The Algorithmic Beauty of Plants (1990) [pdf]

I love L-Systems. I somehow obtained a couple of VHS tapes in the 90s when I was a kid of early computer graphics, and was really taken by the generative artists — like Karl Sims. This was also the heyday of big bookstores like Borders, and I’d spend hours in the “Computers” aisle, where I found “Advanced Animation and Rendering”, which covered L-Systems.

These are fun things to code because they can surprise you. Even when you understand the rules, it can be delightful to see what grows.

Anyone interested in fooling around with algorithmic art should check out the Context Free Art software/language: https://www.contextfreeart.org/ ref: https://github.com/MtnViewJohn/context-free/wiki/About

The primary source for anyone interested in L-Systems. I can also recommend looking into Gielis superformula [1] for e.g. generating realistic petals and leaves on the trees.

[1] https://www.researchgate.net/publication/258238188_Inventing...

As someone who has invested a career into 3D modeling of plants, particularly of real plants from 3D scanner data, I really don't find value in this work, scientifically. It surely is fun and useful for artistic purposes, where the only evaluation is from a human eye. But when you actually compare the structures of the algorithms presented here to the real curves and topologies of any living plant, you find that they are missing a tremendous amount of detail. From a machine learning perspective, e.g., it would be very hard to "fit" these models to actual data. This is because these algorithms and data structures assume far too much homogeneity in the basic structures of a given species, particularly that these structures be repeated with only simple rotations and permutations. In fact, when you look closely at leaves and flowers and roots and stems of a plant, there is more and more detail; just like any living system, this is because plants are very active on a microbiological level that causes a level of microscopic entropy well beyond that of any single Bezier curve (or whatever). So for those who really do love the beauty of plants, and who really do want to model them and therefore be able to propagate their forms in the digital space, I would suggest looking elsewhere, and in particular seeking ways that are more data-driven to model plants. As it stands in this work, there are no objective functions or optimizations, this is just purely art that is disconnected from the reality in myriad immeasurable dimensions.

I first read this book back when I was just starting programming, learning about fractals and chaos and graphics and recursion and all those good things. I got the hard cover version out of the library several times (just to show how long ago that was!) and implemented a few of the things in it. Very happy to see it available online!

I came across this book when exploring L-Systems based on this earlier HN post - https://news.ycombinator.com/item?id=21749806

If the book is inspiring it might be worth going through some of the insightful comments in that post.

I recall a SIGGRAPH around that time on synthetic life covering that material.

A nice thread from 2014: https://news.ycombinator.com/item?id=8723728.