Design
Why We Still Keyframe by Hand
Every week this summer someone sends our motion team a generated video clip with the message "is this the end of you guys?" and every week the answer gets easier to give. Look at how it moves. Really look. The lighting is plausible, the textures are impressive, and the motion is wrong in a way that is hard to name but impossible to unsee.
Motion design is the discipline of timing, and timing is where the current generation of tools falls apart. So no, it is not the end of us. But it is changing how we work, and it is worth being precise about what changes and what does not.
Motion is a language, and most generated motion is mumbling
When something on screen moves, your brain reads intent into it before you consciously register anything. A logo that settles with a soft overshoot feels friendly. The same logo snapping to rest on a sharp curve feels precise and technical. Ease curves are tone of voice. This is why serious brands are building motion into their identity systems, right next to typography and color.
Generated video has no opinion about any of this. It averages the motion of everything it has seen, which produces movement that is smooth, weightless, and characterless. Objects drift when they should snap. Fabric floats like it is underwater. Nothing has mass. For a background texture on a stage screen, fine. For the two seconds where a brand mark resolves at the end of a film, absolutely not. Those two seconds are the brand.
Where the tools genuinely help us
I want to be fair here, because our pipeline has changed a lot this year and some of it is wonderful:
- Previsualization is unrecognizable. We can show a client three moods for a 3D sequence in a day, work that used to take a week of styleframes.
- Cleanup and rotoscoping, the tedium that burned out a generation of junior artists, is largely automated now. Nobody mourns it.
- Texture and environment generation for 3D scenes saves us real days, as long as a human art-directs the result hard.
What we refuse to automate is the animation itself: the keyframes, the curves, the twelve frames of anticipation before a move. When we rebuilt the motion system for Aperture's product launch this spring, the entire personality of the work lived in decisions smaller than a tenth of a second. There is no prompt for that. There is only a person scrubbing the timeline at two in the morning saying, not yet, not yet, there.
The bar just moved up
Here is what I actually think is happening to our field. The flood of generated motion is not replacing craft, it is exposing it. Audiences are drowning in smooth, empty movement, which means work with real timing and real weight stands out more than it did two years ago. Slop raised the value of the thing it imitates.
For brands, the practical takeaway is to treat motion as an identity asset, not a production afterthought. Define your ease curves. Decide how your brand accelerates, how it settles, whether it bounces. Document it with the same rigor you document your typography. Then defend it, because every generic template and every generated clip will pull toward the average, and the average is now free.
We keep a plaque in the motion room, a quote from an animator on the team: machines interpolate, animators decide. That is the whole argument. The tools are welcome in our pipeline. The decisions stay ours.
Building something this could apply to?
We take on a small number of flagship projects each quarter.
Start a project