Even if it wasn't, the nature of floating-point is such that very close values are never going to result in reliable physics.
I see what you mean. (to be more specific:) I know the recess was less then 16 units wide since the upper edges of the brushes caught the ball first. Then the texture on the basic-ball seemed to vibrate more rapidly than on the other two recesses, which were narrower. Next, the ball clipped through the brushes after I had tilted the level for a couple seconds. Maybe also the slant of the edges (obliqueness to the axes of the level in Radiant) as something to do with this?
In my level, "Pyramid Zone," there are quite a few ramps with the same reflective texture on the top surface. There is a certain section where the space between the floor and the bottom face of the brush above is the same as the diameter of the ball, and several of these ramps are visible. In playing the level on a friend's machine, the game lagged a lot when I rolled in this section. This could mean that rolling in gaps the same width as the ball would be more cpu-taxing, or something (emphasis on this word).
Challenge Space Yard wrote:
I can either (1) make the tunnels slightly wider (making them easier in the process),
Wouldn't it be negligible to draw back each edge of the chasm 0.125 units? If you use NetRadiant, and if this does not go past the 1 unit grid, I would do this:
1. Draw a 1 unit large cube.
2. Use "arbitrary scale" and set the x/y dimension to 0.125.
3. Hide all brushes other than this brush and the other brushes of interest.
4. For each brush, place the auxiliary brush inside the edge and use CSG Subtract.
--It the it
It the it's level set: http://neverforum.com/fmpbo/viewtopic.php?id=29226/19/17: a bit of help needed here, plz: http://neverforum.com/fmpbo/viewtopic.php?id=2988
I created my avatar with ANVIL-1000MD (c) and scaled it down with GIMP.