Geoff the Medio wrote:
Why do you say self-shadowing is not relevant? The asteroid has a complicated concave surface with lots of places where a star-side prominence would cast a shadow on a flat area further away from the sun. Just using the normal to determine lighting only takes into account the local surface orientation, which makes the object look a lot less interesting and realistic when primarily lit by a signle directed light source.
From my experience, self shadowing is usually hardly noticeable(if absent), even more so in our case(example .avi
). I thought it's not worth it, especially with the idea of baking it into the texture and this alignement issue in mind. However, if it can be done as tzlaine described it and it's easy on hardware resources we should go for it.
Would it help if you didn't make so many frames?
It's still 9 (planet types) x 3 (color maps each) individual
light map sets. Multiplied by the races, if we are going to vary them between races.
Compare this to just 1 set per race, that gets individualized by masking out certain parts. Also I argue, that players won't look that closely at planets anyway. He wouldn't notice if the lights follow the terrain, especially considering, that the terrain isn't even visible at the dark side.
The light map growth won't be animated as in your example; at any given time, the player will only be shown a single frame for each planet. We don't really need so many farmes as you've shown... three or four per colour map would probably be adequate (plus a blank one for no development).
I've provided 10 steps. The animation is done by blending those. Not really that many, considering the population goes from 0 till 30 or something in some cases. Which brings me to the question, should those lights rather show the population level or maybe the infrastructure?