Placing 'glass_lightcover.mdl' into 128 unit spaces.
Hey folks,
I'm getting back into mapping here with Portal 2 after an approximate eon (last time I fooled around with this stuff was in the Team Fortress Classic days). I'm happy to find the basics are the same, but I ran into one oddity last night while putting together a really basic test chamber.
I've got a section of one 512 unit high wall cleared out (the cleared section is 128 units high x 64 units wide), and have a recessed brush behind it placed with the 'lights/light_panel_cool' texture applied to its inward face. This by itself compiles fine.
I then placed 2 prop_statics with 'models/props_lab/glass_lightcover.mdl' in front of the recessed wall, but the next time I compiled it immediately returned a leak because apparently the edge of the model clips into the surrounding brushes. (Actually sorry now that I think about it, glass_lightcover.mdl may not be the exact filename, it's one of the related glass_ models in /props_lab/ that's 128.5 units tall though)
The model itself looks like it's 128.5 units tall and I guess my question is: why the extra .5?
Do you really need to scale back the surrounding brushes by .5 to accommodate this model or is there something I'm missing here? I can post some screenshots later tonight when I'm back at my home PC if this is unclear.
Thanks for any assistance, I've found the posts here really informative.
There's nothing keeping models from intersecting with world geometry.
Please do not Private Message me for assistance. Post a thread if you have questions or concerns.
If you need to contact the staff privately, contact the Global Moderators via Discord.
What msleeper said. Models can collide with geometry and usually do. What matters is that the origin of the model is not completely outside of the map. Have you looked at the pointfile to see exactly where vbsp thinks the leak is?
Portal 1: Portal Pro, Resurrection, Dyad, Rexaura
Portal 2: [Coop] Electrophobia, [PTI] EotM Collection, [SP] Aperture Halloween
Ahhhhh, okay great. The origin placement makes sense, I'll give that a shot tonight and report back.
Oddly when I got the initial leak it didn't look like vbsp created a pointfile for the map at all. I checked the same directory I've saved the .vmf to but there were no .lin files in there. My map had a number of these prop_statics placed at that point, and I culled it back to two before determining that they were the issue. The original compile log reported something like 'too many errors' at one point, possibly due to all of these prop_statics with out of wak origins, so perhaps that aborted the pointfile creation since there were too many leaks.
In any case, I'll look into the origins here first and try to narrow things down. Thanks for the help guys.
Post your compile log next time it errors and we can try to help out more.
Please do not Private Message me for assistance. Post a thread if you have questions or concerns.
If you need to contact the staff privately, contact the Global Moderators via Discord.
Got it figured out. Looks like you need at least 24-32 units of space behind the model before your enclosing brush to avoid a leak. I initially only had about 16 units, once I made the recess deeper it sorted itself out. 
Thanks much.
Please do not Private Message me for assistance. Post a thread if you have questions or concerns.
If you need to contact the staff privately, contact the Global Moderators via Discord.