How it works
Key poses, in-betweens and what the model fills
In traditional animation, key poses mark the important moments of a movement and in-betweens connect them. Kling 4.0 works in a similar way: the first frame is required and opens the clip, intermediate keyframes mark points along the way, and an optional last frame says where it ends. Up to ten frames can be used in total.
The model creates the motion between your frames, so it cannot know about anything that happens off your poses. If a character should pick up a cup between two keyframes, say so in the prompt. The frames show appearance and position; the prompt explains how one becomes the next.
Plan a sequence of poses
Neighbouring frames should look like moments of the same shot: same character design, same outfit, similar light and a consistent camera angle. Small, readable steps connect more smoothly than big jumps in location or framing.
- Three poses: an adult dancer standing still, mid-spin with her skirt flaring, then bowing. Prompt: one continuous spin that slows into a bow, stage light, soft music.
- Four poses: an adult archer drawing, aiming, releasing and lowering the bow. Prompt: a steady draw and release, the camera holds still, wind moves the cloak.
- Two poses: an adult office worker at a desk, then standing at the window. Prompt: she stands, stretches and walks to the window as rain begins.
- Five poses of an adult swordsman for a turn-and-strike, each on the same rooftop background, with the last frame showing the finished slash.
Duration, resolution and frame spacing
Duration runs from 3 to 30 seconds and resolution is 720p or 1080p. More keyframes need more time: as a starting point, allow two to three seconds for each change between frames, then adjust after a test. Squeezing many poses into a short clip tends to make the motion abrupt.
With no aspect ratio selected, the clip follows the shape of your first frame, which avoids cropping drawn key poses. You can also choose 16:9, 9:16, 1:1 or 21:9. Keep every frame at the same shape and resolution so the model does not have to reconcile different crops.
When two frames are enough
A simple move from one pose to another needs only a start and an end. Vidu Q3 Start/end frames requires exactly those two images and adds movement strength and background music settings, with clips of 1 to 16 seconds. Use Kling 4.0 when the path between start and end matters and you want keyframes to guide it. In either case, results interpret your frames; they do not guarantee smooth or identical motion.
Questions
- How many keyframes can Kling 4.0 use?
- Up to eight intermediate keyframes, plus the required first frame and an optional last frame, for no more than ten frames in total.
- Do my keyframes have to be drawn by hand?
- No. You can draw them, or create them with the character creator. They should share one design, outfit and lighting so the motion connects.
- Does the order of uploads matter?
- Yes. The clip moves through keyframes in the order you add them. Reordering uploads changes the motion.
- How long should a keyframe clip be?
- Allow roughly two to three seconds per change as a starting point, within the 3 to 30 second range, and adjust after a short test.
- Is there a simpler option with just a start and end?
- Yes. Vidu Q3 Start/end frames takes exactly two images, the first and the last, and creates the motion between them.
