pouët.net

anyone know a good algorithm to draw trees? not the nature kind, the graph kind.

http://paulbourke.net/fractals/ifs_tree_b/

what about "not the nature kind" wasn't clear maali :)

if you spots IFS trees in nature, please gimme a call!

IFS often if not always uses chaos/random. What he wants is most probably (something like) this:



Which should be easy enough. Anyway, curb your attitude :)

b-trees!

Align depth along one axis, width along another. Boom, you have drawn a tree-structure.

I was thinking the depth would shove over other nodes of the same parent

A graph can be formally defined as a set of nodes. linking with a set of edges:

G = {N, E}

theres many graph-types too choose from, for whatever purpose those graph may have.

see graph listing here: http://en.wikipedia.org/wiki/Graph_%28mathematics%29

first make a graphnode-baseclass with index to the node which you can build from. then make a graphedge-class with offsets connections too and from the nodes, and a cost-variable (or calculate cost on the fly your platform is slow). then building a sparse graph-class and you're set to go experimenting with the different graph-types.

sigflup: i just remember some time ago you where in hospital. how are you now?

I'm well, I was in there for about a month over two stays. medication helps a lot. they said I'm schizoaffective and I am learning what this means to other people. I'll take a look at the links when I get home (on the bus)

maail wins thread!

oh absolutely, after all he was wrong, and then wrong again

for self-similarity in nature, go and buy a fucking romanesco broccoli.

You can use particle linked with springs, with 1/r^2 repulsing force between particles. It's simple but will make nice trees. Works also for graphs. You might look for further inspiration here http://www.graphviz.org/

plek, damn you!

in my defense, it was early in the morning.

i thought this would be a thread about weed. oh well.

marmakoide, that sounds correct. A slight weight downward as well

Yes, you have to throw some additional tricks to make it work. A downward force, yes. Also some small viscosity force (f = -kv, v being the velocity, k something close to 0) will help to dampen eventual oddities, like 2 particles coming very close. You also use springs across internal node to shape the tree to your convenience. It's rather flexible and tinkering-friendly.

For the particle motions, I recommend verlet integrator, it's cheap, unconditionally stable, we don't care much about accuracy and you get the viscosity force for free.

I like kusma's approach better to be honest, i.e. find the "widest" level or "widest" sibling group and then align accordingly.

If you ever need to store treestructure, consider implementing "nested set model"

I'm not sure how relevant to this discussion it is, but I've found it highly useful (and vastly time-saving) whenever I've had the chance to implement it. Not sure it applies directly to your need, though.

punqtured: that is nice!
i'd also use kusma's method. i usually do :)

Ok, I walked the tree downward and stored where the nodes are, for each parent I walk up parents shoving over the size that children take up.

For quickly toying around I can also recommend GUESS Graph Exploration System.