Stay as-is, or rewrite?
Hi, all two of you.
In the last half-hour, I spent some time doing something I should have done beforehand: google implementing voxels.
Turns out what we/I are/am doing is nowhere near the most efficient route, nor the one that is easiest over time. What we're currently doing is blindly adding voxels ('cause I'm a n00b) whenever we feel like it. What we could do is something awesome (and complex) based on octrees (or some other form of tree), that could be a lot more efficient and allow for things like procedural rendering and precision settings.
So, we've basically got three options:
-Continue as we are. I haven't actually noticed any problems yet with performance, and the idea itself is sound, just a little inefficient.
-'tack on' features that make it more efficient. Technically we can still implement a lot of the aforementioned stuff with what we've got, it's just a little more difficult/less efficient. For example, implementing octrees would be possible with a (current) voxel as the base unit.
-Go (almost) from scratch again. This is probably a better (if more complicated) approach. We could also re-use most of the code. However, there's the onerous burden of re-doing most of it, and we were getting somewhere with what we have now. (I think we're quarter of the way to something useful, or thereabouts).
What do you guys think?
Question information
- Language:
- English Edit question
- Status:
- Solved
- For:
- GlVoxel Edit question
- Assignee:
- No assignee Edit question
- Last query:
- Last reply: