PROCEDURAL VEGETATION
I began experimenting with growth.
I took nodes
as a starting point.
I recursively added further child
nodes to build a network of branches.
I required an algorithm capable not
of physically accurate growth but of visually
pleasing vegetation.
I took the concepts of gravity
stimuli and recursion.
I began experimenting with constraints
to produce vine-type structures.
The vines would creep along floors
and drape over edges.
They appeared natural.
The ray-casting method was just too unstable—
vines would exhaust space
and intersect scenes.
I created growth based upon sun
direction & previous growth. I cast a ray
in that direction.
Then I discovered ivy
generation,
implemented brute
force against all triangles,
clamped the new child
to this surface.
The results were far more stable.
HI THERE
I started working on a procedural tree/
vegetation generator yesterday. It generates a vast
variety of trees/
plants/
shrubs dynamically.
Most are quite convincing.
Here is a video showing variations
of African acacias.
*
Great work. You generate very convincing trees.
Procedural is the future.
*
It would be cool to see a landscape
without bark or leaves.
(Am I the only one that got enjoyment
out of building a tree, flinging it
into a spin, then building
another and doing the same?)
*
Have you explored plant succession?
This would use natural systems
to bring about forests (the end game).
It would be rad if the longer you painted
the farther you succeeded, or if erasing
would turn back time.
(This would actually be quite simple to implement, I think.)
PROCEDURAL VEGETATION DISAPPEARS
Procedural vegetation
is not viewable from great distances.
It's basically just a bunch of palm trees,
cliff bushes and other stuff.
It doesn’t really work. So what's the trick
so it won’t disappear?
Procedural veg is only good
for small things like grass,
stones.
I took the tree generator. It’s a bit laggy but
I can generate trees.
I also added a fern factory.
Sounds great.
I'm guessing this is the finished product?
I'll have to try this out until
we figure out how to merge.
Have you considered just preloading a couple hundred
or so procedural trees?
With enough possible trees
and no two trees in sight ever
being identical, it would be impossible
to notice repetition.
You could also regenerate some of the trees over time
as you transition to a different
world.
NOTES
I recently learned of a 3D modeling technique called “procedural
vegetation.” To my limited understanding, its technological basis is a
simple program written for the purpose of creating“natural” growth
patterns by setting up a high number of chance operations to occur
during rendering, in combination with the use of such mathematical-
but-biologically-occurring devices as the Golden Mean and Fibonacci
numbers. The final animated effect thereby mimics the look of “real”
vegetative growth with its irregular, “natural” patterning. Procedural
vegetation is the best method currently available for generating
“convincing” flora in the 3D environments of gaming and film.
All language in this series was culled from online gaming articles
and fora.
Sarah Heady is a poet and essayist interested in place, history, and the built environment. She is the author of Corduroy Road (dancing girl press, forthcoming 2020), Niagara Transnational (Fourteen Hills), winner of the 2013 Michael Rubin Book Award, and Tatted Insertion, a limited edition letterpress chapbook with artist Leah Virsik. She is also the librettist of Halcyon, a new opera about the death and life of a women's college, currently in development with composer Joshua Groffman and producer Vital Opera. Her manuscript "Comfort," a poetic meditation on female solitude, agency, and relationship set on the prairies of the American West, was a finalist for the 2019 Ahsahta Press Sawtooth Poetry Prize and the 2017 National Poetry Series. She lives in San Francisco, where she co-edits Drop Leaf Press, a small women-run poetry collective. More at www.sarahheady.com.