The "Imagine with Meta AI" website allows users to generate images based on various prompts, with each generation creating four 1280×1280 pixel images that can be saved in JPEG format. The images include a small "Imagined with AI" watermark logo in the lower left-hand corner. The AI model can create photorealistic images and handle complex prompts, but struggles with text rendering and different media outputs. The images of people it generates include diversity in ethnic backgrounds.
Key takeaways:
- Meta has released a free standalone AI image-generator website, "Imagine with Meta AI," which uses its Emu image-synthesis model trained on 1.1 billion publicly visible Facebook and Instagram images.
- The AI model can render a novel image from a written prompt, with examples including a muscular barbarian with weapons beside a CRT television set, a cat in a car holding a can of beer, and a photorealistic Mickey Mouse on the moon in a spacesuit.
- Creating images using the new website requires a Meta account, and each generation creates four 1280×1280 pixel images that can be saved in JPEG format. Images include a small "Imagined with AI" watermark logo in the lower left-hand corner.
- Meta's model generally creates photorealistic images well, but not as well as Midjourney. It can handle complex prompts better than Stable Diffusion XL, but perhaps not as well as DALL-E 3. It doesn't seem to do text rendering well at all, and it handles different media outputs like watercolors, embroidery, and pen-and-ink with mixed results.