The basic building blocks of pipelines.
Cache
Saving an intermediate result to avoid recalculation, latency, or dependency craziness.
Bake / Baking
Baking is caching one set of data in the context of another.
Conform
Changing representations on an aggregation.
I have just a few patterns done.
I chose these because they were, for various reasons,
a good exercise of the tech stack and diagramming constructs,
What follows is planned content.
These mostly came from analyzing the various case studies.
I hestitate to mention planned content before it is ready,
but several people have asked me what the book will cover.
Planned Representation Patterns
Assembly and Schema
Building structure from components.
Aggregation
Bringing lots of stuff together.
Baked Response
A type of baking involving two applications of the data,
a cached (baked) pre-calculation, and a final calculation.
(Better definition needed!)
De-Duplication
Optimized storage.
Extended Schema
The full-pipeline footprint of an asset.
Flattened Representation
Getting rid of hierarchy and structure.
Sidecar
The out-of-schema component.
Streaming Formats
Organize the data in the order it needs to be read.
Binding
Mapping the execution parameters of one object into the attribute schema of another.
Build
Converting a source assembly into a run-time context.
Cache, Access
Optimizing the lookup of data.
Cache, Execution
Optimizing the calculation of data.
Harness/Adapter
Making one representation look like another.
Instancing Engine
How to represent lots of stuff.
Level of Detail
Different representations for different visualization requirements.
Planned Revision Patterns
These define how a pipeline handles collaboration and review.
How do artists share work, and—just as important—how
do they work in isolation?
Publishing
Making work available downstream.
Subscription
Acquiring work downstream
Convert-on-Publish
Could be thought of as a representation pattern.
Convert-on-Subscribe
Could be thought of as a representation pattern.
Cached Context
A quasi-assetized revision state, useful to save and restore.
Asset Management
Immutability, Coherence, Schema, ...
Indirection and Pin
Hiding the explicit revision.
Override / Sequential Resolution
For example, path-based file overrides.
Query-Based Resolution
Using metadata, tags, or labels.
Sandbox
The ultimate collaboration design pattern.
Push vs. Pull
Push is not always better!
Planned Variation Patterns
These define how a pipeline handles artistic complexity.
Object vs Instance Variation
Variation that can be subsumed in the object definition,
vs the irreducable complexity of parenting and placement.
Attribute Override
Redeclaration.
Activation / De-Activation
In the scene but not visible.
Speculative Attribute
Setting an attribute that doesn't exist yet.
Attribute Groups
Orthogonal Attribute Groups
Tall vs Short, Male vs Female, Blonde vs Brunette.
Procedural Variation
How to get a lot of variation in the scene.
Variant
Giving a set of specialization attributes
a distinct name or handle, by various means.
