How to Paint the Figure in Trompe l'Oeil

Here is a detail from the finished painting, enlarged because I used very small brushes.

You can use any background you like. I used an old faux-painted limestone sample I had lying around, but you could just as easily do this on faux bois to imitate the look of carved wood. The only stipulation I'd make is that to get the best effect, your background should be no darker than mid-range on the value scale.

In the next image, you can see the brushes I used. One small pointy one, and another splayed out and busted one. When doing shadows, I'd paint with Mr. Pointy then stipple and soften with the busted one. I'm not very fussy about materials or brushes. When I was young, I used to read all those manuals and study hard to learn the "secret" materials that would give me the edge. The only secret, I learned, is your eyes.

Step 1

 If you look closely in Step 1, you can just make out the pencil lines I used to establish the basic design. I took a photo of a bas relief panel at the Met as my reference. This is actually the first time I've used a photograph of actual relief as reference material. Mostly, I'm either inventing relief or copying another trompe l'oeil painter's work (who probably also invented the relief). This was a great chance for me to exercise Rule # 1 in illusionistic painting: Paint what you see, not what you know. 

It's all too easy to fall back on what we "know" about Form rather than simply using the evidence presented before our eyes. "Shadows are dark; highlights are white; reflected light goes here; etc." These are all learned rather than observed truths. [Painters use old tricks such as flipping their image upside down, or examining their work in a mirror, to escape the natural tendency to paint what we think we see.]

Step 2

Step 2 shows the completed shadows. There are no highlights at all here. You can see that the effect is 99% complete already. We could easily leave it like this, and call it a day.

I build the shadows very slowly using Ultramarine and Raw Umber acrylic paint, mixed with a little matte varnish (which dries quickly so I can keep working) as a medium.

The shadows are built up in layers. I never try to establish the darkest dark in the first pass. It's much more tentative then that. I build up darks in translucent glazes, always erring on the lighter side. I tend to work back and forth all over the image, as opposed to finishing each area completely as I go.

Step 3

Step 3 is my first pass at highlights. They might be hard to detect on your screen, as they are very subtle, but they are there. I used an opaque white, mixed with yellow and a little purple. I've heard that some people like to shift the hue of their highlights in opposition to the background color in order to make them pop more. [In other words, if the background hue is yellow (as with mine), they might shift their highlights into purple so that they jump out at you.] I don't do this. I use a lighter value of the background.

Use highlights very sparingly! As subtle as mine might look to you, when I look at the opening detail image of this post, the highlights jump out at me as being too strong and brushy. The image looks like it's been dusted with snow.  The shadows are soft and muted, as they should be, but the highlights are harsh, overused and overly delineated. We want to avoid this at all costs. (It was too late for me).

Step 4

 In Step 4, the only difference is that I carefully glazed the chest, top of the head, and the right knee with a second highlight. I decided that those three areas should be a little more prominent than they were.

It will be easier to see the differences between stages if you view these images in slideshow mode, and scroll between them.

Step 5

 Step 5 might be hard to see, but I think it made a difference. Since all my highlights were already laid down, the only way I could make the relief pop a little more was by darkening the background. I explained the reasoning behind this in a separate post about relative value, here.

I certainly didn't want to add any more highlights, as mine were already too bright. Instead, I used a very washy ultramarine/raw umber glaze and stippled it all around the top right corner, outside the main figure.

I also used a glazed version of my highlight color and subtly lightened the bottom left corner (again, outside the figure only).

Finished.

To give you an idea of how light/dark my values are, here is a chart that shows you (below)...

Along the top you see 3 swatches of color. These represent averaged tones taken directly from the finished piece. Directly underneath them, you see 3 grey values (A, B, and C). These are the same colors from the top line desaturated so as to see value only. Below that, I plotted A, B, and C against the Munsell value chart. You can see that the entirety of my painting occupies roughly three value steps on the Munsell chart (from 6 to 9). No white, and nothing at all on the lower half of the chart.

Some more detail photos...

Shut that Door! How to Paint Doors in Perspective

Domenico Remps

Artists are often confronted with painting doors, windows and shutters ajar. This creates a problem: How do we make sure that the door actually fits the frame when it's seen half open?

Surprisingly, mistakes are more common than you'd think, even among master painters.

Vanitas trompe l'oeil, by Cornelis Gysbrechts

It's a head-scratcher that's bested the best. Cornelis Gysbrechts was a Flemish trompe l'oeil master known for his paintings of cupboards containing household objects depicted with stunning realism. Often, he included an open glazed door that revealed the contents within. However, his realism falls flat every now and then because if we were to close over his cupboard door we'd see that it does not fit the frame in which it is sitting.

