Adobe makes its text-to-video feature in Firefly generally available. It now also allows its users to generate images using models from other companies, such as OpenAI and Google.
Adobe is gradually transforming Firefly from an AI image generator into a general platform for generative AI. This evolution becomes clear during the Adobe Max conference in London. There, Adobe announced the general availability of Firefly’s text-to-video feature and a series of new integrations.
Generating Images with GPT Imagen 3
A quick refresh: Adobe Firefly launched in 2023 in response to the image generator DALL-E. Firefly users can now not only choose from Adobe’s own models such as the new, hyper-realistic Firefly Image Model 4 and 4 Ultra, but also generate with external AIs. OpenAI’s DALL·E, Google’s Imagen 3 and Veo 2 are now available. Other popular alternatives such as Runway, Pika, and Luma will be added soon.
The announcement states that in addition to model selection, Firefly also introduces “Boards”. This allows users to combine multiple images into new prompts. Photoshop gains even more AI functionality with ‘Composition Reference’ and improved selection tools. Premiere Pro expands its generative capabilities to automatically extend scenes in videos, and introduces automatic subtitles in 27 languages.
Finally, Adobe is strongly focusing on ethical AI applications. With the new Content Authenticity app, creators can verify their work and indicate that their creations should not be used for training AI models.