I want to generate visuals from references as well as from text prompts.

This feature lets you generate visuals using either text prompts, reference images, or a combination of both. It helps you maintain consistency with an existing style or subject while still allowing creative control through prompt instructions.

This feature supports visual generation from two main inputs: text prompts and reference materials. You can start with a written prompt to describe the scene, style, mood, or composition you want. You can also provide reference images to guide the output toward a specific subject, look, or aesthetic. Combining references with text prompts enables you to preserve key visual traits while still directing changes such as lighting, background, pose, or art style. The workflow is designed to be flexible, allowing you to iterate quickly by adjusting prompt wording or swapping references. This is useful for creating variations of a concept while staying on-brand or on-model. It can support tasks like concept exploration, marketing creative, character or product consistency, and style matching. By grounding generations in references, you can reduce randomness and get outputs closer to a desired target. By layering prompt instructions on top, you can explore multiple creative directions without starting from scratch. Overall, it helps you move from inspiration to usable visuals faster with more predictable results.