June 14, 1998. Over the weekend I was starting a model of wood-blinds for windows.
Although the model was going well geometrically as far as I could tell, it did not look
like I wanted. The slats were not catching the light from the other side of the window
I tried a couple angles of the sun and angles of the slats, with no avail. I figured I had not done anything wrong with the model as everything seemed to be working OK, so the effect had to be "real". I put it aside for the evening.
On Monday morning, I looked closely at the glowing wood on a set of real blinds and examined it closely. Changing the angle of the slats cycled through many different lighting effects. The exact effect I wanted would have to get the angle just right, so just poking around at it would not find it. That makes sence — It is the complex chaotic lighting that made me interested in it as art in the first place.
When I got to work on it again, the first thing I did was apply a "finish" to the texture. It seems that the stock wood textures are pigment only and are "unfinished". Just to make sure the defaults were just too dull to show anything interesting, I applied a reasonable guess at a "finish".
Then I reset the angles of the slats and the light to what I figured was in the right ballpark, and gave it a spin. The image at the left is a detail (see tip 1) shows the outside light only (lights inside were turned off) to better see the effect of the outside lighting. That's when everything clicked and I understood how to get the effect I wanted.
My first idea was to reach for some graph paper and a handbook on trigonometry. Then, remembering that I've been accused of being too much of a geek at times, I thought about a simpler way. The solution: look at the image to the left. The part I've cropped out shows the light beams coming through the window, and I can clearly see the angle of the light beams along side the angle of the slats.
It's clear, if you look carefully, that the visible tops of the slats are dark because each light beam misses the slat below. The light is too horizontal. I needed the sun to be a little higher before the slats would catch the light.
|Here is the resulting image. I didn't calculate anything, but just tried a few different sun angles until I got the effect I wanted. Isolating this element of the image (and turning off other lights) I could quickly (2 minutes) render this detail and — here is the key — get visible feedback as to whether it was "too high" or "too low". That allowed me to home in on the value I needed rather than just taking shots in the dark.|
For comparison, here is the final image with the inside lights on as well. The lesson
I learned is to look for clues within the rendered scene when possible.
It reminds me of something I heard about building shelves. A homeowner who fixed things up had always heard "measure twice, cut once". But he was asonished to see a professional building some shelves in a closet. The pro didn't measure anything ahead of time, but got measurements directly off the workpiece. Each board was cut to fit the space available, rather than being cut to a pre-planned length.