Rosa Bonheur's fantastic "Horse Fair" is a massive canvas that caused quite a stir when it first appeared at the Salon in France back in 1853. Standing in front of the painting today at the Met, the effect of the circling horses was dizzying. I could almost hear the commotion and clatter of hooves. The visceral sense of "being there" is accentuated by the way Bonheur constructed her painted space.
|Solution 1: Objects on the same perpendicular plane get smaller to the left and right|
The panoramic sweep of her canvas seems to adhere to this Hyperbolic perspective grid. Imagine that these black silhouetted figures are all standing in a perfectly straight line from left to right. I've exaggerated the effect a bit in the illustration, but to the real-world viewer it would appear that they shrink towards the horizon line as they get further away from us. We know that the line they are standing on is straight, but it appears curved in reality.
Our brains do lots of similar tricks, lining up our reality with what it knows to be true rather than what it actually sees. Imagine you arrived late to a movie and you're sitting in the front row of the theater, squashed all the way over on the furthest seat to the right. A giant square shape appears on the screen. Your brain knows for certain that that's a square, despite the fact that it looks nothing like one from your current vantage point.
You can see the same effect in this stitched-together panoramic photo I took of Storm King seen from Breakneck Ridge. It's not a bend in the Hudson river you're looking at here. It's just that it adheres to the same "optical" grid as Bonheur's painting whereby objects in the center appear larger (because they're closer) than objects to the distant left or right. This happens when we try and squeeze anything beyond about 60º (our normal - undistorted - field of vision) into the frame. In this case, the extreme distortion of the river is due to the 180º view.
|Solution 2: Objects on the same perpendicular plane stay the same size|
Linear perspective takes a simpler solution to the problem of how to represent objects in space. Its solution is Euclidean in that its geometry is based on straight lines. The standard approach would have us portray as the same size any objects on the same plane, perpendicular to the viewer. This works just fine when we stick to portraying a scene that's within the undistorted 60º field.
When painting especially wide scenes we are faced with a dilemma. Maybe we just paint everything the same size (as in Solution 2 above) and let reality do the shrinking for us. If you've ever stood in front of Veronese's gigantic canvas in the Louvre you'll know that he had no need to artificially reduce the scale of his figures in paint. The outer figures are so far away from the viewer that they naturally appear smaller.
|Fish-eye distortion isn't just for large canvases|
It's not that painters didn't understand the problem. It seems that they simply found this fish-eye distortion unattractive, so (for the most part) they chose to ignore it and paint everything in straight line grids.
Pierre Puvis de Chavannes chose to ignore the issue. He painted incredibly long paintings, where each figure was the exact same size.
|"The Triumph of Aemelius Paulus," by Carle Vernet|
A third solution is also possible: Why not just kind of blend the first two solutions together? You know: fake it a bit?
Carle (Antoine Charles Horace) Vernet did just that. He painted "The Triumph of Aemelius Paulus" in 1789, as a bravura performance of pre-Revolution pro-Empire sentiment. Vernet painted the architecture in perfectly straight, horizontal lines, but his painting avoids the stiff frieze-like nature of De Chavannes. How? He manages to inject a little hyperbolic perspective into the scene in two ways.
|Central figures are close to the bottom of the canvas, and have strong value contrasts, thus appearing closer|
Firstly, the central figures are lower on the canvas than those to the left or right. This simple trick makes them appear closer (even though the are the same size as those to the left or right. It also approximates the optical effect we would have if viewing this scene in reality, in that figures to our distant left and right would appear closer to the horizon (and thus further away).
I say he "approximates" because, in reality those peripheral, distant figures would also appear to be smaller. Vernet, however, defers to the same unnatural rules of linear perspective as Pierre Puvis de Chavannes in that he presents these peripheral figures the same size (Linear perspective tells us that figures on the same plane should be painted the same size).
|Figures on the right are higher up and less "contrasty", making them seem further away|
The second little trick he pulls to avoid de Chavannes' "frieze" trap is to reserve his strongest values for the central figures. Although the figures to the far left and right are painted the same size as those in the center, they appear to be further from us because they are painted with a narrower range of values. While he does use strong lights, his darks are greyed-out. This pushes the figures back in space.
Anyway, there you have it. A few thoughts I had while strolling through the Met the other day. A comment by Annabel Armstrong on my last post got me thinking about this subject. One of the great benefits of working on an apartment literally across the street is that I get to spend my lunch break with this stuff.