The State of AI Animation in 2025

AI animation has gone from “uncanny curiosity” to “actually usable in production” frighteningly fast. Here’s my current map of the landscape.

Video generation models

The big story is consistency. A year ago, characters would morph between frames. Today’s models hold character identity across shots, which is the difference between a demo and a tool.

The pipeline is shifting

Traditional animation pipelines are being reshaped:

Stage	AI role today
Storyboarding	Fully AI-assisted
Keyframes	Partially automated
In-betweening	Nearly solved
Compositing	Mostly manual

Manga-style generation

This is where things get wild. Panel-to-panel consistency for manga is still the hardest problem — keeping a character on-model across pages requires either strong reference systems or LoRA-trained custom characters. But the pace of improvement suggests we’re months, not years, away from fully AI-assisted serial manga.

What’s next

I’m most excited about real-time generation — animation that renders as fast as you can storyboard it. The gap between “idea” and “watchable output” is collapsing.

If you’re experimenting in this space too, I’d love to compare notes — get in touch via the contact link in the footer.