usage: q3map -light [-<switch> [-<switch> ...]] <mapname>
(scale) This scales the intensity of area lights.
Creates a debugging border around the lightmap.
(N) (where (N) = number of
radiosity passes)
This enables radiosity calculation. It will re-diffuse the light emitted onto a surface (N) times. In addition, it will write out the bsp after every pass, so it can be cancelled without loss. The tint of light that’s reflected is the lightmap/vertex * texture color, and subsampled to certain granularity across every lit surface. Use q3map_lightimage in a shader to override the reflected color.
This enables radiosity affecting the lightgrid.
This stops Q3Map from calculating light on a sample after it exceeds (255, 255, 255). This may produce odd artifacts on maps with lots of saturated colored lighting. In addition, do not use cheap with radiosity, if you wish to preserve all the emitted light.
This is the same as cheap, but it is used only in lightgrid calculations.
(default = on. To disable, see nocollapse) -nocollapse disables comparison of raw lightmap a to raw lightmap b. If they are exceedingly similar, (off by 0.0025 in 3 * x * y), then they are merged.
Enables lightmap debugging.
This colors unused lightmap pixels hot pink.
This dumps prefab? files, when used with radiosity, for each bounce. (not used in light?)
This exports lightmaps to .tga images.
see extrawide
Extra is the equivalent of -super 2. If you have a -super (N) argument after -extra, it effectively masks it. Use one or the other. Also, -extrawide is the same as -super 2 + -filter.
This enables light envelopes for area lights, and enables light culling. With some maps, using it could speed up light calculations by 50x or more. For maps with large numbers of dim surface lights, fast also has the side effect of producing dimmer maps. Fast is just fine for final compiles, though, as long as you light accordingly. To disable light culling, and at the same time disable compile time speed increases, do not put fast in your command line.
This enables fast for radiosity passes only.
This enables fast for lightgrid calculation only.
Gives you a gaussian blur for shadows, and applies a 3x3 box filter (similar to Photoshop's blur) to the lightmap. It breaks down on the edges of a lightmap, because there isn't anything to sample from off the edge. Filter averages a given luxel with its 3 neighbors. This is what the old extrawide did, but it's a bit smarter about it.
This disables identical lightmap collapsing. Collapse compares raw lightmap a to raw lightmap b. If they are exceedingly similar, off by (0.0025 in 3 * x * y), then they are merged. (unless you use nocollapse)
This disables calculation of the light grid for dynamic model lighting.
This colors lightmaps by their facings, and is good for debugging phong shading.
This disables the surface tracing of detail brushes and patches for shadow calculation.
No light tracing is performed. As a result, no shadows will be cast.
This disables calculation of vertex lighting.
This enables the casting of shadows by patches.
(scale) This scales the intensity of point lights.
(N) This sets the lightmap pixel size to (NxN) units. The default is 16. (16x16)
This turns on phong shading, but may only be useful under certain circumstances, and on a per-shader basis.
(Nº) This also turns on phong shading, and gives you the ability to specify the angle of light, determined by (Nº). Shadeangle, as a global switch, is a bit dodgy though.
temporarily non-functional
This makes for smoother shadows. A smarter version of extra, smooth only subsamples lightmap pixels that are shadowed. It produces results comparable to extra, but in approx.1/3 the time. It can also be combined with super and filter, for ?16 or ?48-tap sampling.
(N) Arbitrarily ordered grid supersampling of lightmaps. This replaces extra & extrawide. Extrawide has always had a bug in it. This replacement does true N-tap supersampling, rather than the weird filter that extrawide did. Note that -super requires an argument >= 2
By default, lightmap samples are 16x16. ( ? )
(N.N) Thresh sets the recursive triangle subdivision threshold. The default is 1.0.
(N) The number of threads used to compile the map. For the fastest compile times, the number of threads should be set to the number of system processors. 1 is the default.
This outputs verbose information.