Cogmind > Bugs

FOV Could Use Improvement

(1/2) > >>

Cogsnack:
You should always be able to see down the length of a straight wall when standing next to it. Try it IRL. Make fully occupied cells (like ones with walls) more permissive. Reason is you only need to see an edge or corner of such cells to identify them.

A solution: If cell is fully occupied and diagonal cell closer to viewer is in FOV then permit (handle for each quadrant).

Kyzrati:
That's essentially how FOV already works, at least out to a significant distance. There must be some other factor at play in your particular screenshot there, something special about those particular cells? (Not sure. What's the seed and map? Or better yet, a save there :P)

Whatever it is, FOV is symmetrical in all directions so it's obvious something is going on in that particular setup. Walk around more and you'll see that your image doesn't reflect normal behavior. See here for an example of standard symmetrical behavior and wall distance:

Cogsnack:
Yes current algorithm works fine in most cases. Every so often I encounter this issue however. So far noticed in SE and NE quadrants. After encountering I investigate and walls seem to be normal, so I don't know why it happens.

Note: Additional requirement for solution above is that diagonal cell does not block line of sight.

Kyzrati:
Right, again it could be the map, which is why I asked for a seed. The Mines screenshot is not normal behavior, either. (I haven't seen that before and it hasn't been reported in years of plays, so it's rather odd that you're seeing this.)

I know the FOV algorithm itself is solid so there shouldn't need to be any changes there. Gotta be something else...

Cogsnack:
I don't have a reliable seed as those shots may have come from earlier runs, but I do know they are Beta 4 and/or 5. Will send seed in future encounters. Using uninitialized array elements during FOV calc would be one suspect.

Navigation

[0] Message Index

[#] Next page