MarKI

Tell MarKI which image it should generate for you and it will do it.

Just describe a scene, no matter how crazy it may be. MarKI then generates the right image for you instantly. MarKI now uses a diffusion model, which performs its calculations on 128 super-fast A100 graphics cards in less than a second. Use the smartphone microphone to describe your desired image with your spoken words or use the text input field. At the moment MarKI understands english and german.