Frame to frame, fully controlled

First and last frame video is a mode in AI Video Pro where you decide how a clip begins and how it ends. You upload two images, and the model generates the motion that leads from the first to the last. It is the most controlled way to make an AI clip.

Models with first and last frame

Veo 3.1 has a dedicated First Last Frame To Video mode, plus faster Fast and Lite variants. It makes 8-second clips in 720p or 1080p, as 16:9 or 9:16, with sound on or off. Kling v3 Image to Video takes a start frame and an optional end frame, with lengths from 3 to 15 seconds and a 1:1 format. Seedance 2.0 and 2.5 also accept an optional end frame.

When to use it

Use it when the last shot matters: a product turning to face the camera, a closed box that ends open, a room before and after styling, a logo reveal at the end. It also helps when you join several clips, because the last frame of one clip can be the first frame of the next.

Ideas for product clips

A skincare brand can start on a closed jar and end on the open jar with the cream visible. For a bakery, the clip goes from raw dough to a golden loaf. Furniture shops move from an empty corner to the same corner with a new sofa, and a watch store can start on the strap and finish on the dial facing the camera. Every one of these needs two matching images and one short move.

Preparing the two frames

Both images should have the same format, the same light and the same product at the same size, so the model only has to invent the motion. Change one or two things between the frames, such as the angle or the lid, not everything at once. Create matching frames with AI Image Editor or AI Photoshoot.

Writing the prompt for first and last frame video

Describe the path, not the frames. „Slow 180-degree orbit around the bottle, steady speed“ tells the model how to travel. If the result jumps or warps, make the two frames closer to each other or shorten the move. For a free camera move from one photo, use image to video.