I have been working in CG for 20 years now, and I have seen a lot of different pipeline configurations, more or less effective.

Nowadays, a small show has 300 shots and big one 1000+. A strong pipeline is increasingly important and that's the one thing pretty much everybody agrees on. But everybody has their idea of  what a strong pipeline should be, depending on their skill set and experience.

It is tempting to create an all-encompassing framework that will categorise, validate and organise data flow. We have seen countless time how easily show can veer off the road and those moments what we need is to adapt.

We know the ingredients of a typical show

• We are routinely tasked by our client to accomplish things that are either very difficult or impossible.
• It is impossible to predict how a show will evolve. Entire sequences could be cut or added at the last minute, the original director fired, deadline shortened, etc.
• The best laid plans only work to an extent. There is always an unforeseen technical glitch, timeline change, etc.
• Every show is a race against the clock, a time versus quality trade-off.

However prepared we might think we are, we're not.

There are a few questions we need to ask ourselves

• Can we define a global data flow that will fit every single show ?
  • I used to think so. Nowadays I see agility as the main quality we should try to achieve. Every single show has differ requirements and we need adaptable solutions. A pipeline should not be a single tube with exactly one entry and one exit.
  • We can pipeline sub-parts of the data flow in an attempt to guarantee quality and make our artists' lives easier. This would include :
    • Enforcing global conventions (color spaces, naming, publishing, etc.)
    • Pre-packaging sequences of actions to make some organisational tasks less tedious and more predictable. (Publishing, asset validation)
• Where are our main bottlenecks ?
• What happens when something totally un-expected happens ?
• What about user-experience ?