# Image-to-video versus text-to-video: which should you start with?

Both approaches generate moving pictures, but they solve different problems. Text-to-video is best when you are exploring ideas and do not yet have visuals. Image-to-video is best when you already know what the first frame should look like.

If brand or character consistency matters, start from an image. A reference frame fixes colours, proportions and composition, so the model only has to decide how things move. With text alone, every generation reinterprets the subject.

A common workflow is to use text prompts to find a look, pick the best still, and then move to image-driven clips for the final shots.

If you want to test both modes in one place, text to video AI is worth a try.