A while ago I was approached by my friend DJ of STEM Films to work on a project for AI Media. From the start I wanted to focus on creating a unique aesthetic that would lend itself to the the high-tech data saturated content that AI deals with. I came up with style-frames that were inspired by long exposures taken at night in urban areas, in this case the headlights of the cars representing individuals. At the same time though I wanted it to feel really small, almost like we were looking inside a processor of a computer. When you look at the master shot it is a densely populated almost chaotic stream of data. As we drill down deeper and deeper it becomes more organized and we are able to see leads, which are represented as orange lights.
The turnaround was short on this project. From the time I first started scribbling on the whiteboard to the time the final render was complete was around 5 days. This tight turnaround meant planning a pipeline accordingly. Originally we were going to use After Effects for the entire project but it wasn’t producing the results I wanted so I ended up doing most of the scenes in Maya. From there the renders were brought into After Effects for composting and final grade.
It was going to take 48+ hours on one machine but we were able to do it in 4-5 hours across 12 machines.
For the “city” that houses all of this data I wanted something that felt organic and very dense so I got on Google Maps and started looking a large cities, I eventually settled on Tokyo. In the end I think we crudely modeled a couple square miles of a section of downtown.
From there I used nParticles animating along a path for the lights. I wanted them to move in a very specific way which wasn’t possible with AE, so it was all done in Maya.
When it came time to render the final version of the project out of Maya, we were already down to the wire so we ended up going to my Alma Mater, UT Dallas, to use the render farm. It really saved us. It was going to take 48+ hours on one machine but we were able to do it in 4-5 hours across 12 machines.