Coblet
Jun 15 2004, 12:48 PM
Right then, lets structure this correctly:
The aim of this topic is to discuss whether or not this would be possible.
My aim therefore is to:
a) Clarify what I think this would entail
b) Ask if it's actually possible
c) Complain when it isn't and cause ALL WALLS EXPLODED
So compilers can check the location of leaks through grid references etc, and by seeing what entities are in the sightline of the leak. Therefore surely it would be possible to specify the (two) exact brush(es) that is/are nearest to the leak/leaks and therefore give a brush number feedback to the user.
However, it doesn't end there. If the compilers could calculate grid distances (just one unit grids ala TEH HAMMRE) then surely the compiler would be able to calculate the size of the leak, determine which brush is nearer and then (you guessed it) translate the brush so that the leak is saved.
Of course this would probably create wierd results when coupled with entity leaks (such as func_door as a barrier to TEH VOID) but perhaps it might be worth thinking of something like a brush equal to the size of the door, textured black, being placed through the door to make it plugg'd.
So then, thoughts? I'm not sure how this would go about (as I have no real knowledge of coding), but I am assuming that some of the code would be somewhat similar to normal leak finding.
Trojan
Jun 15 2004, 12:55 PM
Im not one to really be answering this, but an educated guess would say this would be near impossible. I have a <0.01 unit gap somewhere in my map, the pointfile has over 3600 points (x,y,z) in it. The map is only but a room and a few corridoors so far... Trying to make a program figure out which one is the leak, and which polygons to change, doesnt sound too feasable to me
Coblet
Jun 15 2004, 01:01 PM

Well I wanted to help you out Trojan, thats why I made this
-13:28:49- (Trojan) man, screw world peace, if I ever had a wish it would be to find every little 0.0000000000001 unit leak that ever pops up
-13:28:49- (Trojan) ever
Trojan
Jun 15 2004, 02:12 PM
Yeah I figured as much, still, it would probably help everyone

Not just me
Freezer
Jun 15 2004, 02:29 PM
We have tools that automatically create detailed textures,
we have tools that make particles easier to make and place,
we have tools that export better to map type,
we have tools that compile twice better than old ones,
we hove tools that fit several parts of maps together.
I think thats enough.
The idea isn't bad but at some point mapping will be almost automised.
Isn't it fun to find leaks? Everytime you see one you know what you can make better next time.
Your idea is not bad but

for me
BlackPlague
Jun 15 2004, 02:36 PM
i think i use somethin to find leaks
my eyes! :-o
Coblet
Jun 15 2004, 02:37 PM
| QUOTE (Freezer @ Jun 15 2004, 03:29 PM) |
We have tools that automatically create detailed textures, we have tools that make particles easier to make and place, we have tools that export better to map type, we have tools that compile twice better than old ones, we hove tools that fit several parts of maps together. |
Most of these aren't about automating the mapping itself... they're about making the map playable. Therefore your point about mapping becoming totally automated is unjustified.
Yamazaki
Jun 15 2004, 02:50 PM
Grab Hammer 3.4 and you'll be able to load up pointfiles within the editor. Then you just need to start from the entity that reported the leak and follow the red line until it goes through the hole, doesn't take long at all.
Keep in mind that this is more than most editors give you. UnrealEd for ut2k3 and ut2k4 will just crash to desktop when you ask it to Build with bad BSP geometry. No error, no indication of what the problem is, just BOOM... desktop. Hope you saved
brute_force
Jun 15 2004, 02:53 PM
I think leaks are easy enough to find right now. We have pointfiles, I do believe QuArK has an inbuilt leak-finder, you can turn on flat shaded to more easily find leaks, you can run a leaked map to see if you can spot the HOM, you can change the background colour to pink so it will stand out more, there is the box method, you can check the entites that the compiler reports, etc. I think a tool like this wouldn't really be very useful, since I can find most of my leaks within 5 minutes (often less than that to, I'd say my record is 30 seconds). It would only help the lazy people or the complete newbies. Better for them to learn how to solve it by themselves instead of using an automated process.
NEO_Phyte
Jun 15 2004, 04:36 PM
are you saying you ENJOY spending however long on finding a leak just to discover that those brushes that look fine in hammer actually have a miniscule gap in them?? (once took me ~30 min to find one of them)
brute_force
Jun 15 2004, 05:53 PM
| QUOTE (error 404 name not found @ Jun 15 2004, 05:36 PM) |
| are you saying you ENJOY spending however long on finding a leak just to discover that those brushes that look fine in hammer actually have a miniscule gap in them?? (once took me ~30 min to find one of them) |
No, because I do not make invalid brushes. And even if I do, I would check the .map file with the .lin instead, and then it would be just like a normal 1 unit leak.
Vahn_Paktu
Jun 15 2004, 06:12 PM
Wouldn't it be possible to make a program to do the box method for you?
Coblet
Jun 15 2004, 06:21 PM
There is already a program for people to make horribly inefficient maps in, it's called Hammer. A simple look through some of the latest posts/some mapping forums will confirm that we don't need another.
Jean_Luc_Picard
Jun 15 2004, 09:08 PM
But, the problem is, when hammer finds an invalid brush, if you let it auto-correct it (I made this mistake b4) it moves the brush EVER so slightly off the line... by maybe .000000000000000000000000000001 units... THAT is enough to form a leak... and the pointfile doesn't help because it becomes a mass of like 300+ lines randomly scattered along the entire leak line... and the leak hole is too small for human eyes to see anyway... so
DarkSeraph
Jun 15 2004, 09:38 PM
if you can tell which boxes the leak is in between, this should be feasible, and just make a program to find the boxes where the leak is, and expand them ever so slightly. would prolly have to be editor based. but i dunno how to find which boxes leak is in between.
Ripzaw
Jun 15 2004, 10:00 PM
The HL2 thing is the best... Every error it finds it replaces the errored model/whsprite/whatever with "ERROR!" in giant red letters. Pretty hard not to find them tbh.
DarkSeraph
Jun 15 2004, 11:08 PM
i know this is off topic, but is NS gonna be in HL2? and/or is there going to be a map editor? maybe if i make my map for HL2 ppl won't **** about the w-polys as much
Cronos
Jun 15 2004, 11:17 PM
A program was made a while ago that automatically generates random maps, so mapping is technically ALREADY fully automated.
Oh wait, why hasnt randmap replaced the mapper outright? Oh thats it, a mapper is a thousand times better then that hunk of crap.
I'm for this idea personally. I've been stuck trying to find leaks before, and at one point I had been hunting a leak for four frigging months.
DEATH TO LEAKS! DEATH TO LEAKS! DEATH TO LEAKS!
GiGaBiTe
Jun 16 2004, 08:08 PM
well why dont we go about the approach to finding leaks the quake III way. if there is a leak in your map, the editor finds the closest entity to the leak and draws a fat red beam from that entity and it continues twisting and winding untill it goes through the leak point and out in to the null space. i would show you this but i havent messed with q3 maps in so long that the editor dont work anymore.. lol
Recoup
Jun 16 2004, 10:05 PM
You know, this is why I use point files....
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please
click here.