 |
How performance houngry is the hiding spot calculation actually? If the impact is not to high, the esiest way may be to recalculate it in specific intervals. |
 |
I've never measured it, but creating the list of spots doesn't seem to take long. 2 or more AI can share the same list, though, so the designer(s) must have been worried about how long 'list creation' took. Since it can generate hundreds of spots, and each spot must undergo an illumination test (which is time-consuming), it seems to be a very complicated way of telling an AI to "look here, then look there, then look ...". I'm not a big fan of the hiding spot mechanism, since it can cause odd search behavior, especially when 2 or more AI are searching. Eventually, I'd like to see it replaced with a mechanism that uses two bits of information: 1 - nearby architecture (where are the doors? where do nearby stairs go? if someone was here, where's the most likely place they would have gone?) 2 - how can I use my friends to best search the area? (squad behavior, posting guards at doors, sending someone to raise an alarm, etc.) |
 |
Hiding spot code was substantially changed in 2.04. Is this still relevant? |
|
|
The 'hiding spot' approach was abandoned in 2.04, though the code is still present but unused. Given the dynamic nature of the 2.04 changes, adding and subtracting lights during a search will automatically modify the search. This issue is closed. |
- Anonymous
