Adobe Introduces Firefly Image 3 Foundation Model to Take Creative Exploration and Ideation

• 24 Apr 2024
• TM Team
• 0 Comments

The Creativity Conference – Adobe debuted Adobe Firefly Image 3 Foundation Model, the next major release in its family of creative generative AI models, available  (beta) in Photoshop and the Firefly web application. Firefly Image 3 will bring massive advancements to quality and control to users. This includes higher-quality image generations, better understanding of prompts, new levels of detail and variety and significant improvements that enable fast creative expression and ideation. Adobe’s newest Firefly model delivers photorealistic quality like never before with better lighting, positioning, attention to detail, advancements to text display and more. 

Available today in the Firefly web application, users can access Firefly Image 3 in features including Structure Reference and Style Reference in the Text to Image module and a new Generative Expand feature in the Generative Fill module, which enables users to expand and change the aspect ratio of images. 

“In just over a year, Firefly has become the image generation tool of choice used by millions of creators to ideate every day, and we’re just getting started,” said Ely Greenfield, chief technology officer, Digital Media at Adobe. “As we continue to advance the state of the art with Image 3 Foundation Model, we cannot wait to see how our creative community will push the bounds of what’s possible with this beta build.”  

Firefly has been used to generate over 7 billion images worldwide since its initial debut in March 2023. Adobe built it for direct integration into workflows Adobe customers use every day, including Adobe Photoshop, Adobe Express, Adobe Illustrator, Adobe Substance 3D and now Adobe InDesign. In just one year, Firefly has transformed image editing, template creation, vector design and 3D texturing and staging for the better. With each new foundation model, Adobe releases the technology as a beta to its creative community to give feedback and advance the outputs.

The New Firefly Image 3 Foundation Model

Firefly Image 3 equips users with the power of fast ideation and experimentation. It allows creators to bring their visions to life more easily, work more productively and generate higher-quality, more detailed images. 

Firefly Image 3 delivers: 

• New levels creative expression with auto-stylization capabilities: Powered by a new style engine, new auto-stylization capabilities in Firefly Image 3 deliver higher quality outputs with more variety, giving users more control and personalization over the styles of images they generate. Image outputs include new varieties of styles, colors, backgrounds, subject poses and more, giving users a wide array of options for creative ideation and enabling faster exploration. 
• Advancing creative control with Structure Reference and Style Reference: Firefly Image 3-powered Structure Reference and Style Reference capabilities deliver exceptional user control and state-of-the-art visual quality. With Structure Reference, users can quickly generate new images that match the structure of a reference image, eliminating the trial and error of having to write the perfect prompt. Style Reference then delivers higher quality outputs, giving users more control and personalization over the styles they generate. When these capabilities are combined, users can reference both the structure and style of an image to quickly bring their ideas to life. 
• Stunning photographic quality: Firefly Image 3 delivers new levels of photographic quality with better lighting, positioning and more variety. Significant improvements can be seen in people rendering (with detailed features and a variety of moods and expressions), complex structures and crowds.  
• Astonishingly rich detail and prompt accuracy: Firefly Image 3 has a better understanding of text prompts and scenes, enabling image generations that accurately reflect long, complex prompts and include richer details. Text renderings are also improved, resulting in image generations with clear text displays. This makes ideation for assets like posters more efficient.   

• Broader range of styles for illustrations and icons: Firefly Image 3 makes significant improvements to illustrations in outputs enable the quick creation of icons, logos, raster images and line art.

Firefly Image 3 is available in beta. Users can access it via the Firefly web application at firefly.adobe.com. Adobe offers both free and paid plans that include an allocation of Generative Credits for the use of generative AI features powered by Firefly in Adobe applications.

Adobe Firefly was trained on licensed content, such as Adobe Stock. It was designed to generate content for commercial use that does not infringe on copyright and other intellectual property (IP) right such as trademarks and logos. Adobe uses a multi-layered, continuous review and moderation approach to block and remove content that violates Adobe’s policies and offers customers IP indemnification for Firefly generated content.

Accelerating Content Creation and Production Across Enterprises

Adobe recently introduced Adobe Firefly Services and Custom Models to fundamentally change how brands create and produce personalized content at scale. Firefly Services provide access to generative and creative APIs, tools and services for content generation, editing and assembly, which can be integrated into any production or workflow so organizations can maintain quality and control when automating the production of content at scale. In addition, Custom Models enable enterprises to train generative AI models based on their IP, product and brand styles to support brand consistency amongst creative and marketing teams.

Commitment to Responsible Innovation

Adobe is committed to developing AI in accordance with the company’s AI Ethics principles of accountability, responsibility, and transparency. In a recent Adobe study, 76% of U.S. consumers emphasized the importance of knowing if online content is AI-generated. To provide transparency around the use of AI, Content Credentials are automatically attached to Firefly generated content. Like a nutrition label for digital content, Content Credentials are tamper-evident metadata that can provide more information about the digital content, including whether AI was used in the creation or editing process. 

Content Credentials  are built on the C2PA open standard and supported by the Adobe-led Content Authenticity Initiative (CAI), which was founded in 2019 to increase trust in the digital ecosystem. Today, the CAI has grown into a global coalition of over 2,500 members across tech, policy, media companies, creative professionals, researchers and more, all working together to add transparency to digital content.