Adobe has announced a new feature for Photoshop that uses generative AI to expand images. The feature, called Generative Expand, allows users to drag the crop tool beyond the original picture size and add material with or without a text prompt. This can help when an image is simply too small, of course, but Adobe also believes it can help when you want to change aspect ratios, fix a cut-off subject or otherwise touch up artwork.
The Generative Expand feature is powered by Adobe's Sensei AI technology. Sensei is a suite of AI-powered tools that are used in a variety of Adobe products, including Photoshop, Lightroom, and Illustrator.
“Powered by the same Firefly-based technology as Generative Fill, Generative Expand enables users to seamlessly expand and resize any image by using the Crop tool. After clicking “Generate” the empty space is filled with AI generated content that naturally blends with the existing image.”
And specifically underpinning Generative Expand is the company's Firefly image-generating AI. Introduced in May, Firefly is a technology is trained on a massive dataset of images, and it is able to learn the patterns and relationships that exist in images. This allows the technology to generate new images that are realistic and visually appealing. It is available in Photoshop through the Generative Fill.
Currently Locked in Beta for Photoshop Users
The Generative Expand feature is still in beta, but it is available to all Photoshop users. To use the feature, simply open an image in Photoshop and then drag the crop tool beyond the original picture size. You will then see a prompt asking if you want to use the Generative Expand feature. If you click yes, Photoshop will generate new content based on the text prompt you provide.
“Suppose your subject is cut off, your image is not in the aspect ratio you want, or an object in focus is misaligned with other parts of the image — you can use Generative Expand to expand your canvas and get your image to look like anything you can imagine.”
Expanding Market of AI Image Generators
One of the leading AI image generators is DALL-E 2, created by OpenAI. It can make anything you can imagine, from a cat wearing a hat to a castle in the sky. Microsoft has also joined the AI image revolution by integrating DALL-E 2 into Bing Chat, creating the Bing Image Creator. This tool lets you make images from text right inside your chat box. You can also choose between “Precise” and “Balanced” modes to fine-tune your image generation process.
Microsoft has also made its Designer tool with DALL-E 2 available to everyone. This tool lets you make personalized designs from text, giving you a new way to express yourself.
Meanwhile, NVIDIA has been pushing the boundaries of generative AI research, developing new techniques to make AI-generated images more realistic and high-quality. OpenAI has also expanded the scope of AI in image creation with ShapE, a generative model that can make 3D models from text.
Another AI image generator is StableStudio, an open-source version of Stability AI's DreamStudio web app. StableStudio uses Stability AI's Stable Diffusion model to make images from text prompts, allowing you to create diverse and unique images. Stability AI has also c, a feature that lets you make multiple variations of an image using Stable Diffusion.
Meta joined the party June with its I-JEPA image generator AI. Image Joint Embedding Predictive Architecture (I-JEPA) was developed by Meta's AI research team, and it is based on the company's generative transformer model AI. The model was trained on a massive dataset of text and images, and it is able to learn the relationships between words and images. This allows I-JEPA to create realistic images from text descriptions.
Earlier this month, Alibaba launched its AI for generating images. Tongyi Wanxiang is a generative AI image generator that can create images from text descriptions in both Chinese and English languages. It can produce images with different artistic styles, such as watercolors, oil and Chinese painting, animation, sketch, flat illustration, and 3D cartoons.
It uses Composer, a large model developed by Alibaba Cloud, which allows users to adjust the image output parameters, such as the spatial layout and the color scheme, while keeping the image quality and originality high.
Last Updated on August 4, 2023 2:04 pm CEST by Markus Kasanmascheff
