Game Development Community

Portals lighting up and seams showing in .DIF

by John Spivey · in · 03/10/2006 (6:14 am) · 9 replies

Been playing with the TLK last night and placed a light entity into a small interior that was portaled off. The model was scaled up by 10 (only way to get these huge models to run in TGE without problems) but i have noticed this problem as well on non-scaled models.

Everywhere there is a seam in the models you can see a line of light comming through from a distance. You get closer and it disapears. Also the portal brushes sometimes give the area where they are at a "TRON" kind of lighting accent. Not consistantly, but when you move around the entrance to the interior you see this effect that the lights have on the portal. Coupled with the light leaking out through the seams of the model it just looks terrible.

I suspect that this is something simple I am overlooking but I cannot find it on the boards or by searching.

Any help would be greatly apreciated.

#1
03/10/2006 (6:30 am)
John, I have had to change my modelling techniques to avoid this kind of behavior. Bevelled corners and joins instead of straight-aligned sections, etc. So I am in the same boat and will keep an eye on this thread if someone posts some tips!

One other thing I just thought of is to slip a small poly barrier into the wall at the seam, a spline or wafer-thin cube which would block light from coming through the seams.
#2
03/10/2006 (8:16 am)
Beveled Corners? Gah, map2dif gets the co-ordinates wrong alot as it is with placing brushes, beveled corners means even more headaches.
#3
03/10/2006 (8:21 am)
Hey guys, can you post some screen shots, I'm not following what the problem is - especially the portal/TRON effect.

Are you talking about the light leaks that Torque can have? These are generally cause by low lighting detail (lighting scale is too high) and thin walls.
#4
03/10/2006 (8:27 am)
Yep, beveled corners. Works great, saves faces too.
#5
03/10/2006 (9:13 am)
**double post** hehe dont ask me how but it did it
#6
03/10/2006 (1:26 pm)
i50.photobucket.com/albums/f338/jbsss/Tron_movie001.jpgTRON the movie......

i50.photobucket.com/albums/f338/jbsss/tron_light.jpgThe effect I was talking about.

LOL! I was wondering if my analogy was off after you said something but nope I was pretty much on. =)

On some models this effet is on every seam. I am currently working on a 8 grid with quark and this particular model is scaled up by 10. I am using Quark as the modeler.
#7
03/10/2006 (1:37 pm)
i50.photobucket.com/albums/f338/jbsss/light_effects002.jpg
Here is another example of the "tron" =) effects I was talking about. The intensity is at half and this is a portaled off area.
#8
03/10/2006 (2:24 pm)
Ah yes, those are TGE's dynamic lights. They use a technique to avoid z-fighting that creates small 'edges'. These are not as noticeable when using small radius dynamic lights, which are recommended also for performance considerations.

If you change the light to static (which you should anyway if it isn't animated or moving) those go away. Generally the static lighting looks much better, because it's calculated on mission relight instead of every frame and the engine can take the time to them look awesome.
#9
03/26/2006 (9:28 am)
Part of the problem with scaling your difs is that you might create the interior originally with integer vertices, but the scaling process might produce vertices with decimal values. At that point, I suspect rounding errors might be producing the small edge gaps you are seeing.

But why are you having to scale to begin with? Why not build it that size?