instruction-based generation

Create, edit, and transform visuals from both text and image prompts — powered by DreamOmni2 multimodal.