Meta bundles its impressive Imagine artificial intelligence image generator with the new MetaAI chatbot, a significant upgrade compared to the previous version.
Imagine has a live generation feature that generates an image in front of you as you type, changing within limits for each new word. You can turn this into an animated GIF to demonstrate the creation process. This is similar to the live generation feature for StabilityAI-based models.
Once you're satisfied, click Generate and it will generate four images, often very different, based on the prompts you entered. However, the images are only square for now.
However, one of the standout new features is the ability to convert generated images into animation clips very quickly. It produces a few frames rather than a full video, so it's more like a gif than a film, but it's still an impressive new feature and a lot of fun to play with.
MetaAI Imagine has some features that are not available in other image generators, especially those built into the chatbot interface. This includes real-time previews, the ability to create videos of the generation process, and the ability to animate one or all of the generated images.
In addition to this, click on the image,[編集]It has a very efficient editing system that tells MetaAI how to change the selected image by clicking .
In one example, we generated an image of the Doctor Who TARDIS orbiting Earth, but the result was grey. So I clicked on edit and typed “make it blue” and the problem was resolved.
Images are created based on media such as Doctor Who, but not images of real people, especially those in the public eye such as politicians or actors. Even if you ask them to create a person that resembles a celebrity's face, they usually avoid creating such images.
Imagine is the same model used by other meta products like Facebook and Instagram to generate background images for chats with friends and posts. However, this is the first time it can be easily accessed as a standalone product to create fully realized artwork.
There are no chatbots that actually generate the images themselves. These act as intermediaries between the user and the image generation model. For ChatGPT it's DALL-E 3 and for Gemini it's DeepMind's Imagen 2. In MetaAI's case, it's a model called Imagine.
Imagine is impressive for several reasons. It's extremely fast, can create a variety of image styles, from surreal to artistic, and can generate generally readable text on top of images.
Built with the new Llama 3 large language model, MetaAI can provide comprehensive instructions to Imagine, even if the initial prompt from a human user is a bit limited in scope.
It takes a prompt, interprets what the user really wants, sends it to Imagine, and it returns some cool images. Initially it runs in real time and then produces a more rounded image.
I haven't used MetaAI Imagine as much as Midjourney or Ideogram, and I don't think it's inferior to either in terms of image quality or feature set, but it's as good as any image generator built into a chatbot. DALL-E 3 and Google Imagen 2.
It also has some features that are not readily available elsewhere, such as the ability to create gifs from generated images. But the best thing about it is the interface and how easy it is to use, edit, and adjust your photos. MetaAI does a great job of fast compliance.
Another advantage is that this is worse than ever. Llama 3 400B, arriving in the summer, will improve image quality, add new features and make the wide range of MetaAI assistants five times better.
