Map2Dif Problem : HEAD
by Prairie Games · in Torque Game Engine · 04/20/2002 (4:38 pm) · 10 replies
*EDIT*
The below has been remedied... I have had some good success with map2dif now... HEAD is touchy about missing resources...
From my post below: Is something more graceful possible? Maybe just a declaration of WHAT is missing before bomb? Would help in wrangling...
*END EDIT*
Compiling any .dif with map2dif (either 1_1_1 or HEAD version of the tool) and then loading it into the HEAD demo app causes a crash after a complaint about a network incorrect version...
The 1_1_1 stable release bug of coming in contact with an unlit interior causing a crash makes it somewhat unusable.. I will hunt around for a fix to this... if there is one, it would have been nice to find it already in the CVS..
Inital tests show that map2dif only works with _extremely_ simple architecture... it barfs on architecture that works in Quake2... even the sample .maps included in the sdk giving warnings about orphaned polys? I have experienced quite flakey results with this tool... I have written some BSP compilers myself and know how touchy the data is...
I could be wrong... but after feeding in a range of .map files I doubt it... it seems I either a) need to be satisfied with primitives or b) work out a solution outside of map2dif...
One example of what seems a very robust compiler is at: http://www.fluidstudios.com/publications.html , this bsp compiler could likely be worked into Torque... and may give better results...
Another thought I have is to alter the Quake2 tools to work with Torque... they are pretty darned solid... the problem is that they are $5k to license... woot!
Then there is this.. lets say I have some complexish data... would the runtime system choke on it?
-J
The below has been remedied... I have had some good success with map2dif now... HEAD is touchy about missing resources...
From my post below: Is something more graceful possible? Maybe just a declaration of WHAT is missing before bomb? Would help in wrangling...
*END EDIT*
Compiling any .dif with map2dif (either 1_1_1 or HEAD version of the tool) and then loading it into the HEAD demo app causes a crash after a complaint about a network incorrect version...
The 1_1_1 stable release bug of coming in contact with an unlit interior causing a crash makes it somewhat unusable.. I will hunt around for a fix to this... if there is one, it would have been nice to find it already in the CVS..
Inital tests show that map2dif only works with _extremely_ simple architecture... it barfs on architecture that works in Quake2... even the sample .maps included in the sdk giving warnings about orphaned polys? I have experienced quite flakey results with this tool... I have written some BSP compilers myself and know how touchy the data is...
I could be wrong... but after feeding in a range of .map files I doubt it... it seems I either a) need to be satisfied with primitives or b) work out a solution outside of map2dif...
One example of what seems a very robust compiler is at: http://www.fluidstudios.com/publications.html , this bsp compiler could likely be worked into Torque... and may give better results...
Another thought I have is to alter the Quake2 tools to work with Torque... they are pretty darned solid... the problem is that they are $5k to license... woot!
Then there is this.. lets say I have some complexish data... would the runtime system choke on it?
-J
#2
(contact with unlight interior)
which im pretty sure Mark fixed
and you are talking about a geometry export problem
I would like to see the geometry that you speak of.
Other than the obvious bsp limitations I find no problems exporting geometry.
04/20/2002 (4:49 pm)
if your not talking about the lighting bug (contact with unlight interior)
which im pretty sure Mark fixed
and you are talking about a geometry export problem
I would like to see the geometry that you speak of.
Other than the obvious bsp limitations I find no problems exporting geometry.
#3
When I compile the sample maps from the sdk.. some of which are only a cube or two.. there are warnings about orphaned polygons... perhaps these warnings are extraneous...
I need to test further... all bsp's aren't created equal btw... a giant leap in stability was had between quake1 and quake2... by a subtle change in the way the bsp was calculated... brush based vs. strictly by planes... the collision system is MUCH less flakey in quake2... in addition, the bsp compiler allows more complex geometry... quake3 uses an all but identical solution
If there is a problem with map2dif in the sense of it not allowing geometry that the runtime can support.. an alternative would be a good thing...
I don't want to lose the part about the HEAD version not being able to use any .dif as compiled by the HEAD or 1_1_1 version of map2dif... so here it is again :)
-J
04/20/2002 (5:04 pm)
I have fed the utility some quake2 maps/submaps... which have caused map2dif to die(winding/plane problems).. or output nonworking geometry... these .map files are more complex then what I have seen in the Torque engine... When I compile the sample maps from the sdk.. some of which are only a cube or two.. there are warnings about orphaned polygons... perhaps these warnings are extraneous...
