Monday, February 3, 2014

Seeing Brightness in the Season of Little Light

 FIGURE: Robert Pepperell’s “Christmas Scene.” Reprinted with permission of the artist.

 I am about to talk about how we experience brightness as we struggle with the winter solstice.

Brightness is an attribute of sensation of light. It is hard to pin down, partly because it is easily confused with luminance (whose equalization across a visual boundary makes the boundary minimally distinct) and with lightness (the apparent reflectance of a non-self-luminous body).

A light’s spectral content contributes to its brightness, but is not the whole story. To show that brightness doesn’t necessarily increase with light intensity, Bill Thornton used to project three circles of light (red, green, and blue) on a screen, in such a way that the areas underwent mutual overlap. In his demonstrations, the whitish center (with all beams present) was invariably the least bright, and the separate red and green areas appeared the brightest.

Psychophysicists and artists have long acknowledged the contribution to brightness of geometric factors that are at least as influential as the spectral content.  Ernst Mach (from whom Mach bands are named) observed that a visible region appears brighter when it is next to a darker region. Mach’s phenomenon is clear when the gray screen of a turned-off TV set develops black regions when the set is turned on.

Truly, the brightness of a light depends on its spectral content and on contrast with its neighbors, but even this is not the whole story. The influences I described above can be incorporated into a vision theory that has three kinds of receptors (cones) and comparisons of spatially adjacent visual inputs, but that is just low-level vision. High-level vision---which involves recognition of whole objects and their spatial context---may also enter. Over the years we have seen some striking demonstrations of global influences in the visual field (the work of Alan Gilchrist comes to mind), but I have noticed an effect that is not widely noted. I call it the “Christmas-tree-light effect”, because I first had my attention drawn to it by Robert Pepperell’s painting “Christmas Scene”.

I met Pepperell and his painting at the 2012 SPIE Electronic Imaging Symposium, a conference at which I always seek the explanation of “new” visual effects by old theories. Accordingly, looking at the “Christmas Scene”, I was surprised to see that the Christmas-tree lights were blurred and yet seemed brighter than they might have appeared otherwise. Classical simultaneous-contrast models would have predicted a diminution of brightness by blurring the lights, but it seemed the opposite was true. You can see the Christmas-tree-light effect in the figure here because Robert Pepperell has graciously given permission for me to reprint it.

At this moment you may remark (as our ISCC News editor did) that my assertion of enhanced brightness is not quite supported by Pepperell’s painting. Indeed, this objection is correct: I have not controlled the visual experiment by, say, Photoshopping all the tree lights to remove their blur patterns. Besides the obvious explanations of personal laziness and reluctance to upset the artist with my experiment, I respond to the objection by saying that enhancement of brightness by blur has a sound ecological story. Also, this kind of trick has been used by photographers for a long time.

Why should high-level visual processing interpret a blur around a light as evidence that the light is very bright? Well, any light seen through the eye is surrounded by blur due to light scattering by the eye. The blur is invisible if the light is very dim, but it becomes more visible as the light becomes more intense. In a dark living room, a point of light from (say) an LED is very intense, and saturates (dazzles) the visual response. The only remaining evidence of the light’s brightness will then be the blur, which (as a fixed fraction of the intensity of the light) will be quite visible.

And what is the photographic trick that uses the “Christmas-tree-light” effect? It is  called the “star effect” [1], and involves turning each point of light into a star-shaped pattern using a periodically ruled grid called a star filter.  There’s some nice Fourier-transform theory here, as well as good optics. But for purposes of the present essay, it is merely further evidence that points of light can be “enlivened” (which I interpret to mean, “enhanced in apparent brightness”) by surrounding them with patterns of lower intensity.

A parting problem: How is it that I see star patterns around points of light at night? Surely I don’t have a ruled grid in my eye that piles up Fourier components of light in an orderly way…

1. Tiffen, I., Star effects: enliven points of light – how star filters work. Print edition: Student Filmmakers Magazine, June 2008, p. 44; online edition at

Michael H. Brill


Carl Jennings said...

The photographer's trick is based on an old painter's trick of specular highlights, which can be seen in highly reflected objects found for example in still life life paintings like this one by Willem Kalf (1656)

The blurred effect is also found in paintings of candle flames, as well as stars and moons, though I think atmosphere plays a role here also. Clearly the usual explanation of brightness contrast models break down here, as they do when it comes to something like White's Illusion, for which there is still no accepted explanation. I agree with you Mike - there is something much higher up the processing food chain going on here. I would suggest it relates to our empirical experience of the world, our ecological interaction with our visible environment. After all, that is all painters had to go on. And the same would apply to the 'star pattern' of stars - how else could we explain the use of the star pattern in art all the way from the Egyptians through the Renaissance? Here is a Baroque painting by Adam Elsheimer (1609)which is very exact in it's observation, with colored stars and all, yet even here we have several prominent 'star patterns'.

Ezra Bradford said...

This seems related to the visual effect of "bloom", which quantizes the effect of the scattered light.