News | Forum | People | FAQ | Links | Search | Register | Log in
Screenshots & Betas
This is the place to post screenshots of your upcoming masterpiece and get criticism, or just have people implore you to finish it. You should also use this thread to post beta versions of your maps.

Need a place to host your screenshots? Upload them here:
http://www.quaketastic.com/
Username: quaketastic
Password: ZigguratVertigoBlewTronynsSocksOff
File size limit is 128MB.
First | Previous | Next | Last
 
Oh, most any modeling app will work. OBJ is an ancient old format. 
 
Kinn - will do! 
V-hacd 
Found a thread were someone tried using v-hacd in a SketchUp -> Hammer workflow, and it sounds like it didn't work out because of the v-hacd results only being an approximation.
http://sketchucation.com/forums/viewtopic.php?f=323&t=60709

I found another reference to this library, which seems to do exact decompositions, might be promising:
http://doc.cgal.org/latest/Convex_decomposition_3/index.html

re: texture coords, I did get conversion of UV's from .obj to .bsp tex coords working when I was playing with qbsp. I could probably contribute code to convert from .obj UVs to .map coordinates if you want (would probably need to offer the user a choice of Valve 220 format, or standard) 
 
Wow, really? Texture coord conversion ... that would be interesting! Being able to UV in MODO rather than Quake would be cool. 
 
I'd say UV in 3d modelling would be only go so far because of the way textures tend to skew in the standard .map format. I think it may be better in 220 as it is more accurate. 
 
Imagine stone statues and things like that. Would be KILLER. 
 
Yeah, from what I could tell, the conversion is lossless for triangles with UV at each vertex, to the format .bsp uses. (If you have a 4 or more -sided face, I think you can make it stretch weirdly at one corner in a way that can't be stored in the .bsp format, but that shouldn't usually be an issue.) Valve 220 is almost the same as the internal .bsp format, so that should also be lossless. You'll get possible distortion with standard map format, but nothing can be done about that.

This was a .md3 I grabbed from nexuiz, converted to .obj, and ran through my patched qbsp: https://www.dropbox.com/s/o4fbwqc9lq5uk93/obj.jpg?dl=0

Messy code is here: https://github.com/ericwa/tyrutils/blob/obj/qbsp/brush.c#L784
basically you just have to solve a "known 3-vec = (known 3x3 matrix) * (unknown 3-vecor)" equation.

I could give it a shot if you don't mind sharing the source (either just privately or here?) 
 
Wait ... those spider things are BSP? I'll be in my bunk. 
 
Let me get the first real version out and then we'll talk. I'd LOVE to get some sort of convex decomp thing going along with texture coords ... oh yes. 
Errmmm 
that's amazing! Right?!

the possibilities for this seem ludicrous 
 
Warren, cool, sounds good!
Yeah, the spiders are in the BSP - each one's a func_illusionary, but instead of being made of brushes, qbsp is getting the faces directly from the .obj file. It can also insert them directly into the world instead of into bmodels, but you still don't get collision, and the model may or may not block gunshots depending on the obj file.
I think going via .map is the way to go in the long run, though my approach has the benefit of handling textures and concave models right now... I can post it if anyone wants to mess around. 
Ericw 
omg - keep working on it, you'll get there! 
RE: Gear 
I probably didn't do that right. I think the real way would be to convexorize the stuff myself and then rotatecopy it. That way I could use your default setting to generate several brushes instead of 1 brush per face.

auto convexorizoring would be awesome though, but likely really mathy and complex. 
 
Yeah, I had that thought while looking at the gear ... doing it yourself would definitely be WAY more efficient in terms of brush use.

But, the other methods will work now so it was worthwhile as a test case. 
OBJ2MAP 1.0 
I'm going to do a proper news item tonight or over the weekend but for now, I think we're at 1.0 and ready to release.

+ logging should be more reliable now
+ added more info to log
+ user can specify a texture name for visible and hidden faces

https://dl.dropboxusercontent.com/u/161473/OBJ2MAP/OBJ2MAP_Release.zip

https://dl.dropboxusercontent.com/u/161473/OBJ2MAP/ReleaseShot.png

Run it as administrator! Apparently that's necessary for some people. 
Still Runs With Wine 
I just tested the latest version with Wine, and it worked fine. Since it's going to be a 1.0 release, I thought it was a good time to confirm this.

There are a couple of minor typographical errors in the log file. It says:

"Copy To Clpboard : True" - should be Clipboard
"Beginnning MAP construction" - should be Beginning

Glad you shared the program with us all. 
 
My attention to detail is stunning.

Thanks, will fix! 
Convexorizing Yourself 
is definitely the way to go: http://i60.tinypic.com/zm1c7p.jpg 
 
Nice. 
Bumped 
I can't wait trying it out, but I'm stuck on the message:
it's not a valid win32 application. 
 
That's odd. It's definitely a 32bit application. What OS are you using? 
 
'not a valid win32 application' is typically a dependancy issue. make sure you have .net 4.5 installed (I think that's its only dependancy). 
 
Yeah, should just be .NET. Although that's a terrible error message if that's what it is...

Also, run it as an administrator. Don't know if that's relevant. 
 
i think the run as admin is only if you're writing to a folder that is normally protected, eg: program files, or documents?

and yeah, that is a shit error message for a missing dependency. :\ 
Passed 
I'm using WinXp servicepack 3, but it's fresh installed.
Also I hate adding an administrator and password as I'm the only user,
so I can't even reach that prick.

I think it is the .net.45 not installed.,
but in that case I'm not lucky
because it won't fit WinXp.

I'm an ancient fossil, I know. 
First | Previous | Next | Last
You must be logged in to post in this thread.
Website copyright © 2002-2024 John Fitzgibbons. All posts are copyright their respective authors.