Even further offtopic..)
...effects as wide-ranging as the Moon-tilt 'illusion'...
Haven't heard of that one, but I do know about the "gigantic" full moon as it rises behind a hill or distant trees.
It's my personal favorite.. Pick a time when both Moon and Sun are in the sky but are a reasonable distance apart, around sunrise or sunset. Look at the angle from which the Moon is illuminated and mentally draw a straight line back to where the Sun *should* be... It misses by a significant margin! (which to some tinfoilhatters is proof that the sky/sun/moon is a hologram.. It's really quite a weird effect if you've never seen it before.
The reason for the 'illusion' (it isn't really an illusion at all) is simple perspective distortion - the same thing you get when standing near and beneath that long straight wall or standing on railway lines, the same thing that causes (anti-) crepuscular rays to fan out from the sun overhead and yet re-converge at the vanishing point on the opposite horizon. Apart from those that run along the horizon and those directly overhead, those rays *cannot* be straight lines *from our viewpoint* - after all, they go from the two vanishing points spread by 180 degrees, but then fan outwards up to 45 degrees as they go over your head - no straight line could do that... 
But yes, of course the light rays do follow a straight line in 3d reality - it is the projection of that straight line on our 'spherical' - yet 2d - viewpoint that causes the curve (and it's a real curve..). I've probably described that badly, but it's hard to express in words..
Interestingly, the effect vanishes ('shifts' is a better word) if you alter your 'horizon' to match any curved ray, but by doing so, the actual horizon now becomes bent... The fascinating thing is that because we *always* perceive the horizon as level and straight, your brain doesn't like doing that and will do its best not to let you perceive the curve.
A really short lens (fisheye) will demonstrate that well: keep the horizon level, and every other horizontal line curves; select another reference, and the horizon curves. Of course, with that type of lens, there is deliberate barrel distortion in the entire image. Also, go to
www.widescreenmuseum.com and look at any of the Cinerama images where the camera was panned up or down and the horizon is visible - because the image is on a flat plane (your monitor), and not a curved screen, you'll see a bent horizon.
If you think about it, every lens that images anything less that 180 degrees is really magnifying the center of that fisheye image, so the curves become less and less apparent. What lens makers work so hard at is managing edge distortion by keeping a straight line straight while maintaining focus to the edges of the frame. Poorly made lenses will exhibit either pin-cushion or barrel distortion. Actually, a well made lens has deliberate distortion to satisfy our perception of reality.
If you want to prove this to yourself, take a camera outside and shoot a 180 x 180 panorama using a normal lens (35 mm for most modern digital SLR,'s), and try to stitch the images together, without using a photo-stich software. You will find that there will be discontinuities at the edges of the images where the lens maker kept the lines straight (and now you know why you pick joining references in that software - they force a fix).
It's the inverse of the problem of matching aeronautical or nautical charts at the edges because the lines of latitude and longitude were kept straight.
