I tested Adobe’s Firefly 3 image generation tool and was amazed by its photorealistic output

by

in

– Adobe has released Firefly 3, an AI image generation model with improvements in photorealism and prompt adherence
– Firefly 3 was trained on billions of licensed stock images to improve overall output
– Adobe is also launching generative expand for the Firefly web app, allowing users to change image canvas or orientation with Firefly filling in the gaps

Adobe has introduced the latest version of its AI image generation model, Firefly 3, along with enhancements to generative fill in Photoshop. The new version of Firefly offers improvements in photorealism, prompt adherence, and overall control of the final image. Trained on billions of licensed stock images, Firefly 3 includes detailed labeling for lighting, structure, and style to enhance the output quality. It is accessible through the Firefly website and will eventually be integrated into other Adobe products, such as Premiere Pro alongside AI video models like Sora.

In addition to generative fill, Adobe has also launched generative expand for the Firefly web app, allowing users to expand the canvas or change the orientation of an image while Firefly fills in the gaps. Firefly 3 has been well-received for its photorealistic images and quick ideation process, enabling designers to create fully-fledged images from ideas with minimal friction. The new version offers a wider variety of outputs across different styles, such as photo, art, and illustration, and includes options to adjust the mood and lighting of the image.

Adobe is investing in its Firefly web app and is committed to advancing the state-of-the-art with its Image 3 Foundation Model. Firefly 3 has the ability to mimic the style or structure of existing images, providing creators with more flexibility in their design process. The overall reception of Firefly 3 has been positive, with users praising its photorealism and speed in generating high-quality images. Adobe continues to push the boundaries of what is possible with AI image generation and looks forward to seeing how the creative community will utilize Firefly 3 in their work.

Source link