GitHub - TechnoSam/vtray

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 53 Commits
scripts		scripts
tests		tests
.gitattributes		.gitattributes
.gitignore		.gitignore
CMakeLists.txt		CMakeLists.txt
README.txt		README.txt
Vagrantfile		Vagrantfile
camera.cpp		camera.cpp
camera.hpp		camera.hpp
catch.hpp		catch.hpp
color.cpp		color.cpp
color.hpp		color.hpp
light.cpp		light.cpp
light.hpp		light.hpp
message_queue.h		message_queue.h
object.cpp		object.cpp
object.hpp		object.hpp
plane.cpp		plane.cpp
plane.hpp		plane.hpp
ray.cpp		ray.cpp
ray.hpp		ray.hpp
ray_tracer.cpp		ray_tracer.cpp
ray_tracer.hpp		ray_tracer.hpp
scene.cpp		scene.cpp
scene.hpp		scene.hpp
sphere.cpp		sphere.cpp
sphere.hpp		sphere.hpp
test_camera.cpp		test_camera.cpp
test_color.cpp		test_color.cpp
test_light.cpp		test_light.cpp
test_plane.cpp		test_plane.cpp
test_ray.cpp		test_ray.cpp
test_ray_tracer.cpp		test_ray_tracer.cpp
test_scene.cpp		test_scene.cpp
test_sphere.cpp		test_sphere.cpp
test_vec3.cpp		test_vec3.cpp
unittests.cpp		unittests.cpp
vec3.cpp		vec3.cpp
vec3.hpp		vec3.hpp
vtray.cpp		vtray.cpp

Repository files navigation

FIXES SINCE THE BETA:

Memory issues. Was accessing a 3x3 array at element 3. Windows set this
value to 0, Linux left it uninitialized. This caused my random issues
from before and my failing the memory tests.

Lighting issues. Planes allowed light in from the opposite side relative
to the camera. This did not affect spheres, but it made planes brighter
than they should have been.

Modules:

-Scene-
Contains a camera, vector of lights, vector of planes, and a vector of
spheres. I wanted to do a single unified vector of objects instead of
breaking it up into planes and spheres, but I kept running into issues
that should have been simple to fix but I just couldn't for some reason.

The scene contains all the information about what to draw, so it is in
charge of returning pixel color values using ray tracing.

-Camera-
Contains all the attributes as given in the JSON.
Supplies a method to generate rays to a given pixel from the focus.

-Light-
This probably could have been a struct, it's just a container for the JSON
data.

-Object-
Contains a color and lambert, which all objects must have.
Also has a method that returns the distance to intersection from a given
ray. This distance is infinity if there is no intersection. Arithmitic
will give us the actual intersection location if we want it.

This allows all objects to be known to have certain behavior. The benefits
did not pan out due to complications.

-Sphere: Object-
Inherits from Object and adds a center and radius.

-Plane: Object-
Inherits from Object and adds a center and normal.

-Ray-
Contains two geometric vectors, origin and destination. From these, the
direction can be easily computed.

-Vec3-
A 3D geometric vector. Defines dot/cross product as well as overloaded
operators to make math easy.

-Color-
Contains RGB values with overloaded operators to make math easy. Also supports
scaling.

Putting them together:

Ultimately, the Scene is responsible for everything. Colors, Vec3s and Rays
form the base operations, following the Ray Tracing algorithm. This algorithm
got quite tricky once I had to account for clipping and roundoff. I'm not too
happy with the current state of the algorithm.

Sources:

https://www.scratchapixel.com/lessons/3d-basic-rendering/introduction-to-ray-tracing/how-does-it-work

This site provided most of the high level algorithms I used like object
intersection.

I used Dr. Wyatt's help for camera rotation.