In a stride towards fostering creative exploration responsibly, Google introduces ImageFX, a cutting-edge image-generation tool utilizing Imagen 2, the latest text-to-image model from Google DeepMind. This innovative tool aims to provide users with a robust interface for safe and swift image creation, pushing the boundaries of generative AI. Alongside ImageFX, enhancements have been made to the existing generative AI tools, MusicFX and TextFX, to offer users an even more seamless creative experience.
ImageFX: Unleashing Creative Potential with Simple Text Prompts
ImageFX, the latest addition to Google’s Labs, allows users to craft images effortlessly using simple text prompts. Recognizing the significance of creative exploration for new users of generative AI tools, ImageFX facilitates experimentation by incorporating an intuitive prompt interface with “expressive chips.” These chips empower users to swiftly experiment with various dimensions of their creations, fostering inspiration through diverse prompts and concepts.
MusicFX: Elevating Generative AI Text-to-Music Experience
Building on the success of last year’s release, MusicFX, a generative AI text-to-music experiment, now enables users to create tunes up to 70 seconds in length and music loops. With expressive chips, users can explore prompts and share their creations with friends. Notably, over 10 million tracks have been created since the tool’s launch, and recent updates promise higher-quality audio and faster music generation, with early testing revealing a threefold preference for the new music samples.
TextFX: Inspiring Writers with Generative AI
TextFX, a collaborative effort with Lupe Fiasco, serves as a generative AI experiment for lyricists, wordsmiths, and creative minds. This tool, designed for exploring creative possibilities with text and language, has received usability updates to enhance the overall user experience and navigation, further empowering individuals to express their creativity through writing.
Also, Google emphasizes its commitment to developing technologies responsibly, aligning with Google’s AI principles. With Imagen 2’s ability to generate photorealistic images, Google has invested significantly in training data safety and implemented technical guardrails to mitigate problematic outputs. This includes filtering out violent, offensive, or sexually explicit content. The company conducts thorough adversarial testing to identify and address potential harmful and problematic content.
To ensure transparency and identification, all images generated with ImageFX and songs produced using MusicFX are marked with SynthID, an imperceptible digital watermark developed by Google DeepMind. Additionally, ImageFX-generated images include IPTC metadata, offering more information to users encountering AI-generated content.
These tools, ImageFX, MusicFX, and TextFX, are now accessible in the U.S., New Zealand, Kenya, and Australia. Curious minds and creative enthusiasts can explore these experiments and more by visiting labs.google, continuing the journey of responsible and inspiring generative AI.