Still, I guess that’s the expected outcome from spending a couple of days with your peers listening to a variety of speakers whose objective it is to make you think.
Ex-Weta Digital technology head Jon Labrie gave a very cool talk on his role at Weta. At the beginning of his presentation he was almost apologetic that he wasn’t going to be very business-oriented. In the event, nothing could have been further from the truth.
Sure, Labrie’s business is pretty different to the rest of ours and none of the other speakers were able to start their presentations with a sneak preview of the next instalment of The Lord Of The Rings, but at the end of the day, he had a bunch of business processes to support just like the rest of us and his story was all too familiar.
When he was provisioning Weta for LOTR, he had some industry metrics on which to base his calculations. Based on the average shot length, average storage per shot, average shots per artist, average render time per frame, average layers per shot and average renders per layer per shot, he was able to calculate the amount of processing power and storage he had to provide. Guess what? He was well short. Why? In reality, the way the facilities were used by the artists was very different to the way he had anticipated they would be.
The migration of data between online and offline storage became a major issue. Getting stuff shifted around took time and became a bottleneck. Same deal with the rendering facility. At first Labrie tried a rules- and enforcement-based approach. But, as you do in any environment populated with smart users, this turned out to not work so well either because they soon found ways to work around the best laid management plans.
Eventually Labrie stepped back and took a long hard look at what the required outputs were. It turned out to be pretty simple, really. In the visual effects world, the final quality of the work depends on the number of work-view-review iterations the work can go through. The artists needed direction on their work to progress it through the next iteration. For them to get direction, their work had to be viewed by the creative directors at the morning dailies. For their work to make the morning dailies, the infrastructure needed sufficient capacity to process and manage the volume of data.
The deadline was fixed – the movie’s release date was an immovable target – so stretching the required number of iterations out over a longer period wasn’t an option. The capacity of the infrastructure had to grow and Labrie produced a business case for more disk and processors on this basis.
It became a simple case if IT supporting and augmenting the business rather than trying to be prescriptive. Sometimes you have to accept that, while there might be a better way, any given business generally behaves the way it does for good reasons, and imposing IT-based constraints on the business isn’t going to help anybody. Least of all you.