Re:...empirical data says no (Score 2, Informative)

Good point, and I agree heartily. However, I'm not suggesting you use a single LED to illuminate an entire image -- you SCAN the image, and rely on human image retention in the eyeball to make it look like a solid image. That'd be one key question -- whether you can scan fast enough to do this. You might even need to double-scan images (twice per visual field) or somesuch.

So you are going to use the LED to illuminate one pixel at a time. For a 640x480 screen there are more than 300,000 pixels. At 30 frames/sec, each pixel is going to be illuminated for around 0.11 microseconds.

The problem is the eye has an integration time of about 1/30th of a second, but the pixel is only illuminated for 0.11 microseconds. This is equivalent to integrating the same illumination level for the 0.11 micro second time. But when you are just shining the LED on the wall you are integrating to the full 1/30th of a second. The let's assume the illumination level is 1. So the integral for the Photon on the wall is 1 integrated for 1/30 of a sec or .033. For your scanning LED the integral is 1 integrated for 0.11 microsencond or .00000011. For a given pixel size, this means you need to make the light 300,000 brighter if you scan it than if you don't if the eye is going to see the same amount of light.

Your TV solves this problem by having a phosphorous surface on the back of the screen. A powerful electron beam excites the phosphorous and then moves on. The phosphorous holds the image while the electron beam is elsewhere.