Apple released an open-source AI image editing model
Apple researchers have released MGIE, an open-source multimodal AI model capable of editing images based on simple, natural language commands. This development highlights Apple’s quiet but steady progress in artificial intelligence, particularly in bridging the gap between vague human instructions and precise digital manipulation. By leveraging large language models to interpret expressive prompts, MGIE offers a more intuitive user experience than current tools that often fail with brief descriptions. The model’s ability to infer context from minimal input allows for sophisticated edits, such as modifying a pizza image to look healthier or adding specific atmospheric effects to a landscape. This capability addresses a common limitation in existing AI image editors, demonstrating significant potential for making creative software more accessible to non-expert users. This release is highly relevant to open_data as it provides the community with a transparent, accessible tool for studying multimodal AI applications. By publishing the model on GitHub and Hugging Face, Apple contributes valuable resources to the open-source ecosystem, encouraging collaboration and further research into how AI can better understand and execute complex visual tasks.
Source: mashable.comPublished on 2024-02-10
Related news
- ITWeb TV: Open-source AI: Unlocking potential for South Africa's SMEs
- China’s fastest deflation in 15 years poses risks - Financial Mirror
- Lynx.MD Builds Massive Gastroenterology Data Repository For Research and AI Development | BioPharma Dive
- Here's a quick glance at unemployment rates for January, by Canadian city