|Posted by ericw on 2017/12/30 22:59:42|
|This is the continuation of the previous
tyrutils-ericw thread now that the project is renamed.
Github issues - feel free to report bugs / upload screenshots / test maps
Github releases (downloads) page
v0.16 is not the most amazing release because I am still working on fixing lightmap seam issues. However it has 2 somewhat important fixes:
- Fix MarkV .lit rendering (do the color->greyscale conversion in a way that forces MarkV, FTEQW to render identically to Fitzquake, Quakespasm)
- better detection of when BSP29 format bsp exceeds the format limits. Previously it would silently produce corrupt bsp's if you exceeded 64k leafs/nodes without manually enabling -bsp2 mode.
Yeah, it should be easy to add a default; I was thinking func_detail.
How complicated would it be to add random texture tiling feature based on texture names during bsp generation similar to Half-Life's technique which does this during run-time?
This would help break monotony of large walls and rocky terrains.
Misc_external_map 'classname' Shadows
I want to make sure I have this right. While setting up the entry in the FGD it seemed that for...
func_detail, func_group and func_detail_wall there is no way to disable shadowing?
Setting _shadow to 0 or -1 on the misc_external_map entity had no effect.
Jus' curious, thanks.
@veryslowmapper Yeah, it would be doable - probably not too hard because qbsp already chops up faces to fit within the software limits.
@damage_inc "_shadow" "-1" should work on misc_external_map. I'll double check.
Netradiant Valve 220
"Infinite projection" in 220 format causing light stage to fail. Created by rotating brush without tex lock.
Example map with stretched trim
I'm not sure what to do.. on one hand, that face with an invalid texture projection crashes software quake (at least winquake.exe), so light.exe failing with a hard error isn't too unreasonable.
maybe qbsp should warn about the face, and then save it to the .bsp with default texturing.
1) the qbsp defines the sizes of surfaces.
2) infinite projection means infinitely sized lightmap.
if the qbsp is generating a lightmapped surface larger than the specified limit then that's the qbsp going wrong, not the light util.
(unless its turb or sky, anyway, in which case the light util shouldn't be trying to light them and thus shouldn't care).
Just reading about the new misc_external_map feature, and it sounds awesomebiscuits.
Although, I'd have thought it could be even more powerful if it worked like: whatever is in the external map, including multiple entities (e.g. func_door, func_detail, light_*, whatever) would just get copied across, so e.g. you could have a brush torch sconce + a light or two associated with it.
I'm sure there's a damn good reason for the current implementation but just thought i'd mention it :}
Are you talking about Quinstance style external bsp support?
I'm talking about the misc_external_map entity in ericw's bsp tool
I started with "brushes only" as a first step because it simplified things a bit, especially "-onlyents" compiles can ignore the external maps right now, but they'll need to load them when I import entities as well. No special reason it can't be done though.
Well, I think one of the biggest uses for me would be the aforementioned torch sconces where I'd maintain just a small selection of lighting prefabs; each containing some func_detail_* sconce thingy, plus a torch entity and maybe complimented with a couple of secondary filler lights.
I'm fairly schizophrenic when it comes to lighting, and often decide to change every single light in my map in a certain way, and then some time later, I end up changing them all again, etc. Being able to do this through prefabs would be a "game changer", to coin a phrase :}
What Kinn said is why I wanted instanced groups in TB so much. You could just stick your light setup (brushes + ents) in an instance group and copy it around the map, thus easy to iterate over time.
I've avoided directly addressing this, for fear of seeming to shit on eric's efforts in the same vein, but I feel I should clarify that Quinstance
is not for external BSP files, but for MAPs, and already does what Kinn's looking for.
It is an external tool, which of course complicates compile setup, and it (along with VMFII
, which does all the real work) is written in C#, which demands the .NET Framework in Windows or Mono in Linux and macOS, so I understand the value of doing a similar thing in a native tool, especially one already so widely used. Until entity support arrives in ericw's QBSP, though, you could use Quinstance to have easily tweakable prefab entity configurations. Nesting, entity name fixup, @ syntax, material replacements, most of the Source func_instance features work with Quinstance. The "origin" and "params" entities found in that engine don't, but VMFII doesn't support them either.
If you need example usage, I could pack some of my own maps up for you. The Jam 7 and Retro Jam 5 distributions include source for my entries that separates the main map from the instances, but only in JMF form, which I'm only now realizing I shouldn't have done. For most other Jams I only sent in collapsed versions, i.e. after the instances were resolved, but I've used instances on all but one of my maps going back to Jam 6 (including the "bubbler" ornaments in my Xmas Jam map), so I have plenty of samples if anyone's curious.
Eric, if you're interested, both my wrapper code and Metapyziks' is MIT licensed (really "X11", pardon me), so despite it being a different language you're free to browse it for ideas of how to approach the problem without fear of running afoul of some oddball license terms. Or how not to, if the techniques prove unappealing. I know the VMFII dev has expressed a desire, at one point or another, to redo the project to greater or lesser extent, so how much you'd want to imitate I don't know.
the super useful _dirt_off_radius / _dirt_on_radius stuff seems to be missing from the website documentation.
Anyone else seeing buggy behaviour with the "mangle" setting on "_sun" "1" lights? Try setting a few different yaw values on the mangle...
