Meta recently launched a free AI image generator, Imagine with Meta, that uses the company’s Emu image-synthesis model. It’s essentially the same feature they introduced to their messaging apps a few months back, but now as a standalone website.
The tool lets you create images from natural language input, and the generative model has been trained with over a billion images from Instagram and Facebook, with aesthetic and decent results.
The new Imagine with Meta is very similar to any free AI-image generator you have heard of: it requires a login –you can use your Instagram or Facebook account or an email login– and it has a field to type in your descriptive text prompt. Once you hit the “Generate” button, it produces four image variations for the given prompt, in a 1280×1280 size and with a small “Imagined with AI” watermark in the corner.
This tool is powered by the corporation’s native image generative model, Emu, which also fueled the AI stickers for messaging apps and AI image editing features for social media platforms that were introduced in September and rolled out in October.
For now, it is free to use but available for users in the US only.
Where Emu –and Imagine with Meta– distinct itself from similar products is in the training datasets used to teach the software to synthesize visual content.
Unlike others like Dall-E, Midjourney, or Stable Diffusion, which use large training sets scraped from the web, or apps developed by companies like Adobe or Shutterstock, which use legally authorized and licensed datasets, Meta trained its image generation model with 1.1 billion publicly visible photos from Facebook and Instagram –that is, images that users of these platforms uploaded and shared in public mode– previously selected and refined.
This kind of training has interesting results. For example, some early users see a remarked Instagram-ey aesthetic in AI-generated people images created with Emu.
In addition, the firm is said to address potential harmful outputs with filters, which stop users from creating content about famous people or public figures or around concepts like violence or hate.
Meta is also taking on one of the current pressing issues of the AI-generated image field today, which is transparency and appropriate content provenance.
Right now, the images created by their free image generator come with a small, visible, and “traditional” watermark, but they expressed they will begin to implement an invisible watermark into the visuals created with Imagine with Meta in the near future.
This invisible watermark will identify the image as AI-generated and will be unaffected by common removal methods such as cropping, screenshotting, or filtering.
This way, the company is assuming a commitment to the responsible and transparent use of AI-generated media.
Are you going to try Imagine with Meta? Tell us what you think of its results!
I am an experienced author with expertise in digital communication, stock media, design, and creative tools. I have closely followed and reported on AI developments in this field since its early days. I have gained valuable industry insight through my work with leading digital media professionals since 2014.
AI Secrets is a platform for tech decision-makers to learn about AI technology. Our team includes experts such as Amos Struck (20+ yrs ICT, Stock Photo, AI), Ivanna Attie (expert in digital comms, design, stock media), and more who share their views on AI.
Get AI news in your inbox & join thousands of engineers and managers using AI to boost sales and grow market share