Q3Map 2.5.8 "Melting Crayons"

[ydnar]

Q3Map 2.5.8 - Sundrop

Win32 (x86) (http://shaderlab.com/q3map2/2.5/q3map_2.5.8_win32_x86.zip) - Linux (x86) (http://shaderlab.com/q3map2/2.5/q3map_2.5.8_linux_x86.tar.gz) - Darwin/Mac OS X (PPC) (2.3.36) (http://shaderlab.com/q3map2/2.3/q3map_2.3.36_darwin_ppc.tar.gz) - Readme (http://shaderlab.com/q3map2/2.5/readme.txt)

Lots of fixes and bits in this release. The illest (major) new feaures are light jittering for penumbra (half-shadow) effects and per-shader and per-light lightmap filtering.

Screenshot of penumbra effect, using the following shader directive:

q3map_sunExt 0.95 0.99 1.0 120 60 30 2 16

The additional two numbers at the end are deviation in degrees (2) and number of random jitters distributed over the solid arc (16). This gives you much more realistic shadows from the sun, especially when trying to simulate a cloudy day or a wide sun. Generally values up to 2 or 3 degrees are acceptable. The real sun has a solid angle of about half a degree.

http://shaderlab.com/q3map2/screenshots/penumbra_06.jpg (http://shaderlab.com/q3map2/screenshots/penumbra_06.jpg)

The penumbra effect can be applied to entity lights (point, spot, and sun) with these new entity keys:

"_deviance" "N" -- where N is distance in world units for point/spot lights and degrees for suns
"_samples" "N" -- where N is the number of samples

The sun in the screenshot could be done using an entity sun with the following key/value pairs targetting an info_null with targetname "sun_target":

"classname" "light"
"_sun" "1"
"_color" "0.95 0.99 1.0"
"light" "120"
"_deviance" "2"
"_samples" 16"



The other big new feature is shader-controllable lightmap filtering. The old -filter command still works, but it can be overridden (and is in fact reimplemented) using new code. Add the following to your light-emitting shaders to use the effect:

q3map_lightmapFilterRadius 0 16 -- no filtering for self, but light it emits will be filtered across 16 world units (default lightmap sample size)

q3map_lightmapFilterRadius 64 0 -- filters itself quite a lot (64 units) but doesn't filter any light it casts

If you use q3map_skyLight in your sky shaders, I recommend using q3map_lightmapFilterRadius 0 64 to blur out the "stadium lighting" artifacts.

Entity lights can also specify lightmap filtering using the "_filterradius" key:

"_filterradius" "32" -- will filter lightmaps created by this light by 32 world units



Version history/changes:


2.5.8 (2003-10-02)

- New: Added two new sun parameters: angular deviation (width of the sun in
degrees) and sampling count (jitters). This allows for decent approximation
of penumbra "half-shadow" effects from sunlight with 16+ samples. These
parameters are accessible for entity lights (including spots and suns) via
these entity keys: _deviance and _samples. To use in shaders, use the new
q3map_sunExt <r> <g> [b] <brightness> <angle> <elevation> <deviance> <samples>
- New: q3map_lightmapFilterRadius <self> <other> for light-emitting shaders.
Put *after* any q3map_sun directives or else your sun will be filtered. This
is good for eliminating the "stadium lighting" effect from q3map_skyLight.
Also usable as an entity key: _filterradius or _filter
- New: Quake 2 MD2 model support in PicoModel for misc_model entities
(thanks to Nurail!)
- Re-enabled vertex-cache-aware triangle reordering. Will probably have a
negligible effect on rendering performance, but can't hurt
- Added short-circuit to raytracer: any empty nodes (including children) are
ignored on sun traces
- Added BSP file size printout
- Filtering of any kind now disables adaptive supersampling on a per-light,
per-ightmap basis
- Fixed another _minlight <-> styled light interaction bug (thanks pjw!)


y

[lennyballa]

Woot just in time, my vis is almost ready :D thx ydnar, you're the coolest

[lennyballa]

Btw, could you in the next release at a percent thing, so you know how far it is? Or isn't that possible? And how do i use these new features btw, do i need to add something to my compiling line?

[chavo_one]

This is so incredibly wonderful for the map I'm working on. "Stadium lighting" artifacts, be gone! Smooth outdoor shadows, hello!

Thanks! :drink:

[redfella]

n1

Thanks!!

:banana:

[WeblionX]

I'm going to try the sun extension out in a test map. What, if that's a standard, sky is that, by the way?

[ydnar]

I'm going to try the sun extension out in a test map. What, if that's a standard, sky is that, by the way?

