I always wanted to have my own story, my RPG. So I started this journey, little by little with my kids, for fun.
Biomes, weather, NPCs, animals, etc... A long way ahead, but I hope someone will like to contribute and work towards a RPG engine, with open source assets, code, sounds, rigs and all this!
The next part I'm working is using Azgaar maps to generate the full map (nearly done), build structures like in tiny glade...
I was struggling in a Rust project to build surfacenets with different LODs, I spent weeks trying to stitch them with skirts, geomorphing shaders etc...
Then I read about continous LOD with decimation and I decide to build a threeJS with wireframe view to validate it.
The plan is to validete it before moving into a larger engine: generate same-resolution chunk meshes, weld them into LOD0 pages, merge 2x2 children upward, lock the outer page border, simplify with meshoptimizer, and render a runtime quadtree cut based on screen-space error.
What I have now:
This is still a quick prototype, not a finished. I’m mainly using it to find the practical problems: page-boundary shading scars, simplification cost after edits, how stable the runtime cut feels, and whether the border-locking approach holds up visually.
Repo:
https://github.com/danielsobrado/drusniel-voxels-web
Demo:
https://danielsobrado.github.io/drusniel-voxels-web/
I’d be interested in feedback from anyone who has worked on voxel terrain LOD, Surface Nets, Transvoxel-style approaches, mesh simplification, or terrain editing pipelines.
Still struggling with grass performance, this is a side project for fun, I have no background on game programming.
I was struggling in a Rust project to build surfacenets with different LODs, I spent weeks trying to stitch them with skirts, geomorphing shaders etc...
Then I read about continous LOD with decimation and I decide to build a threeJS with wireframe view to validate it.
The plan is to validete it before moving into a larger engine: generate same-resolution chunk meshes, weld them into LOD0 pages, merge 2x2 children upward, lock the outer page border, simplify with meshoptimizer, and render a runtime quadtree cut based on screen-space error.
What I have now:
This is still a quick prototype, not a finished. I’m mainly using it to find the practical problems: page-boundary shading scars, simplification cost after edits, how stable the runtime cut feels, and whether the border-locking approach holds up visually.
Repo:
https://github.com/danielsobrado/drusniel-voxels-web
Demo:
https://danielsobrado.github.io/drusniel-voxels-web/
I’d be interested in feedback from anyone who has worked on voxel terrain LOD, Surface Nets, Transvoxel-style approaches, mesh simplification, or terrain editing pipelines.
Still struggling with grass performance, this is a side project for fun, I have no background on game programming.