153 lines
5.0 KiB
Plaintext
153 lines
5.0 KiB
Plaintext
QRAD DOCUMENTATION
|
|
|
|
|
|
qrad [parameters] bspfile
|
|
|
|
PARAMETERS YOU MIGHT WANT TO USE
|
|
|
|
-bounce <0-???> default: 8
|
|
Specifies the number of light bounces to perform.
|
|
0 gives direct lighting only, and does not calculate any
|
|
transfer functions, giving a very fast run. Under a minute
|
|
for a level with vis info, a few minutes without.
|
|
Because most of the time in full runs is in the transfer
|
|
function generation, the number of bounces greater than 0
|
|
is not a crucial speed factor.
|
|
|
|
-chop <16 - 256> default: 65
|
|
Determines how finely patches will be chopped up for the
|
|
radiosity process. Smaller number give better results,
|
|
but use quadratically more ram and time. You can use
|
|
larger values like 96 or 128 to get the light properly
|
|
dispersed in the level (no need for a -ambient) but still
|
|
take lots less time than a full process. The older qrad
|
|
used a chop of 49, but it didn't do the direct lighting
|
|
as a seperate step.
|
|
|
|
-ambient <0.0 - 1.0> default: 0.0
|
|
This value is added to every light sample before any lighting
|
|
is done. An ambient of 0.1 (10%) or so can be used in
|
|
conjunction with a -bounce 0 to get a very fast aproximation
|
|
to the full process.
|
|
|
|
-maxlight <1.0 - 2.0> default: 1.5
|
|
If set to 1.0, no surface will ever be overbrighted,
|
|
no matter how much light shines on it. Artist types may
|
|
push for this to keep their textures from being abused by
|
|
the overbright tables.
|
|
|
|
|
|
PARAMETERS YOU AREN'T LIKELY TO USE
|
|
|
|
-verbose
|
|
Turns on more detailed statistic printing.
|
|
|
|
-dumppatches
|
|
Development aid -- writes bounce0.txt and bounce7.txt polygon
|
|
files for visualizing the radiosity process with glview.exe.
|
|
|
|
-threads <1-???> default: number of cpus
|
|
Specifies the number of threads to use for calculations.
|
|
Set to 1 to isolate potential multithreaded errors.
|
|
|
|
-scale <0.0 - ???> default: 1.0
|
|
Multiplies all light values by this factor to brighten or
|
|
dim the entire level. If you are just way off either
|
|
overbright or almost black, you can easily try scales like
|
|
this.
|
|
|
|
|
|
USAGE IN DEVELOPMENT
|
|
|
|
You should do a vis before performing a qrad. All of the
|
|
lighting tests are accelerated by the PVS information.
|
|
|
|
If you haven't vised the map, subdiv will be forced to 9999
|
|
to keep it from taking all day to complete. Even a fast vis
|
|
will help.
|
|
|
|
The first part of the radiosity process is the accurate
|
|
calculation of direct lighting.
|
|
|
|
All entity lights and the bright (5000 or greater) surface
|
|
patches exactly trace lines to the individual lightmap
|
|
samples.
|
|
|
|
If -bounce 0 is specified, thats it for the lighting.
|
|
Otherwise, the lightmap values are summed up to give the
|
|
initial amount of light to be bounced around the world in the
|
|
radiosity pass.
|
|
|
|
|
|
|
|
SURFACE LIGHTING
|
|
|
|
All textures that are on the "light emiting" list will cast
|
|
light into the world, oriented down the surface normal.
|
|
|
|
The lights.rad file is found inside the game directory
|
|
(f:\quake\id2, etc). It contains a list of texture names
|
|
and the amount of light they emit.
|
|
|
|
A value of 100-300 is reasonable for a large, dim surface, like
|
|
slime, lava, or a dark sky. Lights that will illuminate large
|
|
areas with a small surface area will have values of 5000
|
|
or greater.
|
|
|
|
The total amount of light emited from a surface is proportional
|
|
to the surface's area, so the size of the brush face has as
|
|
much effect on the light sent into the world as the light value.
|
|
|
|
Surface lighting falls off with the sqare of distance, so if
|
|
you double the height of a room, you will need to have four
|
|
times the light intensity on the ceiling to get the floor
|
|
the same brightness. The walls would then be much brighter
|
|
near the top than they were with the smaller room as well.
|
|
|
|
|
|
SKY LIGHTING
|
|
|
|
Skys emit light like other surfaces, but have slightly
|
|
different characteristics. They emit light from all angles,
|
|
so it doesn't matter if you see a horizontal or vertical sky
|
|
brush. Light also does not bounce off of a sky in the
|
|
radiosity pass, if it flies out a sky texture it is just gone.
|
|
|
|
|
|
ENTITY LIGHTING
|
|
|
|
You can use "light" entities to have light come from space
|
|
without a surface. A level designer goto. They should
|
|
function similar to quake 1 lights, with linear falloff
|
|
instead of inverse quadratic.
|
|
|
|
The new class "light_spot" can be used to make a spotlight.
|
|
If it is targeted at another entity, the spotlight will point
|
|
directly at it, otherwise it will point in the direction of
|
|
its "angle" field, so you can easily make simple axial pointing
|
|
spotlights.
|
|
|
|
The field "_cone" determines the spread of the spotlight. If
|
|
not specified, a _cone of 15 will default.
|
|
|
|
|
|
|
|
|
|
ISSUES WITH BMODELS
|
|
|
|
Brush models emit and absorb light like other surfaces, but
|
|
they do not block light, so after normalization, a little less
|
|
light lands on the nearby surfaces, because it was absorbed
|
|
both by the door and the surfaces beyond it. You probably
|
|
won't notice anything unless you have a row of spotlights
|
|
on normal walls, then an identical one on a bmodel.
|
|
|
|
|
|
ISSUES WITH WATER
|
|
|
|
Light does NOT cross through water surfaces if the vis has
|
|
been calculated, because the areas are in different potentially
|
|
visible sets, but if you run qrad without a vis, then light
|
|
will be sent through as well.
|
|
|