Apple’s MGIE and Keyframer bring AI-powered creativity to images and animations

Apple recently unveiled two AI tools: MGIE (MLLM-Guided Image Editing) and Keyframer. These cutting-edge technologies promise to redefine the way users interact with images and animations, leveraging the power of artificial intelligence to simplify complex tasks and make creative endeavors more accessible to all.


Apple’s MGIE lets you edit photos with natural language commands, Keyframer animates art with text prompts

In partnership with researchers from the University of California, Santa Barbara, Apple has introduced MGIE, an AI model designed to revolutionize image editing through natural language commands. By harnessing multimodal large language models (MLLMs), MGIE interprets user instructions to perform precise pixel-level manipulations, offering a seamless and intuitive editing experience. From basic adjustments like cropping and resizing to advanced modifications such as object manipulation and color enhancement, MGIE empowers users to unleash their creativity without the need for specialized software.

Apple’s Keyframer represents a significant advancement in animation design, enabling users to animate static images using simple text prompts. Powered by large language models (LLMs), Keyframer generates CSS animation code from static SVG images and textual descriptions, eliminating the need for complex animation software. While still in its prototype stage, Keyframer holds immense potential for future integration into Apple’s product ecosystem, paving the way for intuitive and accessible animation creation. From a paper published by Apple researchers:

Keyframer is an large language model (LLM)-powered animation prototyping tool that can generate animations from static images (SVGs). Users can iterate on their design by adding prompts and editing LLM-generated CSS animation code or properties. Additionally, users can request design variants to support their ideation and exploration.

While one-shot prompting interfaces are common in commercial text-to-image systems like Dall·E and Midjourney, we argue that animations require a more complex set of user considerations, such as timing and coordination, that are difficult to fully specify in a single prompt—thus, alternative approaches that enable users to iteratively construct and refine generated designs may be needed especially for animations.

We combined emerging design principles for language-based prompting of design artifacts with code-generation capabilities of LLMs to build a new AI-powered animation tool called Keyframer. With Keyframer, users can create animated illustrations from static 2D images via natural language prompting. Using GPT-4 3, Keyframer generates CSS animation code to animate an input Scalable Vector Graphic (SVG).

keyframer 2

Both MGIE and Keyframer exemplify the demand for empowering creativity and democratizing design. By leveraging AI and natural language processing, these tools break down barriers to entry, making advanced editing and animation capabilities accessible to users of all skill levels.

Read more:

About the Author

Asma is an editor at iThinkDifferent with a strong focus on social media, Apple news, streaming services, guides, mobile gaming, app reviews, and more. When not blogging, Asma loves to play with her cat, draw, and binge on Netflix shows.

Leave a comment