# llimphi-anim

A Rive-style, **clip-agnostic** animation state machine.

It is the step above linear playback: instead of "play this animation", it models
"given these *inputs*, which state am I in and which do I transition to, with what
*blend*". That is what turns an animation *interactive* (hover, progress,
triggers) rather than a fixed loop.

## Why clip-agnostic

The core **knows nothing about Lottie or vello**. A clip is a `ClipId` (u32) with
a known duration; the machine sequences and blends clips by id and time, and
emits a `RenderFrame` saying *what* to render (the primary clip plus an incoming
clip with its 0..1 blend during a transition). The consumer maps
`ClipId → asset` and paints. Today that consumer is `llimphi-lottie`; tomorrow it
could be a native bone rig or a `llimphi-motion` tween — the same state engine
serves them all.

## Model (mirroring Rive's)

- **Inputs**: `bool`, `number`, `trigger` (triggers are consumed on each
  `advance`).
- **State**: a clip with speed and looping, plus its outgoing transitions.
- **Transition**: a target + (AND-ed) conditions over inputs + a blend duration.
  *Any-state* transitions are evaluated regardless of the current state.

---

Part of **llimphi** — see [llimphi](../README.md).