Not only that, but if we extrapolate the vanishing lines created by the glazed doors, we can see that they don't converge at distance points (above). At least make your lines converge at the vanishing point. This is a pretty big error, but it's not the biggest problem with the painting: it should be pretty obvious that if those doors closed, they'd be too big for the opening and crash into each other.

At least make your lines converge on the horizon

To avoid this mistake, how do we determine the shape and size of our swinging door? Viewed from directly above, any opening window or door would describe a perfect semi-circular arc in space. Seen from an oblique angle, like say, standing in front of it, that perfect arc gets distorted to look like an ellipse.

How do we draw the shape of our ellipse? Viewing distance is the major factor. Here are the rules:

• The shorter the viewing distance (i.e. the closer the viewer is to the action), the wider the ellipse. 

• The longer the viewing distance (i.e. the further the viewer is from the action), the tighter the ellipse.

The above illustration explains how. The cyan rectangle represents Gysbrecht's open cupboard. Draw a circle in perspective (dark green ellipse) with its center at the bottom of the hinged door. My blue dotted construction lines should help you figure it out. Explaining it in words is useless. I set my viewing distance here for about 6', or twice the approximate width of Gysbrecht's 3' canvas. It's too detailed to explain how/why in this post. Just note that now there are green ellipses above and below the door frame on the left and right. Anywhere your door swings (the doors are those pink shapes) its edges should touch the ellipse. Presto; correct perspective.

(A) shows an infinitely long viewing distance. (B) shows a very close viewing distance.

I did another post on viewing distance that might help to fill in. You can see it here.

Now, in the case of the Domenico Remps cupboard the intended viewing distance is very short. Objects are painted more or less life size, and the whole scene is painted as if we were standing directly in front of the cupboard and opening the doors ourselves. However, Remps dispensed with the rules here because if we were truly standing that close, the open doors would be subject to extreme foreshortening (as in B, above).

In actuality, with a viewing distance of 3' or less, the leading edge of the open doors would loom out towards us dramatically and seem quite large. Firstly, this would mean that his doors would be too large to be contained within the frame of the canvas - which would break the golden rule of trompe l'oeil painting. Which is ... all objects should be fully contained within the canvas. In other words; if objects are partly cropped by the frame, then it's a still life and not trompe l'oeil.

Still Life by Champaign (left), Trompe l'Oeil by Hoogstraeten (right)

See the difference above? But that's a different lesson. Back to the topic at hand...

Where were we? Oh yeah, painting doors in perspective. Secondly, it would make an ugly painting. So Remps bent the rules. But they still apply nonetheless. His open doors will clearly close tightly if we swung them shut. He simply tweaked the viewing distance to give himself a tighter ellipse (as if we were viewing from far away) so that he could fit the entirety of the open door within his canvas. Why? It looks better that way. It doesn't matter that it breaks with the illusionistic realism elsewhere in the painting. It was an artistic decision.

Nevertheless, he made sure that no matter what, his doors were still painted according to the rules of linear perspective. They could close if he wanted them to. This is the important point regardless of the viewing distance: make sure your doors can close.

This problem does not apply to doors alone.

Above, we see that a fine trompe l'oeil painting from the Uffizi of an illustrated book is spoiled because of the unnatural size of the pages. Overlaying an ellipse reveals that one of the pages in particular (highlighted in orange) is way too big for the book. If that book were closed, the highlighted page would stick out an inch beyond all the others. If anything, because the open pages are slightly curved, their edges should fall inside the ellipse and not outside.

Below, I took a photo of an antique map and overlaid an ellipse to demonstrate that the edges of the stiff open folds align perfectly along the ellipse.

Tutorial: How to paint trompe l'oeil gold beading

Creating the "gold" beaded molding above this painted curtain was a relatively simple affair. Because there are so many of them across the whole surface (of which you're only seeing a small crop here), it pays to simplify the technique so that you don't lose your mind or, more importantly, your money. Here's how I did it.

First, I used blue tape to section off the area I wanted to paint, and laid down a wash of yellow ochre toned down with some raw umber.

  Next, I striped along the top edge with a line made from cadmium yellow, yellow ocher and a drop of white, and along the bottom edge with a line of burnt sienna, burnt umber and a drop of yellow ocher.

 Then I cut a stencil. This will be for the dark areas between the beads.

 Use a stencil brush to pounce a dark color made from burnt umber, raw umber and ultramarine, repeating your stencil across the full length of the molding.

