 |
This still isn't going to help when the player is crouching behind a box. There'll be no LOS from the AI to the lantern stim for the AI to see it. The AI would have to be able to discern a change in light levels splashed onto surfaces he can see. We have no way of doing that. |
 |
This certainly isn't worth much hassle (and not worth thinking about until after 2.0), but can't visual stims be given a large bounding box? |
|
|
The bounding box (or radius) decides which AI are near enough to be stimmed. In the case of light stims, the AI has to have LOS to the light. Otherwise, he'd be seeing lights in other rooms. Gas and water stims permeate through walls, and LOS isn't used. I recently changed gas stims to use LOS, so that the gas doesn't travel through walls, but does travel through wall openings into neighboring rooms. To do what you're describing, we could create a new stim model that extends the gas stim model to see if there's a LOS path from ANY spot inside the lantern's bounding box to the AI. This would require multiple ray traces from the AI to some subset of these spots (since we obviously can't check all spots w/o freezing the computer), looking for one that has LOS to the lantern and LOS to the AI. So if the lantern splashes light on the floor next to the box the player is hiding behind, if there's a LOS from some spot on that floor to both the lantern and the AI, the AI could notice it. Also, current stim behavior uses delays between stim firings, also to keep the CPU from being overloaded. We'd have to guarantee that the lantern stim fires in the frame where the lantern turns on, then uses the delays if the lantern stays on. That's the only way the AI would see a lantern that's only on for 0.1s. |
|
|
Probably not worth the bother then. |
- Anonymous
