Apple has released a public research project focusing on AI, specifically an open-source model that can edit images based on natural language commands. The model, called “MGIE” which stands for MLLM-Guided Image Editing, combines a large-scale language model with image editing capabilities to perform various image modifications, ranging from pixel-level adjustments to overall image editing.
In a recent research presentation, the model demonstrated its ability to interpret and fulfill specific commands to achieve the desired image modifications. For example, when instructed to make a pizza appear healthier, the model added additional vegetables to the image. It also showcased the capability to make detailed and precise modifications, such as removing people from the background or editing computer screens in the image.
The described capabilities of MGIE include overall image atmosphere editing, Photoshop-like instructions like cropping, resizing, rotating, adjusting brightness, contrast, and editing specific objects identified in the image.
MGIE is an open-source project that can be explored further on GitHub.
Source: VentureBeat
Here are some examples of image modifications performed by MGIE using different commands.
TLDR: Apple has developed an open-source AI model called MGIE, which can edit images based on natural language commands. The model showcased impressive capabilities in interpreting and fulfilling specific image modification instructions, from overall atmosphere editing to detailed object adjustments. MGIE is an open-source project available on GitHub.
Leave a Comment