So I thought to myself, wouldn't it be neat if you could ray-trace an impossible Escher-like scene? It's not too hard to construct an illusion that works when you look at the scene from a specific angle, like those photos of Lego versions of Escher. The trick is to create a model of impossible geometry that lets you move the camera around and view the scene from different angles.

I start thinking of ways of approaching the problem, and it doesn't seem like it would be too hard, which triggers the thought, "I bet someone's done this already." Get out Google, search for "render escher".

Hit #5 is a page at some collaborative idea factory named "halfbakery". Someone posed a similar question in April 2002, wondering about Escher-like computer games. The first comment is basically "delete this idea since it can't be done". There's some discussion, but it doesn't go anywhere. Various people add links related to Escher, but nothing particularly relevant to the question posed.

Hit #3 is a program that someone wrote specifically for rendering Escher-like impossible scenes, as an undergraduate project in computer engineering. Pretty website, animated Escher objects, citations of mathematically abstruse papers, etc.

The Escher renderer was finished and published in October 2001, half a year before the discussion at halfbakery. For some reason, more than a dozen apparently intelligent people were unable to find what I found in half a minute at Google.

I don't really understand this. It's not like I'm using any specialized research ability, it's just simple Google. I've run into similar situations several times. Maybe people don't have the right intuitive judgement of how easy it is to find things on the Internet, so they never think to look?