Pose Animator: SVG animation tool using real-time TensorFlow.js models (github.com)

Article summary

Pose Animator is an open-source tool that animates 2D vector illustrations in real-time using TensorFlow.js models. It uses a skeleton-based animation approach, where the character's surface is defined by 2D vector paths and a hierarchical set of bones is used to animate the surface. The tool is supported on Desktop Chrome and iOS Safari, and can be used to create custom animations by downloading a sample skeleton SVG and embedding it into an illustration. The project is not an officially supported Google product, despite using Google's PoseNet and FaceMesh models.

Main themes

• animation technology
• open-source tools
• machine learning
• privacy applications
• animation industry
• content creation
• virtual webcams
• skeletal animation

What commenters say

• The tool has the potential to revolutionize the animation industry by making it easier to create animated character cartoons.
• Some commenters think that the tool is not yet ready for production use due to its current level of fidelity.
• The existence of similar tools, such as Adobe Character Animator, suggests that this type of technology is not new, but open-sourcing it could still have a significant impact.
• The tool's ability to create animated characters could lead to a proliferation of poorly made 'explainer' videos.
• Forcing commenters to explain their downvotes could improve the quality of discussions, but others argue that it would lead to unnecessary bickering.
• The importance of writing and voice acting in creating a successful show is emphasized, with animation being seen as a secondary aspect.
• The tool's potential for privacy applications, such as virtual webcams, is also discussed.
• Some commenters are interested in exploring the potential of this technology for creating animated content, such as South Park-style shows.