Technology giant Google has announced significant upgrades to its Generative Artificial Intelligence (GAI) technology, Imagen 2, available to “allow listed” Vertex AI customers.
Significant upgrades to Imagen 2 include creating logos, multi-language support, and overlaying text and logos on generated images, whether it’s a business card or a beer can. Most impressively, and as most packaging designers using Midjourney or DALL-E 3 can attest, rendering specific text can prove challenging. Imagen 2 can now give users more control over this and present a cleaner non-gibberish piece of text.
According to Google’s announcement, “Imagen 2 can generate a wide variety of creative and realistic logos—including emblems, lettermarks, and abstract logos—for business, brands, and products.”
Moreover, Google says it’s also developed an indemnification strategy to protect its customers from copyright infringement liability. Imagen 2 also includes SynthID, an imperceptible watermarking technology that can detect GAI images. Tools to see this watermark are not available to third parties, however.
Google’s Imagen 2 is designed primarily for businesses, allowing Vertex AI customers to generate images for marketing purposes. The updates bring Imagen 2 closer to competitors like DALL-E but support multiple languages such as Chinese, Hindi, Japanese, Korean, Portuguese, Spanish, and English.
Customers already using Imagen 2 include Snap, which has updated its app with an AI Camera mode for its Snapchat+ subscribers. Shutterstock also uses Imagen 2 to power what it calls an “ethically sourced” AI image generator. Canva has also made Imagen available as an app on its platform.
Absent from the announcement are, unsurprisingly, details on how and what Google has used to train Imagen, nor was there any mention of an ability for creators and designers to opt out of having their work used to train the company’s AI technology.