During the 'Ask Adam Frisby Anything About Space, His New Virtual World', session, Adam fielded a series of really interesting questions, among them was this one about Landmark's voxels.....
Q: "Why no great voxel build tools like in Landmark? are you stuck on the idea that only imported content & a marketplace can generate the best revenue stream? did not those ideas have a double edge sword effect in Second Life/OpenSim?"
Best tools for the job - I'll get back to voxel based tools in a moment, but mesh based build tools are the cutting edge still. We tend to look at what AAA video games are doing for what we should be doing - we're actively reading papers from SIGGRAPH, GDC, and so forth - and polygonal modelling isn't going anywhere.
Now, that's said - voxel tools are cool; Landmark's tools were written by someone I had a chat with a few years ago; I've even written some myself which are similar. There's a couple of problems:
- Instancing and batching. If you have lots of materials with different shaders on them, Voxels really hurt the GPU. You also can't really efficiently texture voxels in a precise manner.
- Storage and transmission; they're not efficient. You can compress them to a degree (and there's a lot of innovative ways people have come up to do that with). This also means streaming the content is kinda rough - it can be done (Landmark proved that), but if you also stream in textures and everything else, it gets kinda heavy - especially as you increase the spatial resolution
- You can build with those already through some Unity add-ins that are compatible with space. There's one called Voxeland which a few of us have been using. (I built a neat tropical island with that tool: https://www.flickr.com/photos/144994874@N05/31788925204/in/pool-sinewavespace/ a while ago)
- Animation. Voxels can't be animated easily. One of our big features is anything can be easily animated on a timeline visually, within the editor. Voxels are great for static content, not so great for anything dynamic.
Getting a voxel editor like the one I authored a few years ago, into the inworld build tools would definitely be cool - but for now, I think they're still pretty niche in the types of content you can make (great for terrain, with some really sophisticated tools you can kinda do architecture, but not much else); so I don't want to rely on them being the backbone of our tooling.
For reference, the old system I made: https://www.flickr.com/photos/144994874@N05/albums/72157683697918975
Here is the link to the rest of the Q&A: https://t.co/xe1t0F5Guh