We had a thread about reflections recently, and discussed how they work, and especially where the reflected object either does not have a flat lower edge, and / or is at differing heights from the reflecting surface. With that in mind, this image from a camera club field trip is fascinating, and interesting to analyse from a 'how reflections work'.
It underlines that the reflection in the water, is the view as seen from that position, so that you can't see the vineyard posts, or at least, you can only see the upper few inches. You can't see a reflection of the flax plant behind the boat, because the point where that reflection would fall (relative to the camera's PoV) is so close to the boat that the boat shields it. The shadows are interesting as well. Where the ground behind an object is flatter, the shadow is longer than where that ground slopes up, which makes the shadow length shorter. As was mentioned in the other thread, Bert Monroy uses Illustrator to develop where shadows and reflections would fall in a real world situation. Analysing a scene like this is great practice for anyone interested in making composites.
The image is pasted full size so you can click to expand, or copy into Photoshop.
You have to look at it this way on a mirrored surface: angle of incident equals angle of reflection. We all learned that. But if you look at it from a perspective view point, you are actually looking at the reflected subject as if that reflected angle kept going under the lake and at the same distance that that the viewer was from the lake. So in effect, you are actually looking up at the subjects rather than down at them:
So if you want to create a reflection, it's actually pretty difficult to do accurately, as the reflection perspective is normally below ground or water, and just flipping images doesn't really work. I tried to do a reflection photo, but I used two cameras, but later, I would realize that I needed to have one further away and not at the same distance. The image required a lot of tweaking to get lined up, and that was before puppet warp.
We can only make a best estimate of the angles as we don't know, in this particular case, the camera's height above the reflecting surface, and the angle of the slope the camera is on, and thus its distance from the lake. The bottom line here is that it is not always easy to work these things out, and that it is good to study reality when we get the chance, so we can make a better job of faking it.
Clouds are another thing that I think people tend to get wrong. Ask someone to paint in a false cloud, and the chances are they will underestimate how dark the shadows on its lower surface should be. A decent cloud illustration needs a bunch of layers help with whow its depth and the shadows withing a single cloud cluster. It's interesting stuff — I wonder if Matt Kloskowski covers this stuff in his Photoshop Compositing Secrets?