Apple has recently been delving deep into the realm of AI to explore its full potential and chart a new course. The company is blazing trails that no one has ever taken, unveiling a revolutionary paper introducing their latest AI innovation, Keyframer.
This ahead-of-the-curve, modernistic innovation relates to Large Language Models (LLMs) that revolutionize animation creation by transforming static images into dynamic animations based on user-entered prompts.
Titled “Keyframer: Making the Most of Large Language Models to Stimulate Animation Design,” the paper dives into the complexities of integrating language cues with code generation proficiencies to empower users in constructing complex animations.
While one-sentence prompts suffice for basic AI services like DALL-E and Midjourney, intricate and advanced imaginary scenarios demand consideration of temporal and coordination factors, which traditional prompts fail to capture comprehensively. Keyframer bridges this gap by allowing iterative refinement of designs, particularly for animations.
By leveraging a combination of culturally inclined prompts and the code-generating dexterity of large language models like GPT-4, Keyframer emerges as a versatile AI animation tool. It empowers users to bring life into static 2D images through natural language inputs, wherein the system generates CSS animation code tailored to animate scalable vector graphics (SVG).
Apple showcased Keyframer’s prowess through a practical demonstration involving a space rocket-themed SVG image. Users simply input a prompt to depict a sky adorned with three gradient colors and a twinkling star animation.
In response, Keyframer promptly generated the necessary CSS animation code, which could be further emphasized and customized either through direct code editing or additional text inputs.
The author elucidates:
“Our aim is to catalyze the development of future animation design tools that combine the robust generative capabilities of LLMs with dynamic editors, expediting design prototyping and empowering creators to retain creative autonomy.”
This innovation encompasses a new era in animation design, where the fusion of language understanding with code generation facilitates flawless creation and customization, propelling creativity to unprecedented heights.
