LingBot-World overview
LingBot-World is an open-source world simulator built on Wan2.2. It generates video from an input image and text prompt, with optional camera-pose or action control.
Released models
| Variant | Control signal | Resolution |
|---|---|---|
| LingBot-World-Base (Cam) | Camera poses | 480P and 720P |
| LingBot-World-Base (Act) | Actions | 480P and 720P |
| LingBot-World-Fast | Camera poses | 480P and 720P |
The Fast variant uses causal inference with KV caching and processes frames in chunks. It supports minute-level generation and real-time interaction at 16 FPS with latency below one second.
Use Quick Start for installation, model download, required inputs, and reference inference commands.