Treecapitating moretrees-mod trees...

[Ardonel]

Ardonel commented 4 days ago
VanessaE,

I am trying to understand the tree generation in the moretrees mod so I can add support for them to the Treecapitator mod. Is the file Tree_models.lua the file where you define the tree models? I had some of the capitating partially working, but I don't really understand how the moretrees trees are generated, so I am not sure how much changing needs to be done to the treecapitator mod to successfully capitate moretrees trees cleanly.

Thank you for your time.

VanessaE commented 4 days ago
Some of the models are defined there, while some are built up with some code in init.lua, using some simple decisions (e.g. jungletrees). Because of the built-in engine randomness of the spawn_tree() call being added to that, it's impossible to predict the shape of a tree, so you'll have to walk the tree nodes manually.

HybridDog,

If we are going to walk the tree, we could set up a table of single trunks and multi-trunks, and just walk up checking all 8 surrounding blocks for more of that particular tree/leave/fruit to change the border to include in capitation. It will still be interesting when two of the same type tree are close enough to mesh with each other.

[HybridDog]

l already did it.
https://github.com/HybridDog/treecapitator/blob/master/init.lua#L174
it seems that l made a mistake somewhere, sometimes capitating moretrees doesn't work correctly l think
What do you mean by 8 surrounding blocks?

It will still be interesting when two of the same type tree are close enough to mesh with each other.

if l remember correctly, l made it walk up the surrounding trees also and only remove nodes which are not in range/2 of the other trees, or l thought about implementing it

offtopic:
For me it's interesting what happens if you make many lines from a point and add the positions of a lines to a table (x and y), divide the contents of the table by the distance between the current position (because the value at the place the lines begun is the biggest one of course) and then use the result as heightmap.
pictures: https://forum.minetest.net/viewtopic.php?p=177043#p177043

[Routhinator]

Hmm, I just tested this mod and it didn't seem to work and I'm guessing it's because my biomes are using VanessaE's trees. With the way you have the code running, my entire forest would come down with one tree. The leaves and trees criss-cross each other like a real forest.

Perhaps this would be better solved by using the same method that LOTT and mg_villages use for protection. If VanessaE's trees generate IDs, then you can just check for that trees ID an remove the blocks that go with it.

[VanessaE]

Sadly I can't add IDs to the trees, because the engine's minetest.spawn_tree() call, which is used to grow those trees, does not allow for any kind of extra metadata to be added to the nodes it places.

[HybridDog]

@VanessaE could you calculate and use a custom seed when spawning a tree and save the seed into one stem node metadata?

[VanessaE]

In theory, maybe, but you'd have to write a full L-system interpreter and walk the tree node by node. Because it's impossible to know ahead of time what randomness gets used on a given tree (when that L-system feature is used), I don't think will work in practice.

[HybridDog]

Ok, thanks.

l think for moretrees l could implement tree head shapes. For star shapes it may be difficult to detect the tangible form, but if it works, it would greatly hone neighbour tree nodes detection.

[VanessaE]

That might work.