Light stage compile errors

(wedge2211) #1

I checked around and can’t find anything on these two errors. These have got me stuck on a Jedi Academy map, using the latest q3map2:

— SetupSurfaceLightmaps —
WARNING: Lightmap texture coords out of range: S 123.4268 > 123 || T 0.5000 > 102
WARNING: ProjectSurfaceLightmap: Chose a 0 valued axis

Later on in the compile, I see:

— IlluminateRawLightmap —
0…1…2…3…4…5…6…7…8…9… (40)
0 luxels illuminated

Which results in the whole map being displayed black–all the objects in it are silhouettes, with the exceptions of flares and my skybox (which have q3map_nolightmap). Looks really wierd.

The map has a global fog and a foghull skybox, I’m using an entity sun to light the map. I’ve tried commenting out aything funky in my shader file to no avail.

(ydnar) #2

Did you BSP the map with Q3Map2? Are your shader files listed in shaderlist.txt?

Make sure you don’t have any capped cones or otherwise invalid patch meshes (lines or points) in the map.

Also, try the development build of Q3Map2.

y

(wedge2211) #3

I was given the link http://akiba.shaderlab.com/GtkRadiant/install/q3map2 by rgoer, that’s what I’m using to compile. Before that I was using 2.5.13. I’ve double-checked, my shader files are listed in shaderlist.txt, so I doubt that’s the problem.

Ran brush cleanup and got rid of 4 invalid planes, and redid a couple caps that I remembered doing and could have been botched, but the durn thing still won’t work. I didn’t try capping anything strange beyond bent cylinders and bevels…not sure how I would identify invalid caps like that. I can always work my way through the whole map shudder. I’m just about out of ideas on this…I can post my shader file, but I know I’ve compiled and run the map with these shaders before, and I’ve tried deleting any shader codes that could have been causing problems. Hmm…

(rgoer) #4

If it comes down to it, you can try a binary sort, deleting half the map geometry, then half of that, then half of that etc etc until you are able to localize the problem.

why are you using an entity sun instead of q3map_sunExt + q3map_skyLight?

(RabidCow) #5

What size map are you trying to compile? I have been getting the same lightmap error in tests on huge maps and just assumed it was because of the large size. In my latest tests, I used a huge ase model for a terrain and got everything black except the skybox…
RC

(wedge2211) #6

Yeah, it is a pretty big outdoor map, EasyGen terrain, etc. Sounds like you’re in about the same boat I am. Is this something that could be fixed by messing with the blocksize or lightmap scaling? I’ve tried chopping up big, lightmapped surfaces, but that didn’t work.

rgoer- Binary sort? Eeeeuuuch. Man, that would be time-consuming, and I’d be looking for a mathematical point in-editor–kinda hard to see anyways. There’s not any sort of flag I might look for int he BSP process that would spot an invalid patch, is there? (I will keep this in mind as a last resort. I’m desperately trying to remember what I did between the last compile that worked and now…) I’m using an ent sun just because I’ve had pretty good results with them, why, are they undesirable?

(rgoer) #7

undesirable isn’t quite the word I’d use, but yes

(obsidian) #8

IIRC, this problem was addressed in the latest dev-build of Q3Map2.

http://akiba.shaderlab.com/GtkRadiant/tools/quake3/q3map2/Release/q3map2.exe

(rgoer) #9
(wedge2211) #10

Okay…I’ve gone through a long process of deleting map object types and compiling while on MSN with rgoer. So I can now produce a whole list of things that are not causing the problem. Here’s a summary:

light-stage compiles now generate errors of the forms
WARNING: Lightmap texture coords out of range: S 106.4268 > 106 || T 14.9146 > 124
WARNING: ProjectSurfaceLightmap: Chose a 0 valued axis

And IlluminateRawLightmap says “0 luxels illuminated.”

  • When all patches were deleted from the map (select none, filter patches, invert selection, filter patches, invert selection, delete), the same errors were seen on compile.

  • I was using spawnflags 4 on a bunch of models, tried deleting that key from all of them, with no change.

  • Tried the same thing with the models as I did with patches. Blasted 'em, recompiled, no change.

  • I’ve got a fog brush meeting a water brush. rgoer suggested this sometimes confuses Q3 (previous compiles have worked) so I deleted one (the water) and recompiled…no change.

  • Tried deleting all entities in general…the ProjectSurfaceLightmap: 0-valued axis error disappeared (“0 luxels illuminated remained”)…but then again, this is probably due to the fact that I blew away all my lights.

  • I’m using an entity sun in my map, and on the suggestion that this might be bad, I deleted it, with the result that the 0-values axis error disappeared, but still 0 luxels were illuminated!

  • Just in case, as a last resort, I retextured everything in common/mp_blue and tried a compile…still errors.

So…I’m really at a loss as to what’s going on here. If it helps to identify the problem, here’s some added information on my map:

Worldspawn keys:

classname worldspawn
soundset rainy
music music/t1_rail/rail_nowhere.mp3
message Corellian Platform Encounter
ambient 10
_color .75 .75 .8
_farplanedist 2048
_blocksize 1024
_foghull corplat/corplat_sky

The map has 96 func_groups, 1 fx_rain, 106 misc_models, 2 target_speakers, 2 func_timers, 6 func_doors, 1 trigger_push, 2 target_positions, 20 func_breakables, 8 func_usables, 41 lights, 1 target_delay, 2 fx_runners, 2 info_notnulls, 1 NPC_Droid_R2D2, 4 waypoints, 2 info_player_duels, 1 info_player_duel1, 1 info_player_duel2, 1 info_player_intermission, 2 func_statics, 2 ref_tags, and 1 info_null.

corplat.shader (Yeah, a tiny bit messy…I’m not really a coder)
corplat_terrain.shader (EasyGen terrain shader)
Compile log, -v

gaspgaspgasp

Thanks guys…

(ydnar) #11

You must supply a valid -fs_basepath to Q3Map2 so it knows where to find the game assets, for ALL phases:

-fs_basepath “C:/Games/Jedi Academy/GameData”

y

(wedge2211) #12

Okay…just after going through the process of yelling “d’OH” and smacking my forehead, I threw the -fs_basepath, exactly as you provided, into my compile lines. I knew it was going to be something simple and dumb that I forgot. Unfortunately, my map managed to refuse this fix. Still getting the errors.

If this keeps going the way it is, I’m just going to forget it.

(rgoer) #13

Erm, you may well need backslashes in the fs_basepath arg on windows, rather than foreward slashses…

-fs_basepath "C:\path	o\GameData"
(ydnar) #14

Can you post the compile log now, with the full options you use per-stage?

(wedge2211) #15

Sure thing. This was a test compile, using the options:

-meta -v -fs_basepath “C:\Games\Jedi Academy\GameData” -game ja
-vis -v -fs_basepath “C:\Games\Jedi Academy\GameData” -game ja
-light -fast -patchshadows -super 2 -v -fs_basepath “C:\Games\Jedi Academy\GameData” -game ja

The most recent log can be found here.

Thanks for taking the time to look at this.

(ydnar) #16

Can you try with -samples 2 instead of -super 2?

(wedge2211) #17

No luck. :frowning:

(wedge2211) #18

Did some testing with rgoer, and determined that the map actually WILL light even with these errors. So, assuming there’s no way to elimenate them, I’m just going to go with it.