The tree is going to be made of many separate objects. And if you start modeling those individual leaves individually, you will soon see little puffs of smoke coming out of your CPU after several hours of grinding.
A realistic tree, or any other complex object, demands that you pay close thought to “what says ‘tree’ to me?” And then, “how can I get this effect, to an acceptable degree, while doing as little work as possible?” Welcome to the fine art of “faking,” or “cheating,” a shot.
As an illustration of my point, obviously what’s going to be the bugaboo is “those damned leaves.” Calculating the bounce of light over hundreds or thousands of leaves, one at a time, will take hours. :o Hours that you just don’t have.
This approach is “anatomically correct,” to be sure, but always remember that what you need is not a correct model of reality: what you need are a few million pixels on a rectangular video screen! If you can get there, acceptably and quickly, without laboriously modeling “reality,” so much the good.
Perhaps therefore you dream up the idea of taking a sphere, crunching it up by selecting pixels and scaling them in and out, then grabbing the proportional-editing tool and mashing around some more, then making the whole thing transparent and plopping onto it a texture (mostly transparent) which consists of an image upon which you have loosely spattered a bunch of squares of various shades of brown and green. Yes, you are effectively “spray-painting a chicken-wire tree and tossing a hunter’s camoflauge net over it!” It works in the movies; it will work here. The computer can solve that sort of problem very quickly and it will give you a whole lot of very realistic-looking texture, especially if you arrange your depth-of-field so that most of it is out-of-focus anyway. One or two in-focus leaves, right in the foreground close to the camera, will provide all the visual information that the viewer needs to interpret all those other blobby shapes out there as “leaves.” But the computer does not have to render them. Take it as far as you can go. A little more spray-paint on a transparent plane, rocking back and forth in front of the camera maybe halfway through the tree, will suggest just-enough motion of leaves to imply a windy day, and no one will notice that the other leaves don’t move with it.
Never forget that Star Wars Episode IV actually shipped to theatres with the entire crowd in the pod-race scene replaced by a bunch of painted Q-tips. The shot lasted about two seconds, and was completed in about two seconds, and no one noticed. (It’s been fixed on the DVD.) Get really good at cheating. It’ll save your schedule. Get the shot into the can.
Another useful trick is the “highly detailed close-up.” If you want to establish some particular detail of the tree, CUT TO a one-or-two second close-up of that particular area, showing the detail (the rest of the background will be fuzzy since this is a close-up), then CUT TO a medium-shot from the same or a very close-by camera angle. The viewer will “remember” the detail and will “see” it there, but in your medium-shot it does not have to actually be there. The viewer’s own memory is a great time-saver. The reason why the audience never noticed those Q-tips is that they had already been shown what was “there.” In the subsequent passing-shot that I spoke of, they “saw” alien creatures, not Q-tips.
