A start frame is a specific image provided to an AI video generation system as the first frame of the output video, anchoring the visual content, style, subject, and composition of the generated clip to that image. By providing a start frame, the creator locks the opening visual of the generated video to a known, controlled state and asks the model to generate the motion and continuation from that established starting point.

Start frame controls are a significant creative and practical tool in AI video generation workflows. They allow creators to use a carefully composed, generated, or real photograph as the precise beginning of a video clip, ensuring the subject appearance, lighting conditions, scene layout, and visual style at the start of the clip match creative intent exactly rather than relying entirely on the model's interpretation of a text prompt. Start frames are particularly useful for maintaining character consistency across multiple generated clips: by using an image of a specific character as the start frame, the generated video will open with that character's appearance preserved. When combined with an end frame control, the model is asked to generate the transition between two specified visual states, enabling more precise control over what a clip shows from beginning to end.

In workflows that chain multiple AI-generated clips into a longer sequence, using the final frame of one clip as the start frame of the next helps maintain visual continuity across the join point, reducing the discontinuities in subject appearance and scene composition that can make assembled multi-clip sequences feel disjointed. Compose on Morphic supports this kind of frame-anchored generation as part of its video production workflow.