Using AI to generate a spine animation from a 2D image.

[shadowfox87]

Today, there are several image to video converters - Minimax and Kling which are the best at the moment. However, the output is a video which means it is an mp4 or gif. In the context of video game design and other applications, there is another alternative that is smaller in size - a spine.

A spine in 2D animation and graphic design refers to a rigging system that defines how different parts of an image move. It acts like a skeleton made of interconnected bones, allowing for smooth, dynamic animations without needing frame-by-frame drawing. Each bone influences specific parts of the image, enabling transformations like rotation, scaling, and bending. Spines are commonly used in game development, motion graphics, and interactive applications, where animations need to be lightweight and responsive. Tools like Spine 2D, DragonBones, and Godot's Skeleton2D use spines to animate characters, creatures, or objects efficiently.

Basically, take an image, it uses AI to determine what parts of that image should be moveable and splits that image into different body parts. Then, it writes a script which tells it how to assemble those parts and how to move it. This isn't a video. It's a script that is being generated. Unity and Godot engines use spines today but it takes several hours to weeks to months for a human being to create even 1 spine.

I find that this is an avenue that is unexplored currently in AI generation.