Aug 10, 2011

Okay Lighting Experts Bring It ON!!!

Well since I can't be very public about my one and only on going palaeo-art project, I figure instead to engage my very nice and helpful expert readership on how to improve my 3D lighting, as it has been a constant critique of the last while.

Here is my current methodology. I'm taking an object (in this case an unposed Dinosaur) setting it up in a preset lighting rig, and moving the lights around to see what happens. I'm currently playing with a "three-point theory" setup, but am open for other setup suggestions. I will be modifying multiple variables and aspects of my three-points as I can to see what happens.

Here is an animation to help visualize what I'm doing. The floating cones are proxies for the lights themselves (you can't directly see the lights in a final rendering). As they rotate you'll notice I've included the new angle on the right for reference.

After cranking out this first test version I realized that my lights aren't very well individualized. In future batches of these I'll be colour coding them and sizing them so they tell us more about what I've done.

For the time being red is my "Key Light", blue the "Fill Light", and yellow the "Backlight". For colouration the key light is a cooler colour (blue), fill is a warm, and backlight is a low grey. All these lights angled at 45 degrees to the ground from the sky, and are opposed to each other's alignment by 120 degrees. They all are "distant" lights, which is to say they apply light universally throughout the whole 3D scene (as opposed to a spot light or blub that function much like the real life lights that share those names).

Variables I'm planning on testing include strength and colour of ambient lighting, colouration of the three colour sets (which I've obviously started on), and angle from the sky.

Any suggestions on other test variable, light techniques, or things I've missed altogether?


davidmaas said...

If you look at the dinosaur itself, there is hardly any effect visible from the lighting. This makes me think that either the shadows are shadow-maps (as opposed to ray-traced) or that you have incredibly low diffuse color settings on your surface. If they are indeed shadow-mapped, that would explain the missing bounce light. You can fake those by putting a distant light on the ground facing up at the sky and the belly of your dino, lower intensity to ~10% and color it brownish.

Or you can use ray-tracing solutions. (I don't know of any package that isn't trimmed for speedy lighting with ray-tracing anymore.)
What are your global illumination settings?

Traumador said...

Hey David,

Well this is just a test of my test format and methodology. I'm about to unleash about 120+ case studies I've been rendering all afternoon (got to love easy 3D tasks one can do while watching seasons of Mythbusters DVDs).

So yes this particular version wasn't very statisfying. Part of the problem is I forgot I activated soft shadows (which are in my 120 renders). They are ray traced, though I'm not sure Carrara has bounce light...

I'll be hitting the web with my ambient experienments in about an hour. So hopefully that'll answer your question there.

Cheers for the feedback. I'm hoping this mass experinment/tutorial I'm about to launch is worthy of it.

Anonymous said...

I agree with David. Make sure your Keylight has plenty of kick, it represents the sun, or at least your main light source.

Your fill and bounce can have their shadows turned off. Bounce can come from a much lower angle, and be tinted slightly with the ground color.
Does Carrera have a dome or skylight? This would do the job for the fill light and you could turn shadows *on* to achieve ambient occlusion.