Welcome to the website!
If you are reading this, you know the problem:
This website is about the common infrastructure patterns
we all keep re-inventing.
It is very much in progress,
but there is enough here to give a sense of the project.
Goals for the Project
We need to be able to talk to each other.
We need definitions for things like Asset and Pipeline.
Diagrams and Graphical Vocabulary —
i.e. boxes and lines on the whiteboard.
We need to draw pictures that actually convey meaning.
Click on diagram to Zoom/Unzoom.
We need a framework for our concepts,
so we don't have to start from scratch on every pipeline problem.
We need a (growing) list of Design Patterns that help shape our thinking
about complex pipeline problems.
Case Studies —
We need real-world examples, described in this emerging common vocabulary.
I will start with a few that I know, from working on them, or consulting,
or just from conversations over coffee at SIGGRAPH.
I'll propose some ideas for all of these. With your help we'll refine the ideas.
Over time the website will become a clearinghouse.
About the Project
A quick history:
Here's me on LinkedIn.
Pixar — 1996-2015
I spent 19 years at Pixar, most of it doing pipeline stuff.
(At first, we didn't call it pipeline. We had no word for it.
We were just moving files around, adjusting search paths,
making things up as we went along.)
Big software re-write — 2010
We rolled out a next-gen project that changed everything in the pipeline.
(I'm guessing most of my readers can relate.)
And I was a curmudgeon—I felt we were throwing out
all the hard-earned knowledge we'd gained over multiple films.
But I was unable to make my case.
I literally had no vocabulary for it.
And that got me thinking: what was that hard earned knowledge?
Could I write it down? Could I formalize it to be something more than a gut feeling?
Somewhere in there I re-read Design Patterns,
and that got me thinking about patterns.
OpenSubdiv — 2012
I originated and led this effort.
A big part of that was evangelism.
I got a chance to visit lots of studios and pitch them
on subdivision surfaces.
The conversations invariably started with:
“How do you make a simple prop, say, a CoffeeCup?”
This led to pipeline discussions, and I realized we were
all doing the same things and calling them different names.
I gave a talk suggesting that our discipline was maturing.
It was time to standardize vocabulary, codify our design patterns,
speak about best practices, etc.
This talk, and a follow-up at DigiPro 2015,
generated a lot of positive response.
Thus emboldened, I did something crazy:
I left Pixar to work on a book about Pipeline Patterns.
PipeGadget — 2016-17
As I talked to people, they said,
“We like those diagrams! You should sell that as a software package.”
Then they said, “These diagrams would be even cooler if they actually drove the pipeline.
If you build that, we'll buy it.”
So I put the book on hold, and with a great friend, Hans Hermans,
built a prototype.
Then the prospective customers said,
“Wait, you want us to rewrite our pipeline so it can be driven by the diagrams?
Are you crazy?”
I still believe in the idea,
but perhaps the industry is not ready to think about pipelines in terms of maps.
So maybe what's needed is a definitive reference on how to build pipeline maps.
Book / Website — 2018-onward
And so I'm back to the book, but now it's a website.
It can grow as it needs to.
I can add new stuff and fix broken stuff.
I started in LaTeX a few years ago, but recently
switched to a combination of python and markdown—which I extended to
handle site navigation and diagram creation.
I'm translating the LaTeX pages and adding new content frequently.
Check the release notes to see what's new.
I've checked the website in Chrome and Safari.
Let me know if it looks wrong elsewhere.
Various Other Stuff
These are various pages that are outside the main flow of the book/website.
A list of uploads to the website.
I'll be releasing new pages frequently.
Check here to see what's new.
A system for assetizing pipeline diagrams and using them to drive pipelines.
This is the software project I delayed the book for.