Here's how it looks after I take the stencil away (above).

I used a fine pointed brush to lay a stroke of burnt sienna and burnt umber along the left edge of each bead. Notice that my stroke is slightly inside and away from the bottom edge of each bead. This gives the illusion of a reflected highlight.

Then, I added a small stroke of titanium white with a hint of cadmium yellow for the bright specular highlight. It too is placed in and away from the top edge of the bead.

Lastly, I added a dab of cadmium red opposite the specular highlight, and removed the blue tape.

Here's a close-up of how the completed molding looked. You can see that I used a similar method (but different colors) for the grey tongued molding along the top. This technique is quick, and useful in many scenarios.

Prehistoric Painted Illusionism

Praxinotropes use mirrors around a central column to reflect the inside of the drum. The "overlapping" images create the sensation of movement.  

People like to be fooled. We find the notion of illusion entertaining, and go out of our way to believe in magic. Long before David Blaine, Imax theaters, and Florentines faffing about in velvet hats, there were much hairier creatures crawling around in caves, creating a visual magic of their own. 35,000 years ago, painters made their way into the belly of the earth and left behind a spectacular tableau of aurochs, horse, bison and elk, often composed of curious double images. 

The Chamber of Lions, in Lascaux, depicts curious double images, once believed to suggest the motion of their subjects.

Archaeologists have long suggested that the double images are attempts to breath life into the stone; that under flickering torch light the animals are seen to dance and move. Early experiments in kinematics - or motion pictures - built on this principle by using a spinning barrel with staggered images on the inside that appear to move when spun, called a zoetrope or praxinotrope. Recently, however, a couple of artists provocatively suggested that these ghost-like double images are not illusions of movement but are in fact illusions of depth.* 

On the left, the camera is focused on my finger, and shows a blurred double image of the X on the wall.
On the right, the camera is focused on the X, and the double image is instead my finger.

A simple experiment explains their idea: Hold your finger up in the air between your eyes and this page, and focus on it: You see one finger. Now shift your focus to the page behind it, and magically there appears to be two ghost images of your finger hovering. The double image of your finger is thus understood to be closer to you than the single image of the page. Simples! 

Every child is familiar with this basic optical trick. It’s the optical displacement of an object caused by our binocular vision (or stereopsis), known as “parallax,” and it’s a binocular depth cue that babies learn as early as four months out of the womb. In order to navigate 3-dimensional space, we learn from a very early age to interpret our world via optical cues such as parallax, and organize it into "near" and "far," for example.

Perhaps our primitive ancestors were harnessing this basic navigational tool and using it to create images depicting not movement through space, but space itself. If so, they'd have beaten the linear perspective discoveries of Alberti by, oh, 34,000 years or so.

* [In a talk by Ryan and Trevor Oakes, titled “Seminal Notions: The Idea and Practice of Perspective,” given at the Chicago Humanities Festival.]

A New Perspective on Rubens

Rubens painted a set of massive canvases (including two measuring 28 x 20 feet and two others at 40 x 10 feet) for the ceiling of the Banqueting House at Whitehall in 1630. Despite boasting to his customer, James I, that he was “by natural instinct better fitted to execute very large works than small curiosities,” he seems to have made a bit of a mess of the linear perspective. In one of the panels (above), the seated figure of the King, his retinue and even the putti floating above with the royal coat of arms, are all painted as you would expect in di sotto in su painting; they're all foreshortened from below. The columns, however, are parallel to the picture plane which is incorrect.

I took the liberty of sketching on top of Rubens painting (below) using the correct viewpoint (according to linear perspective, at least), such that the columns and the figures appear to recede towards the same - singular - vanishing point.   

This discrepancy in Rubens' paintings was first noted by 18th century painter James Highmore, who published his "Critical Examination of paintings on the Ceiling at the banqueting-Hall at Whitehall," in  1754. Below, on the right, is the row of columns as Rubens painted them, and to the left appears Highmore's engraving of the corrected view. Highmore points out that Rubens used up to three (!) vanishing points in each of the panels. Why? We'll never know. Certainly, he wasn't especially used to painting on ceilings. I leave it up to the viewer to decide which version of the scene is the better view.

Here they are again, side by side for easier comparison.

Enter the Vault

Definitely not a bad Italian restaurant. 

