Adobe unveils Firefly 3 with enhanced photorealism and styling capabilities for Photoshop

Adobe has unveiled its latest image generation model, Adobe Firefly Image 3 Model, featuring enhancements in photorealistic quality, styling abilities, detail, accuracy, and variety. The model is currently in beta and accessible through the Firefly web app, Photoshop, and InDesign.

Adobe's announcement emphasizes the model's ability to streamline the creative process, noting that it was designed to generate content safe for commercial use, similar to its predecessors.

The Image 3 Model reportedly generates higher quality images, interprets prompts more effectively, applies styles automatically to match prompts, and produces more accurate text in images. The model also works in tandem with the newly released Structure Reference and Style Reference capabilities, improving user control and visual quality. Additionally, the model has an improved understanding of text prompts and scenes, enabling image generations that better reflect complex prompts and include richer details.

Alongside the Image 3 Model, Adobe is rolling out the Generative Expand feature in the Generative Fill module of the Firefly web app, allowing users to alter the aspect ratio or size of an original image.

While this announcement is pretty exciting, the performance and realism of this updated model in comparison to competitors such as Midjourney, OpenAI’s DALL-E 3, and other competitors, remains to be seen.