Plan for Stefan Beffy Moises
by Stefan Beffy Moises · 05/24/2004 (3:14 am) · 12 comments
Whee, finally navmesh generation for scaled interiors is working, too! :)

Remember how they've looked like in one of my last .plans?

I was trying to scale the objBox before placing nodes, but ran into some issues with rayCast collisions crashing the engine then etc.
So what I am doing now is I just moved the scaling to a later part of the process, right after the rayscans, before I am transforming the points in worldspace:
easy as that! ;)
Another thing I've been tossin in lately is path smoothing/straightening which is based on an article in Game Programming Gems 1.
The idea behind is pretty simple: walk through all your path nodes, checking for line-of-sight (LOS) between them, so say you have 3 points A,B and C - if you can "see" from A to C (without colliding with anything), you can skip point B and walk straight from A to C.
Here are some pics:
Before:

After:

Before:

After:

It's maybe not that clear to see, but e.g. the second path has around 20 nodes "after" compared to about approx. 120 nodes "before", which makes the movement a lot smoother and straighter...
Next on my plate are dynamic collision avoidance probably using Phil's sensors and further navgraph generation optimization...

Remember how they've looked like in one of my last .plans?

I was trying to scale the objBox before placing nodes, but ran into some issues with rayCast collisions crashing the engine then etc.
So what I am doing now is I just moved the scaling to a later part of the process, right after the rayscans, before I am transforming the points in worldspace:
if(!interior->castRay(start, end, &ri))
{
PROFILE_END();
return;
}
Point3F worldPoint = ri.point;
// beffy: fix for scaled interiors
VectorF scale = interior->getScale();
worldPoint.convolve(scale);
interior->getTransform().mulP(worldPoint);easy as that! ;)
Another thing I've been tossin in lately is path smoothing/straightening which is based on an article in Game Programming Gems 1.
The idea behind is pretty simple: walk through all your path nodes, checking for line-of-sight (LOS) between them, so say you have 3 points A,B and C - if you can "see" from A to C (without colliding with anything), you can skip point B and walk straight from A to C.
Here are some pics:
Before:

After:

Before:

After:

It's maybe not that clear to see, but e.g. the second path has around 20 nodes "after" compared to about approx. 120 nodes "before", which makes the movement a lot smoother and straighter...
Next on my plate are dynamic collision avoidance probably using Phil's sensors and further navgraph generation optimization...
#2
05/24/2004 (6:41 am)
very nice
#3
05/24/2004 (7:05 am)
imporessive. props all the way.
#4
I'm cuirious if you plan on releasing this when its done?
05/24/2004 (7:44 am)
Looking better everyday!I'm cuirious if you plan on releasing this when its done?
#5
@John: yeah, we will release it in any case, but we have yet to decide in which form... as a free resource or as a content pack type thingy or whatever ... and since there is quite some work involved for the 4 of us (and we gotta eat etc. ;)) we *tend* to the second option ;) but we haven't finally decided yet... we'll keep ya posted :)
05/24/2004 (8:09 am)
Thx guys :)@John: yeah, we will release it in any case, but we have yet to decide in which form... as a free resource or as a content pack type thingy or whatever ... and since there is quite some work involved for the 4 of us (and we gotta eat etc. ;)) we *tend* to the second option ;) but we haven't finally decided yet... we'll keep ya posted :)
#6
05/24/2004 (8:24 am)
Nice work, Beffy. I love .plans with pictures in them! :)
#7
05/24/2004 (8:28 am)
Considering how big of an enhancement this if, I think you SHOULD charge for it. I'd pay $20 bucks or so for it. Couple things to keep in mind though if you do decide to charge. It will be expected to be nicely polished, easy to add in and use, and people would normally expect full support as well. Not that that should deter you from charging. I say go for it :) We have some tools/enhancements in development that we plan to release commerically as well.
#9
05/24/2004 (1:11 pm)
I said $20 bucks OR SO :p
#10
05/25/2004 (12:16 am)
Wow! I would pay for this as a content pack, but I'll let John do the negotiating ;) This would be a must have for any TGE owner. The smoothed paths look great.
#11
05/25/2004 (3:00 am)
If this wasn't specific to TGE, as in being able to feed a tool generic topology. I would be interested in it. I imagine others would be too.
#12
05/25/2004 (3:00 am)
Edit: Double post
Torque Owner James Laker (BurNinG)