Meta, formerly known as Facebook, has launched a new generative AI experience called Imagine with Meta. This offline web-based tool allows users to create high-resolution images by describing them in natural language. Like OpenAI's #DALL-E 2, #Midjourney and #Stable Diffusion and other image generation models, Imagine with Meta uses Meta's Emu image generation model to generate images based on text cues.
The tool is currently free for users in the US and generates four images per request. #Meta aims to expand access to Imagine with Meta beyond chat rooms and make it available online for users to create free images.
Given Meta's past problems with imaging tools, there are concerns about the security measures for Imagine with Meta. Meta said that in the coming weeks they will be adding watermarks to the content they create to increase transparency and traceability. These watermarks generated by the AI model will be invisible but detectable by the corresponding model.
In the future, Meta plans to make invisible watermarks a feature in many of its AI-generated image products. The move aligns with growing pressure on tech companies to clearly indicate when works are created by AI, especially in light of concerns over #Deepfakes and AI-generated content.
Overall, Imagine with Meta offers users a great opportunity to create images using natural language, but it is still important for Meta to address any potential issues and provide transparency in the creation process.
