omegga-raytracer

A (cursed) raytracer for Omegga.

Features diffuse lighting, reflections, transparency, and image compression. An experiment in graphics programming. Not designed to be practical or usable for actual purposes.

Installation

Start from the working directory of your Omegga installation (the folder that you run omegga in).

cd plugins

git clone git://github.com/voximity/omegga-raytracer.git

omegga-raytracer has no dependencies and no Omegga config properties. No setup is required.

Usage

omegga-raytracer has two commands that are both host-only.

Command	Arguments	Example	Description
!trace	<yaw> <pitch>	!trace 45 -30	Starts a raytrace with the given camera rotation from the position of your torso.
!set	<setting> <value>	!set renderPlayers true	Sets the value of a setting for the raytracer. Setting values do not persist over plugin reloads/server restarts.

Below are a list of the settings available with !set:

Setting	Description	Default value	Example
res, resolution	The resolution of the scene in pixels.	64 64	!set res 300 200 sets the resolution to 300x200 pixels
fov	The field of view of the scene camera.	50	!set res 90 sets the FOV to 90 degrees
diffuse, diffuseCoefficient	A number from 0 to 1 representing the diffuse coefficient.	1	!set diffuse 0.5 sets the diffuse coefficient to 0.5
ambient, ambientCoefficient	A number from 0 to 1 representing the ambient coefficient (multiplier to color when a ray hits a normal opposite the light vector).	0.2	!set ambient 0.5 sets the ambient coefficient to 0.5
light, lightVector	Three numbers representing the light vector pointing onto scene.	-0.3 -1 -1	!set lightVector -1 -1 -1 sets the light vector to (-1, -1, -1), after normalizing the input vector
shadows	Whether or not shadows are enabled. When enabled, total ray casts are effectively doubled.	true or on	!set shadows false or !set shadows off disables shadows
shadowCoefficient	A number from 0 to 1 representing the shadow coefficient, similar to ambientCoefficient.	0.4	!set shadowCoefficient 0.5 sets the shadow coefficient to 0.5
maxReflectionDepth, reflectionDepth	Sets the maximum amount of times one ray can bounce off of reflective surfaces. Used to prevent infinite reflections off of closed reflective areas.	3	!set maxReflectionDepth 8 sets the maximum reflection depth to 8
renderPlayers	Whether or not players are rendered. Players render as cylinders.	false	!set renderPlayers true or !set renderPlayers on enables the rendering of players
renderGroundPlane	Whether or not the ground plane is rendered.	true	!set renderGroundPlane true or !set renderGroundPlane on enables rendering of the ground plane

How it works

When the command to render the current save is issued, omegga-raytracer saves the game's bricks and generates a "scene" in which ray hits and bounces will be calculated.

The "scene" is composed of arbitrary SceneObjects like the AxisAlignedBoxObject (the object all bricks render by) or the PlaneObject (used to render the ground plane). For every pixel in the image, a ray is shot out from the camera's origin. All objects^[1] in the scene are tested for an intersection and compared. When these intersections are tested, the renderer also tests for properties like reflectance, transparency, and shadows, casting rays out in other directions.

When all rays have been cast and bounced an an image is internally developed in memory, the image is transferred to the quadtree color compressor, which initially precomputes the depth of the quadtree and generates the tree at its fullest depth, with each leaf as deep as possible. Then, the optimizer recursively indexes through every branch in the tree, merging if they are composed of identical leaves. Finally, the quadtree brick builder constructs the tree out of bricks, generates a save, and loads it in.

[1]: No acceleration structures are currently used for testing many objects. I would like to eventually implement a BVH to reduce computation for renders that test large amounts of bricks.

Contributing

Feel free! If you wish to turn this experiment into something more practical, open a PR! I will probably not take this experiment to that great an extreme, but you are welcome to contribute.