The Workflow Is Already Splitting
People keep talking about AI in VFX like it's a future event. It isn't. It is already inside the workflow, just not in the final credits yet. Right now the biggest shift isn't replacement. It's fragmentation. Teams that know how to use AI for ideation, cleanup, previs, and look exploration are suddenly operating at a very different speed than teams still treating every brief like it's 2018.
What most people miss is that speed alone doesn't create better work. It just creates more chances to make bad work faster. Taste, supervision, and sequencing matter more than ever because the tools are flattening the gap between "possible" and "publishable." The bottleneck is moving upstream, away from execution and toward judgment.
Taste Becomes Infrastructure
For 11 years, post-production rewarded people who could survive repetition. Now the advantage is shifting toward people who can build systems. I care less about whether a tool can remove a roto task and more about whether it can preserve continuity across 200 shots without turning the project into aesthetic soup.
That is why I think the next serious VFX pipelines will look more like this:
`txt
reference capture -> intent map -> AI assisted exploration -> human supervision -> final frame discipline
`
If that sounds obvious, good. The obvious part is exactly what most teams still haven't operationalized. The studios that win by 2028 won't be the ones with the most models. They'll be the ones with the clearest creative controls.