livinamuk / Hell2025

HELL ENGINE: rewrite one billion++ This codebase began as a CPU raycaster to explore the viability of a voxel based idea I had for global illumination. The voxel idea didn't work out in the end, the voxel grid caused lighting artefacts, but rewriting it to use a pointcloud did and here we are. This is now the new Hell Engine, the engine I'm writing to house a splitscreen roguelike/deathmatch survival horror game. Controls WSAD: movement Left Mouse: fire Space bar: jump Q: cycle weapons E: interact R: reload Left ctrl: crouch F: Shotgun melee G: fullscreen J: Respawn dead players K: Respawn all players Z: previous render mode X: next render mode B: cycle debug lines O: reload map V: Toggle splitscreen mode Y: show probes 1/2/3/4: switch player -: Lower camera +: Raise camera ~: Toggle debug text TAB: Toggle editor Editor controls F1: File menu F2: Insert menu Alt + click: rotate camera Ctrl + click: pan camera Mousewheel: zoom camera **WARNING: editor menus very buggy still** Download via terminal Older commits (before 18/3/25) were managed by LFS: Usage Run in Release. PhysX runs like shit in Debug. --- December 27 2023 November 14 2023 Huge improvements since the entries below. I scrapped the voxel thing entirely because it was causing lighting artefacts for geometry that didn't fit nicely into the voxel grid, most importantly swinging doors. It now calculates direct lighting for an approximated point cloud of the scene which is in turn used to propogate light through a 3D grid to simulate bounced light, both these passes are performed with raytracing in custom compute shaders. I've got skeletal animation too now and we're running realtime baby. September 12 2023 Things are looking promising. I really need to move to Vulkan so I can do visiblity checks for each pixel against any probe that may contribute indirect light to it. Gonna see how far I can push this in GL first though. All direct and indirect light is still calculated on the CPU, would be nice to move to compute but there's really no point if I'm moving to VK soon. I'll work on voxelizing regular mesh next I think. September 10 2023 Indirect light is working. It's all done with ray-triangle intersection tests at this point, need to look into faster methods for geometry visiblity checks and light propogation. September 4 2023 I started a new project today, to explore using voxels to calculate and render indirect diffuse light at realtime speeds. So far direct lighting only. Will finish light propogation on the cpu, then move it all to compute if it looks any good. Rendered in OpenGL. Will likely eventually port to Vulkan and let them evaluate the raycasts for me, don't really see myself writing something faster, gonna have a crack in GL compute first tho.