token-verse.github.io

TokenVerse

Combine personalized visual concepts in generated images

TokenVerse is a research method that learns objects, accessories, poses, lighting and materials from images and captions, then combines these concepts in new images using a pretrained text-to-image diffusion model.

Tags

personalization, text-to-image, diffusion, multi-concept, research, storytelling

Checked pages: /

Related guides: Best AI tools for image generation