Implementation of a fast ray marcher based on sphere tracing.
This project was part of the ETH Course - Advanced Systems Lab. In a team of 4 members, we had to optimize a specific algorithm following some guidelines learned in class.
Here is the final report.
On the branches of this repository you find all the optimization steps we performed to reach a high performant vectorized implementation of a raymarcher.
The branches are labeled as master_optX where X goes from 1 to 5 (5 being the most performant of all).
No dependencies are required to run our raymarcher. Everything is self-contained.
First compile the project by running these commands from the root directory:
cd build
make
A file called raymarcher will be created.
In the build folder run ./raymarcher.
After the execution the output will be placed into the output directory.
If you want to simply compile and run all the predefined scenes, then run the following:
On UNIX:
team049$ cd build
team049/build$ bash run_raymarcher.sh A Scene is composed of geometric shapes that are each defined by a signed distance function. The combination of all these shapes creates a signed distance space. Our raymarcher algorithm is then applied on this space.
The creation of a scene is defined via a JSON file that gets parsed via a custom JSON parser. With this pipeline we built you can create and customize any scene you want.
The currently available scenes are in the team049/scenes/ folder.
Nihat Isik - (github: devnio)
Qais El Okaili - (github: Xemno)
Alexandre Cavaleri - (github: cavEpfl)
David Graf - (github: fxd24)