Text-to-video vs. image-to-video: which should you use?

Start from a blank prompt when you don't have (or don't need) a specific starting image — brainstorming concepts, generating b-roll, or exploring a scene you're describing from scratch. It's the most flexible option and the fastest way to test an idea, since there's nothing to prepare beforehand.

Reach for image-to-video when you already have the exact look you want — a product photo, a character portrait, a specific piece of art, or a frame from an existing shoot — and you want it to move rather than be reinvented. The model treats your image as the starting frame and animates from there, which keeps faces, products, and layouts consistent in a way a text prompt alone can't guarantee.

On Seedance 2.0 and 2.5, Vidoro also supports handing the model a reference video alongside your prompt — useful when you want to carry over a specific motion, style, or continuation of an existing clip rather than starting cold.

If brand consistency, a specific face, or an exact product shot matters, start from an image. If you're exploring an idea or don't have a source image, start from text. Either way, you can always refine the result afterward using Vidoro's edit/continue flow instead of regenerating from scratch.