It's from my site here (http://shaderlab.com/slrl/?p=textures/skies/skyboxes).

y

[Emon]

Pfft, it sucks. Who wants incredibly realistic lighting that can out class most of the newer games on the market, anyway? :rolleyes:

[Loffy]

Yeah, and speedy comply software for free.
:drink:
Thanks ydnar, so unreal good. Tjohoo!
// Loffy

[megaman]

Great work ydnar \o/

[Godmil]

looks funky :)

I was wondering ydnar, did you get a chance to bump up the Max_edge_lines limit?

p.s. I'm taking your advice and converting a lot of brushes to models (or at least begging modellers to do it for me ;) )

thanks.

[ydnar]

looks funky :)

I was wondering ydnar, did you get a chance to bump up the Max_edge_lines limit?

p.s. I'm taking your advice and converting a lot of brushes to models (or at least begging modellers to do it for me ;) )

thanks.

Not yet. This release was baked and ready to go before I got wind of your problem.

Nice map, but I still advise you change as much of the detailed hallway decorations etc into models for best performance.

y

[Godmil]

OK, thats great thanks :)

[heeen]

hmmm
258 seems to be quite unstable...
i and my mate have crashs on the map we're working on all the time..
257 works fine, tough..
the crash appears in the endmodel phase in bsp

[ydnar]

What kind of map? Does it have lots of misc_models?

y

[MindLink]

Yep. About 444 of em in the latest version.
And what kind of map : http://www.splashdamage.com/forums/viewtopic.php?t=4656

[Loffy]

...
the crash appears in the endmodel phase in bsp

I get the same problem.
When I do the ususal compile, with my bat-file, the program crashes. Right at the -Endmodel-. The Windows dialogue frame comes up, ("Do you want to report this?" and I press "no" - THEN the compile continues!
:clap:
So, in other words, the bsp is fine, but at the end it crashes. I press "no", and the compile finishes (the -vis and light phases).
The map looks alright when I make a pk3 out of it and play.
Strange.
I do not have a lot of models in the map. A command map, a couple of MGs, a truck, "bank_roof.ase" and health/ammo cabinets.
// Loffy

[rgoer]

I get the same crash, right at the end of the BSP phase. I've found that you can avoid this crash if you BSP using the -verbose switch.

[lennyballa]

doesn't work for me, i keep getting that crash :@ isn't there a solution?!

[pazur]

2.5.8 it crashes also on my map :bored: somewhere at bsp -> fix t-junctions

edit:
to be more precise. it crashes here:

0...1...2...3...4...WARNING: Flipped triangle: ( 2385 2652 -1580) ( 2385
2652 -1581) ( 2385 2652 -1581)
5...6...7...8...9... (23)
29580 surfaces merged
358855 vertexes merged
--- FilterDrawsurfsIntoTree ---

[damocles]

I too get the same crash on the filter section as Pazur does :( All previous versions since ET will compile just fine.

[ydnar]

A new GtkRadiant 1.3.12 release candidate has been released. Check the MapCenter forums for more detail.

It includes a build of Q3Ma2 2.5.9 with a quickfix to disable the surface triangle optimization code that's causing the crash.

I had to disable it because I didn't have a map that would let me reproduce the crash.

y

[MMFSdjw]

I want to make sure I understand this right.

where my sky shader says
q3map_sun 1 1 1 100 130 50
I change it to
q3map_sunExt 1 1 1 100 130 50 2 16
and I'm good to go?

[ydnar]

Q3Map 2.5.10-test-1 (http://shaderlab.com/q3map2/2.5/q3map_2.5.10-test-1.zip).

Fixes a couple bugs, including BSP crashes (hopefully) and styled radiosity taking forever at SetupEnvelopes.

Test away... :)

y

[pazur]

excellent work :) 2.5.10 doesnt crash anymore at bsp for me and i like the memory info at end of each stage. ydnar did u do anything to alphashadows? they seem to be more precise... but this could be my compile settings :blah:

[Axedup]

yep 2.5.10 test worked fine for me, but I never tried 2.5.8 or 9 so I can't help there, but .10 is all ok ;)

Axedup

[ydnar]

Q3Map 2.5.10-test-2 (http://shaderlab.com/q3map2/2.5/q3map_2.5.10-test-2.zip).

Fixes sun bug discovered by MindLink.

y

[ydnar]

There's a bug in 2.5.8 that makes lighting take 10x as long to light certain maps.

I'm readying a 2.5.10 release that corrects this oversight. :cough:

Sorry about that... :)

y