8000 GitHub - upf-gti/animics
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

upf-gti/animics

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
8000
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Animics

Animics is a web-based tool for generating 3D character animations, with a strong focus on sign language and embodied communication. It provides two separate workflows — Script Mode and Keyframe Mode — for building animations either procedurally or from video data.

🚀 This project is part of Tàndem, an open suite of tools for expressive, accessible 3D animation.


✨ Features

🎬 Script Mode: Procedural Sign Language Animation based on phonetic representations

Build animations by combining predefined "clips" generated procedurally from and extended Behavior Markup Language (BML) and adapted Signing Gesture Mark-up Language (SiGML) instructions. These clips cover both Manual Features (MF) and Non-Manual Features (NMF) — such as hand gestures, facial expressions, mouthing, gaze, and body posture — synthesized in real-time.

  • Based on SiGML and BML.
  • Animation sequences are constructed through configurable parameters for each action or instruction.
  • Focused on natural, linguistically grounded animation for sign language.

🧠 Public Linguistic Database

  • 1,722 glosses in NGT (Sign Language of the Netherlands).
  • 6,484 phrases in LSF-CH (Swiss version of the French Sign Language).
  • All entries are encoded in SiGML and BML (as a JSON), and can be used within Script Mode.
  • Users can create their own entries or integrate public datasets to expand the database.

📌 Script Mode and Keyframe Mode are separate workflows and cannot be combined.


📹 Keyframe Mode: Pose Extraction from Video

Generate animations from human motion captured in video.

  • Record yourself with the webcam or upload one or multiple video files.
  • Pose and facial data is extracted using MediaPipe:
    • Body landmarks
    • Facial Action Units
  • Data is mapped internally to:
    • Skeleton joint rotations
    • Morph target weights for facial expression control.
  • Includes Inverse Kinematics (IK) for precise adjustments. This feature can be toggled on or off.
  • Includes window-based propagation, which allows edits made in one frame to automatically influence neighboring frames within a configurable time window. This feature can be toggled on or off.

📌 Keyframe Mode and Script Mode are mutually exclusive per animation.


💾 Save, Load & Export

  • Users can register and log in to save their animations (both modes) to the server and retrieve them later for further editing.
  • Animations can also be created and exported without logging in.
  • Supported export formats:
Mode Formats
Script Mode BML (JSON), GLB
Keyframe Mode BVH, BVHE (extended BVH with facial), GLB
Keyframe Extras Raw MediaPipe data, original video

🚀 Import & Visualize Animations

  • Import animation files: Users can import previously saved animation files in both Script Mode and Keyframe Mode to continue editing or refine them.
  • Visualize animations in Performs: Generated animations from both modes can be directly visualized in Performs, the companion application of the Tàndem suite to play the animations in different avatars, customize the scene and be integrated in other applications.

🧩 Modular & Friendly UI

  • Minimalist, intuitive interface
  • Switch easily between Script and Keyframe modes
  • Real-time animation preview and editing

🛠️ Technologies Used

  • Frontend:

    • Lexgui.js - A simple and lightweight GUI library for creating graphical user interfaces for web applications.
    • Three.js - An open-source JavaScript library for creating interactive 3D and 2D graphics in web browsers using WebGL.
    • litefileserver.js - A front-end and back-end library that allows javascript apps to store resources.
  • Animation Engine (custom):

  • Pose Estimation:

    • Mediapipe - An open-source library from Google for building cross-platform multimodal applied ML pipelines.

Developers

Acknowledgements

We would like to extend our gratitude to the creators and maintainers of these libraries for their invaluable contributions to the open-source community.

Support

This project is being developed with partial financial support of:

EMERALD Project (2023-2026) SignON Project (2021-2023)
miciu logomaxr

Releases

No releases published

Packages

No packages published

Contributors 5

Languages

0