Hacker News new | past | comments | ask | show | jobs | submit login

Yes, no SVOs, but in a way sparse trees are used to store higher level information, only they are not octant-based.

Dynamic stuff is not up to the level of Voxatron, partly because it is just a lot more data to deal with, but you can have objects that move dynamically (as seen with windows and doors). That said, I may use old fashioned polygons for somethings like character animation, TBD.

I agree on the AI, I only want to discuss that I have plans to do some stuff with AI, and like I say on the site please be skeptical until you see anything. :)

One easy optimization I saw was that your fill order was rather arbitrary. Ideally you want to fill away from the camera, since the closer you are to the camera the more of the screen you are impacting.

Might want to look into a sparse system since storing 100% of the detail is great for demos like this but will result in huge amounts of wasted performance in most real applications, especially when working with mostly filled objects.

Overall a very impressive demo though.

Applications are open for YC Summer 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact