Apple’s image editing AI model has arrived albeit it’s far from a finished product. MGIE, short for multimodal large language models-guided image editing, is currently available on GitHub and there’s a PDF project paper with instructions.
The model uses text instructions to change and edit images. It can do simple image editing tasks like increasing contrast or brightness and changing the white balance. But it can also try and interpret more complex prompts – like making this pizza healthier – by adding vegetable toppings.
The model can also crop and resize, autofill image borders, change the subject’s hair, eyes, and clothes, and remove objects.
MGIE is just the first step in Apple’s generative AI journey. It’s not likely to make an appearance on any of Apple’s current devices, but it’s likely a sign of things to come.
