Google Labs, Google’s experimental arm, is testing a brand new picture generator referred to as Whisk. This device permits individuals to immediate with pictures as a substitute of textual content, permitting them to remix a photograph by altering the topic, scene, and magnificence.
Whisk makes use of Google’s image-generation mannequin, Imagen 3, to mix three pictures: one for the topic, one other for the scene, and one for the model. As an illustration, you possibly can choose a photograph of your self as the topic, a futuristic panorama because the scene, and an anime model for the ultimate look.
The mannequin routinely generates an in depth caption of your pictures, which is then used to information Imagen 3 in making a remix of the picture. You too can enter textual content prompts to additional outline the specified final result, together with detailed descriptions like “Topic is using a flying bike.”
As a result of Whisk solely focuses on just a few key traits from every picture, the corporate explains that the outcomes could not all the time meet your expectations. For instance, the generated topic may differ in top, weight, coiffure, or pores and skin tone. Google says you possibly can view and edit the underlying prompts at any time.
The experiment is at the moment solely obtainable to customers primarily based within the U.S. at labs.google/whisk.
