Image-to-video
Animating an approved still image rather than generating a shot from a sentence (text-to-video).
Why it matters in production
It is the basis of all our films. We first settle an image (frame, costume, light, period), fix it until it is right, then animate it. The video model starts from what it sees: if the image is good, the shot stands a chance; if it is wrong, no animation will save it. In a prompt, we then describe the motion, never the set: it is already in the image.
Other names
i2v