Gowthami Somepalli Highlights Atlas Text-to-Image Capabilities

Gowthami Somepalli, a multimodal AI researcher at World Labs, retweeted a post from KeunhongP stating that Atlas is a great text-to-image model. She added that under their unified formulation, text-to-image generation is just single-frame generation using the same operations. Somepalli also quoted posts showing additional generations from the model. In a separate quote she expressed personal attachment to another model, noting that lots of her love and taste went into it and that it came out wholesome. The posts form part of visible discussion among researchers sharing work on generative modeling and diffusion models.