I need to test further... all bsp's aren't created equal btw... a giant leap in stability was had between quake1 and quake2... by a subtle change in the way the bsp was calculated... brush based vs. strictly by planes... the collision system is MUCH less flakey in quake2... in addition, the bsp compiler allows more complex geometry... quake3 uses an all but identical solution
If there is a problem with map2dif in the sense of it not allowing geometry that the runtime can support.. an alternative would be a good thing...
I don't want to lose the part about the HEAD version not being able to use any .dif as compiled by the HEAD or 1_1_1 version of map2dif... so here it is again :)
-J
#4
maybe try for the other geometry to export it as a quake3 map so in your map2diff command add -q 3 to the compile line.
Ive not had this problem, then ive not tried to use any old quake maps..
I agree they are not all equal ..
but I would still like to actually try this map file you speak of myself
Im using the Head now and im working on the map2diff adding door support .. Ive had not problems working with it and running the diff's in the engine
PS if you Do have a problem and Have turned up a bug ..
Im sure between you and me we could fix it :)
so lemme know either way.
04/20/2002 (5:17 pm)
Ok..maybe try for the other geometry to export it as a quake3 map so in your map2diff command add -q 3 to the compile line.
Ive not had this problem, then ive not tried to use any old quake maps..
I agree they are not all equal ..
but I would still like to actually try this map file you speak of myself
Im using the Head now and im working on the map2diff adding door support .. Ive had not problems working with it and running the diff's in the engine
PS if you Do have a problem and Have turned up a bug ..
Im sure between you and me we could fix it :)
so lemme know either way.
#5
Also, map2dif can handle almost anything you throw at it (up to 32K surfaces that is). I've converted many Half-Life, Q2, Q3 maps with no problem. But, it won't do bezier patches--that might be what's causing your problem.
04/20/2002 (7:17 pm)
The orphaned polys probably come from previously working with GTK or QERadiant to build your Q3 maps. GTK doesn't eliminate broken polys which leads to the orphaned poly message by map2dif. I've debugged lots of GTK constructed maps and thats one of the first things I look for (I use QuArK btw).Also, map2dif can handle almost anything you throw at it (up to 32K surfaces that is). I've converted many Half-Life, Q2, Q3 maps with no problem. But, it won't do bezier patches--that might be what's causing your problem.
#6
Message Box:
You do not have the correct version of the Torque Game Engine or the related art needed to connect to this server, please contact the server's operator to obtain the latest version of this game. (Invalid Packet.)
Followed by an immediate bomb...
or this in debug:
Message Box:
Invalid connection objectID with a bomb at 760 in sim/netconnection.cc
***NOTE*** The .dif files included in the CVS work...
As for my map compilation problems... I'll try some more once I get this solved... the 1st thing I usually do is try and see how hard I can push this kind of tool... one that is so central to data flow...
-J
04/20/2002 (8:39 pm)
I make a .dif with either 1_1_1 or the HEAD as of yesterday... I launch the test app 1_1_1 and it works... I launch the test app HEAD and I get this in release:Message Box:
You do not have the correct version of the Torque Game Engine or the related art needed to connect to this server, please contact the server's operator to obtain the latest version of this game. (Invalid Packet.)
Followed by an immediate bomb...
or this in debug:
Message Box:
Invalid connection objectID with a bomb at 760 in sim/netconnection.cc
***NOTE*** The .dif files included in the CVS work...
As for my map compilation problems... I'll try some more once I get this solved... the 1st thing I usually do is try and see how hard I can push this kind of tool... one that is so central to data flow...
-J
#7
Nice stuff guys... really nice stuff...
Still can't load .dif files I make into the HEAD test app.. same problem as relayed above..
-J
04/21/2002 (1:15 am)
Hm.. GTKRadiant was hiding clips on me(I don't know the tool very well)... successfully imported Q3dm1sample.map into torque.. well kind of... anyway.. I tilted it on the x,y,and z axis.. try that in your daddy's Quake.. nope... I like it! whee!Nice stuff guys... really nice stuff...
Still can't load .dif files I make into the HEAD test app.. same problem as relayed above..
-J
#8
Now that I have valid textures HEAD works...
-J
04/21/2002 (2:05 pm)
Problem was on my end... 1_1_1 allows for invalid texture information in dif... HEAD crashes if texture information in .dif is invalid...Now that I have valid textures HEAD works...
-J
#9
04/21/2002 (2:39 pm)
That would be because of the new download code that was added om the HEAD builds.
#10
-J
04/23/2002 (12:46 pm)
More user error.. head is pretty darn touchy about missing resources... something more graceful possible? Maybe just a declaration of WHAT is missing before bomb? Would help in wrangling...-J
Associate Ken Finney
Tubetti World