Brand systems at scale
Pixel-perfect brand compliance across 40+ languages: and a pipeline that made turnaround 5× faster with no quality loss.
Agency work: client assets are not shown here. This is the process story, which is the interesting part anyway.
Seven years directing visual output for Canon, Ferrero Rocher, and CAT across EMEA and global markets. The job sounds like design; the problem is logistics: multiple concurrent campaigns, each shipping in 40+ languages, every version perfect against brand guidelines.
The bottleneck
24 episodes, 40+ languages.
Manual multi-language content delivery doesn't scale well. Especially if there are multiple non standard alphabets in use. Every locale is a re-layout, every re-layout is a review cycle, and review cycles multiply by language count.
It was a very tight moment in general - Adobe dropped multi-core cpu rendering that year as well, moving to the Mercury GPU engine. This made a 4k 5min. animation render in brand new Media Encoder for 5hr instead of ~1hr.
Tight deadline did not allow us to assembly a new machine in time. But the solution was outside of the pipeline this time.
The fix
I designed and shipped a data-driven asset pipeline that standardised multi-language delivery: content and layout separated, locale variants generated from data using open source software.
This was first time I've used blender3d extensively in production (it's 2017) - extensible nature of the platform allowed our team to devise a way to render all non Latin OST using multi threaded compositor rendering capabilities and some scripting magic to reverse the order of reading.
Result: 5× faster turnaround with no quality loss: measured across campaigns that previously took the full manual cycle in a broken rendering pipeline.