as I’ll detail below.
Kept at it, and got the second lighter set of colors laid in. Not pictured: flipping the damn thing over again.
Earlier that morning, my friend Jesse Holden had come over to the house; we’ve been trying to do regular weekly art get-togethers, what we’ve taken to calling Art Mondays even though the last several haven’t happened on Monday. We mostly work independently rather than collaborating, but it’s nice to just be in the same room and keep each other honest about getting some art work done instead of lapsing into internet distractions or so on.
And at this point we were both ready for a break and some lunch, so we headed into St. Johns to get food and beer.
We didn’t plan that, but Jesse saw what had happened and lined them up for me to snap a picture of. The one on the right, by the way, was a sweet raspberry beer on a nitro tap, a ridiculous beer but not cloying like you might expect and delivered with an amazing foamy raspberry head on it. Could have been served as a dessert in its own right.
And then back to the house and back to work. The final set of colors, the right-facing portions, I wanted to be darker. For the red I mixed in some bone black and a little touch of my cad green to desaturate it. Adding a bit of a complementary color to a hue brings the mixture a little toward a neutral grey middle, which is part of what I wanted here, to keep the top-facing pure tube paint colors the most vibrant elements of the picture.
For the orange, I mixed in ultramarine. Too much, in fact; I ended up with a distressingly dark first mixture. What I came to after more mixing was lighter than where I started but still a little darker than felt right, a deep-feeling brown color that I worried was too dark compared to its lighter orange coworkers.
(Darkening yellows is fraught in general; they tend to go green when you mix in blacks, rather than seeming to simply darken as you’d expect. I added a bit of red to most of my darks here, to warm them up a little bit in compensation.)
Local color is tricky, and I’m learning to try to think through the whole context of a composition instead of just reacting to what’s in front of me at the moment, mid-process. And so I decided to trust the idea of the finished painting instead of looking at the brown next to the still-white gaps to the right that would eventually be filled in with similarly low-value mixes.
Quick aside: this is what a lot of the fiddly bits look like in action: needing to get a very straight flat line on the canvas when there’s wet paint on the area where I’d otherwise like to rest the heel of my hand. There’s other solutions to this problem: rotating the canvas more often, adjusting the height of the easel, using a mahl stick, having a steadier arm, or having the patience to let things dry over the course of a week instead of trying to power through the whole thing alla prima. But I tend to get caught up and then just hope my wrist isn’t sore the next day. I seem to have gotten lucky in this case; plenty of other parts of my body are sore, but my wrist is feeling fine.
And but so yes. The dark cad yellow I mixed less dark than the dark orange; the dark chrome yellow likewise. And then for the green I let myself go a little darker again after all. If this were a study in locking down a consistent value from one hue to the next this’d have been a failure, but I was reminding myself at this point that that’s not what I was doing and the important thing is for the painting to work on its own. I feel like it does; that dark orange brown I was so worried aobut turned out to feel pretty good after all.
In the end I had fifteen colors, five in each set of base, light, and dark mixtures, filling up my palette. In many cases I mixed too much paint, which is a good place to be if slightly wasteful; the alternative is to mix too little of some complex shade and have to recreate it.
Which happened a couple times yesterday. I would tend to mix too much of the orange and yellow hues, and then with the larger green areas end up with not enough out of an unconscious compensation.
Mixing a close match for a complicated, improvised mixture is a pain; for a chromatically busy painting with lots of colors interacting in non-linear ways, close can be close enough; for something like this with very flat, even areas of color, fudging it is more of a problem. I ended up fudging it anyway, but isolated close-but-not-quite mixtures of the same shade to discrete chunks of the painting, relying on the optical stress of e.g. a couple of green bars interrupting the visible pieces of a yellow cube to distract the viewer from seeing where I didn’t quite get this ochre-ish mixture to match that other one.
I am still trying to convince myself not to be miserly about a dollar’s worth of paint; the stress of having to remix and fudge this sort of thing, spending ten minutes cursing and sweating with a palette knife, is not worth the savings.
And so we’re back to the top, with the finished piece. I’m happy with it, and really pleased that folks have so far been responding well to it in the places I’ve shared it. But the process is often as interesting to me as the final product, so I hope this was an interesting read for y’all as well.
Also: I am continuing to work on, and starting to try to get the word out about, my new art showcase site, art.joshmillard.com. If you’re liking this stuff, go take a look, and consider showing it to folks you think might enjoy my work as well.