DALL-E creates english text. DALL-E 2 creates images from text. i fed these words into the openAI system:
mix of a Rolls Royce car and a crocodile in the style of leonardo da vinci. Light from the left, through a tiny window.
… and got these four images back within a few seconds. the images are no photographs. they are pure AI imaginations. the neural network was trained with billions of images on the web and seems to understand properly how reflections and shadows work, how textures and light harmonize.
with a slight variation of the text …
morphing a Rolls Royce car into a crocodile in the style of leonardo da vinci. Light from the left, through a tiny window.
… the AI understands the word “morph“ and its semantic context correctly and delivers these four images:
here is the first image in its original resolution: