Keep one subject
Several characters doing several things multiplies the ways a clip can go wrong. Build a busy scene from separate, simple clips.
BestAIVideo
Kling 4.0 is the newest line from Kling AI. This page is built for the two ways people actually start: a photo that should move, and an idea that only exists as words. Pick one, add a motion prompt, decide on sound and resolution, and the credit cost appears before you generate.
The generator keeps the choices that change the result and hides the rest. Everything below is a control you can use on this page.
Upload a first frame to anchor the look of the video. Add a last frame when the clip must end on a particular image, such as a product at rest.
Describe the scene, the action and the camera in plain sentences. Text mode lets you choose the shape of the frame for the platform you are publishing to.
Keep audio on when speech, effects or ambience matter, and turn it off for silent loops or when you will score the clip in an editor.
Test at a lower resolution, finish at a higher one, and read the credit estimate update with every change before you submit.
Good video prompts read like a director's note: who or what is in the frame, what they do, and how the camera sees it. These steps take you from a rough idea to a result you can judge.
Pick a first frame if the look is already fixed, or text if you are still exploring. A picture constrains the result; words leave it open.
When you have a frame, describe the movement, the camera and the sound. The image already carries the subject and the palette, so repeating them wastes prompt.
Wide, close-up, tracking, overhead and over-the-shoulder are shot terms that translate directly into framing. Pick one per clip.
Adjust the camera, or the action, or the sound, never all three. A single changed variable tells you which edit made the difference.
A dragon skims dark water toward a harbour town under a smoky storm sky, then dives among explosions near a church spire. It is a lesson in writing large-scale action as a clear path: where it starts, what it crosses and where it lands.
A figure in a wide-brimmed hat stands on stone steps in fog, throws a red-tasselled dart, and the shot cuts to a profile close-up. It shows how atmosphere and a single decisive gesture can carry a short scene.
In a blue forest a robed figure draws glowing red threads between her hands while bronze coins spin in the air around her. Stylised fantasy like this depends on naming the colours and the one repeating motion you want.
Several characters doing several things multiplies the ways a clip can go wrong. Build a busy scene from separate, simple clips.
When a first frame is attached, resist re-describing it. Use the prompt to say what moves and what the camera does.
Record the prompt, the starting frames, the sound setting and the resolution of any clip you like, so it can be reproduced and varied later.
Animate a packshot and finish on the hero angle with a last frame.
Bring an illustration or a portrait to life with a small, believable movement.
Use a first and last frame to show a transformation, a renovation or a change of season.
Prototype a dramatic beat with weather, camera and sound before committing to a larger production.
Create the same idea in landscape, square or vertical from text mode.
Keep audio on and write the line in quotes to try a short exchange or a voiced caption.
You can make a video from a written prompt, or from a first frame with an optional last frame. You choose resolution, decide whether audio is generated and read the credit cost before you submit.
Start from a photo when the look is already settled, such as a product or a character design. Start from text when you are still exploring ideas and want the model to propose the visuals.
It tells the clip where to end. Use it for a product resting in its final pose, a transformation or any shot that has to arrive at a known picture.
State who or what moves, how they move and how the camera follows. For example a slow tracking shot as a figure walks away, rather than a list of adjectives.
Turn it on when speech, effects or ambience are part of the idea, and off for silent loops or clips you will score separately. Mention the sound you want in the prompt.
The cost depends on length, resolution and whether audio is on. The generator displays the exact credit figure and updates it whenever you change a setting.
In text mode, yes: landscape, square and vertical are available. When you start from a frame, the video follows the shape of your image.
Say who is speaking, give the tone and put the line in quotation marks. Short lines are far more reliable than long speeches.
Because it makes cause and effect visible. If you alter the camera, the action and the sound together, you will not know which change improved the clip.
Check the subject and the main action first, then camera movement and continuity. Leave sound for last, and replay the clip with the picture hidden to listen.
Choose a photo or a prompt, set the sound and resolution, and see the credit cost before you generate.