Some semi-random thoughts on terrain generation

[NiLSPACE]

I've tried it and spawned in a mega taiga but the podzol should not generate everywhere. More like this:

---

And the savanna M should be . (and realy rare)

[xoft]

I have only done the mesa biomes, no other.

[xoft]

How about this for megataiga floor, composition-wise?

[tonibm19]

Seems good. Waiting to see it with trees .

[xoft]

I'm having a mental block with trees. I can't decide whether it'd be better to rewrite the trees completely to use cBlockAreas instead of blocklists. It might be faster, but it also needn't. And it would allow us to define custom tree images, but then again, only the predefined images would be used, no random variety. What a dilemma.

[tonibm19]

It woild be awesome to use custom tree images, and if it's faster, better.

[xoft]

I keep wondering, how many chunks worth of biome data are queried to actually generate one chunk? The log output for cBioGenCache would suggest that quite a lot, perhaps we could optimize that somehow? Investigating now...

[xoft]

One chunk = 90 biome queries total, 25 unique queries. That's quite a lot. No wonder that the cache adds so much performance gain

[xoft]

If we ever want custom tree images, or any custom-built structures, we need a server where they could be built. And for such a server, we need a plugin that will keep people apart from each other and will generate the starting plots for everyone. I'm starting work on it now.

[bearbin]

I could run the server for a month or 2 if you want me to. (Would be raspberry pi, as it's cheap and always-on ;P)