All news

ImageGPT v3 — a new image generation model

image-gen
  • Creates complex layouts from a single prompt, including infographics, UI concepts, posters, slides, product cards, and information-dense diagrams.
  • Renders small and multilingual text, including Russian, better than many other generators.
  • Delivers highly realistic lighting, materials, reflections, and skin in portraits. These images work well as source frames for further animation in Seedance.
  • Follows prompts more accurately while preserving the structure and aesthetics of complex scenes.