I’ve been thinking about this topic for about a week now, and I can’t seem to wrap it up into a neat package. So I’m going to just sketch out the issue and then plan on revisiting it in later postings.
For photographers trying to learn how to make realistic composite images, the single most difficult challenge, the skill that they most totally lack, is how to manually apply light to a scene. I separate lighting into two categories; cast shadows and on-object or self-shadowing. In fact, a shadow is a shadow is a shadow. It’s simply a light deficit relative to the rest of the scene regardless of where it happens to be.
However, when you go about creating lighting, manually, the two categories are profoundly different in technique, importance and in difficulty. Especially in difficulty.
Because the viewer of your pictures usually can’t see the light sources, and can’t see the figures in the round, you have a tremendous amount of creative leeway in how you make your cast shadows. In addition, even an obviously wrong cast shadow doesn’t necessarily wreck a picture.
On the other hand, self-shadowing makes the form on which it is applied (tonally, by you). By that I mean that in a two dimensional picture, the on-object shadowing is, quite literally, the ‘material’ of that shape. Change the shading and your change the form.
Given that one of the rules of compositing (one of my rules, anyway) is that you don’t make composites of anything that could just as easily be done in a ‘straight’ photograph, how does one know what the on-object shadows should be for an invented scene? It’s a conundrum.
You have to start really looking — and really seeing — what light looks like on everything around you. Study all kinds of forms in all kinds of light. This is something that photographers probably think they do all the time when in fact, they never really look at the light or the form. They respond to the light and the form; they don’t understand or know it — because they don’t need to. Photography is emphatically not about ‘making’ anything. It’s about choosing. Even when one works in a studio setting I believe the lighting is usually done by a combination of rote placement and reactive response as opposed to knowing, before the lights are even turned on, what will happen with given light/object combinations.
In addition to really looking at form and light, I would suggest hands-on modeling. You don’t have to make anything, just get some clay and play with form and deformation under different types of light.
Also, if you are (very) comfortable with computers, and have the time and desire, 3D modeling applications can be instructive in the same way as hands-on modeling. Please note that these are absolutely not for computer novices. They make Photoshop look like kindergarten. And most of them are very expensive.
I have experience with Lightwave (don’t particularly like it), Poser (easier but limited to pre-made models), and Rhinoceros (which I love, but have not kept up with). The latter is a NURBS modeler, which means you can make freeform models. It requires the associated Flamingo app for raytrace and radiosity rendering.
Such 3D applications are useful because you can make, light and render imaginary shapes or you can light and render human, (other) animal, or inanimate models in whatever configuration you like. They are also instructive in what fail to do. They never look really real.
If you visit the photorealistic Poser gallery page, notice how even the best human models look fake. Again, this is because the smallest details of self-shadowing make a tremendous difference. Those models don’t have living, changing, responding muscles in their ‘bodies’ and you can tell this by the rigidity of the on-object shading. This is true even of what I believe are the best human model/texture combinations — which are found at Daz.
For further negative proof of the importance of on-object shading, look at the sample gallery from the Retouch Shoppe which was linked from The Online Photographer, not too long ago. In the After images, notice how ‘homogenous’ the body shading is. They didn’t know how to manually apply lighting according to the anatomy of the reformed image.
Small differences in on-object shadowing make big changes in that objects appearance, and they are changes that are of critical importance to the success of the image.
On a tangent to the above, back (many years ago) when I was really into using Rhino, I used to read their forums (which are excellent). One person posting there was a woman named Bathsheba Grossman who makes sculptures using Rhino. I ended up buying one of her small chrystal sculptures and I really enjoy her stuff. Who else has made a sculpture of the entire universe, or of various protein molecules or of the “Schwartz’ D Surface (triply periodic minimal surface) truncated to a rhombic dodecahadron” — ummmmm… yeah, like I know what that is — but it’s very beautiful.