Arches appear in trompe l'oeil murals with predictable regularity. I don't know how many times I've sat in a bad Italian restaurant, staring at an appallingly painted Bay of Naples through some row of wonky arches. Overcooked pasta, bad wine, swirly stucco and a gilded gorilla on a glass pedestal staring me out as I go to the bathroom. We've all been there. With even the smallest of effort, however, these murals could be made better. Not good exactly, but better. 

To begin with, I'd argue that the arch is not the best structure to imply depth. A single row of arches in the foreground, perpendicular to the line of sight and revealing a landscape behind (for example), creates two distinct planes. We are made aware of the picture plane itself, as signified by the arches, and we are made aware of the plane of the background. A distant landscape exists very much on a flattened plane. Murals painted according to this standard model are distinguished by two discrete planes, foreground and background, usually painted without much physically linking them. It's a very static and finite solution to creating depth.

The vault, on the other hand, creates a continuous bridge between the foreground and middle distance. It may even appear to extend towards us, implying a continuous ceiling right over our heads. The painting above provides both examples in one. An arch revealing a scene (on the left), where the illusion of depth is created almost exclusively through use of value, and the receding vault (on the right), which uses both value and linear perspective to create its spatial illusion of depth. The scene on the left appears static, while the one on the right is dynamic.

[Oddly, the light on the background building seen through the window comes from the right, while the light on the vaulted corridor comes from the left.]

Replica of Ghiberti's Baptistry Doors at Trinity Lutheran Church in Hicksville (!)

The number of architectural structures that inherently suggest depth is quite small. If you want to create the illusion of three dimensions in an architectural mural or large-scale painting, you're pretty much limited to the vault. 

A barrel vault may even extend all the way to the background spatially, as in the wonderful Baptistry doors by Ghiberti. The carefully delineated outline of a vault is all that's needed to suggest great depth, despite the actual relief being no more than an inch or so. The action stops at the altar, but or eye travels all the way to the back of the nave.

Massaccio's Trinity

Massaccio's Trinity is a perfect example. The arch does nothing to create depth besides delineating a plane, it's the vault that creates the illusion that wows everyone.

The original "cassetted" vault: Temple of Maxentius

"The cassetted vault was well known from the Roman temple of Maxentius and other buildings. It became one of the great themes of humanistic architecture. Masaccio used it in his Trinity, generally accepted as the first extant work in perspective. Borromini's use of a variant nearly two centuries later in the Palazzo Spada attests to the enduring fascination of this illusionistic form."" [Kim Veltman, The Sources of Perspective]

Architects such as Borromini,  Scamozzi and Bramante, understood that architectural trompe l'oeil was best served by the vault, using it to spectacular effect in their illusionistic built environments.

An arch is more of a decorative frame. It sits in the foreground as a mute barrier. Arches serve as portals, a decorative meme intended to entice us to walk through into the scene beyond. Taken on their own, however, they serve to isolate the viewer from the scene, acting more like a simple barrier than, say, a pathway running sinuously from us and disappearing into a landscape. A long barrel vault is the architectural equivalent of a forest path in landscape painting. Paths are irresistible. They trigger an irresistible urge to explore and see what lies around the corner.

Arches are not. I see an arch in a painting and I feel like there's going to be a red velvet rope with a bouncer telling me I'm not on the list. Whatever drama is unfolding in the painting above, we’re clearly not part of it, just as I'm viscerally aware that I am not invited to Veronese’s Feast at the House of Levi. I'm a spectator, not a participant. On the other hand, Raphael’s School of Athens makes me want to march right up those steps to join those conversations in the crowded vault.

Veronese’s Feast at the House of Levi

Veronese circumvents the static nature of the single row of arches in the foreground by plonking a second row directly behind it. This implies a vault running perpendicular to us between the rows of arches, and creates depth. If the architecture is to run perpendicular to the line of sight, then it's a handy little trompe l'oeil trick to simply place a second row. Notice (below) how much depth is implied by adding the second row of columns. 

Colonnades running away from the viewer, along the lines of sight, imply great depth. "By the end of the (Fifteenth) Century, artists such as Bramante realized that the representation of colonnades was particularly suited for perspectival purposes because these permitted one not only to look into but even look straight through a space." [Veltman]

From Antiques in Italian Interiors, Vol II

And, of course, the last word goes to Raphael. As if there was any question on the matter. He pretty much closes the argument. How tempting it is to march up those steps and enter the vault, listening in on the conversations of all those great thinkers. Notice too, how Raphael has chosen to position his key players at the precise point of greatest implied depth in his painting. He knew that we are viscerally drawn to the "deepest" spot, and planned the hierarchy of his incredible painting accordingly. Another example of how depth is intrinsic to composition.