What does external bsp support offer that would be different from setting the model key on a func_illusionary to a .bsp? (Ignoring precache requiring engines of course)
its external_map not external_bsp :P
So it's a simplified quinstance that's editor independent. Cool.
Would be awesome. I’m think it could be used for lots of interesting stuff
Any love for quake 2? :)
Nothing usable right now, but I made some progress a few months ago (light.exe can read q2 bsp's and write them out with working lightmaps). What's missing is mapping the Q2 contents / surface flags to the what the tool can understand.
Small update, mostly with changes to how invalid valve220 texturing is handled:
I'll do another release soon with _phong_angle_concave which lets you disable / change the phong shading threshold for crevices / concave joints.
Yeah I've tested it and it's really useful for rocky structures where you want to smooth out the convex areas, but keep the concave parts hard (like where boulders intersect basically).
Also useful on some weird low-poly arch setups with angled trims where you'll only want concave smoothed but not convex.
Another small update:
- light: tweak phong shading to use area and angle weighting
- light: add "_phong_angle_concave" key
- light: fix -bspx option
full explanation for _phong_angle_concave is in the light manual: http://ericwa.github.io/ericw-tools/doc/light.html
I was going to add a toggle for the new area/angle phong shading weighting, but it's too subtle to notice in most cases, but I can make it optional if needed.
Thanks again for _phong_angle_concave, it's quite handy!
I have some ideas of stuff to do with the new area weighting that will probably show off better how it helps, need to find some time to do it.
Silly question, does light still act really strange around certain geometry (example on github here
) or is 0.15.9 still the most reliable version?
I haven't had those kinds of artifacts in a while with the new versions, at least not on sealed maps (sometimes on leaky maps the lighting is a bit weird).
CLEric, Hope You Don't Mind @Mankrip
Been just dabbling a bit with my own flava of a compiling utility when tonight I had this idea:
After you set all your options just click "mini mode", stays on top also, and then no alt-tabbing to the compiling UI!
Anyhoo's... beyond that I almost have the "simple compile" interface done. then it will be on to the advanced UI design part. It will also have the map conversion and bsp operations(extract textures etc) incorporated as well.
"Debug" Preview Of The "simple" Compile UI
I say Debug because I have screen elements that will not be in final design, they are there just so I can see what's it going on.
Here it is:
I know "central" part looks identical to WC/Hammer/Jack etc, I just like that layout. Still not set in stone, if I "see" and easier to view arrangement I will ofc change it.
This is windows-only gui tool or crossplatform?
@Doomer, Most Likely.
I loaded up my 'Nix distro, which is severely aged at this point, and it partially worked(Under Wine). Which is ofc not good enough.
Wine Its Not Very Good.
What was used to create this tool - MFC, .NET (WFp or WinForms), Delphi?
An off-the-wall Basic IDE! Yes, laugh... hehe
I'm no programmer, and in reality this is just a glorified batch file/commandline tool. Mostly string declarations, manipulations and paths saving.
If there is an easy to use, freely downloadable cross platform IDE let me know. I would consider rewriting it.
All I need is access to File and Folder dialogs plus write to .INI files.
Working With Func_detail Stuff
Is a bit of a pain. Would be real handy to just rely on standard func_groups/etc for organizing in editor and then just slap a detail flag on whatever as needed. Maybe handling it like q3map2 does is possible. Would feel more consistent and be less of a headache to work with for sure.
Like...a Spanwflag? Interesting.
But maybe I missed something. What is the point of func_group if you can just group a bunch of func_detail or standard brushes in the editor?
Just Like In Q3 + Radiant,
a simple "detail or not" flag for the compiler with filtering in editor. I don't know if anything more is needed for quake? To me func_detail_ seems unnecessary and it'd be much better just marking brushes as detail. You could have detail stuff and worldspawn linked together in func_groups then atleast.
I did notice too that func_detail_wall and func_wall don't behave the same in game; custom entity keys were removed from func_detail_wall I think.
To me func_detail_ seems unnecessary and it'd be much better just marking brushes as detail.
That requires a new map format.
That's why func_detail exists. To allow you to flag detail using the existing quake map format.
I can add support for the quake 3 .map format's detail flag (stored per-plane in the .map file). @Qmaster in GTKRadiant there is a shortcut key to turn the selection into detail, so it could be handy if that's your editor.
I did notice too that func_detail_wall and func_wall don't behave the same in game; custom entity keys were removed from func_detail_wall I think
They are completely different beasts.
func_wall is a separate entity in game, potentially interactive.
func_detail_wall becomes "baked" into the world geometry on compiling, like func_detail. Unlike func_detail, it doesn't chop up world faces that it butts into.
An alternate approach is to adopt just the UI aspect from q3radiant (the keyboard input) -- under the hood the editor could be using that flag/unflag operation to move the brush from world to func and back. Don't show it to the user as an entity, show it as world brushes with detail flag, but store it as a func.
you'd also still need to support _phong and all the other keys that you'd want on detail too...
Would be awesome to have q3 style detailing as option eric. Wanted to ask this ages ago.. Before pestering niger to add support -)
Will be easy enough to just use func_groups to add keys I assume.
Website copyright © 2002-2018 John Fitzgibbons. All posts are copyright their respective authors.