Video Generator

Create AI-generated video clips from text descriptions or reference images

Sign in to Generate Videos

Create AI-generated videos from text or images.

Features

• Text-to-Video - Describe any scene and get a video clip
• Image-to-Video - Upload an image as a starting frame and animate it
• Multi-Shot Storyboard - Build sequences with up to 6 shots, each with its own prompt and duration
• Start and End Frames - Pin specific images as the first or last frame of your video
• Reference Images - Upload reference images and use @Element1, @Element2 in your prompt to keep characters consistent
• Native Audio - Generates synchronized audio with dialogue support (Chinese and English)
• 3-15 Second Clips - Flexible duration from quick shots to longer sequences
• Negative Prompt - Specify what you don't want in your video
• Prompt Adherence - Fine-tune how closely the model follows your description

Layout

The generator uses the same section order in all modes:

1. Mode selector
2. Scene Input (single prompt or storyboard shots)
3. Reference Frames & Characters
4. Render Controls
5. Advanced options

Tips for Best Results

• Describe the scene, subject, and motion clearly in your prompt
• Include camera directions when relevant (e.g., "slow pan", "dolly zoom", "tracking shot")
• Mention lighting, mood, and visual style (e.g., "cinematic", "golden hour", "noir")
• Keep prompts focused on a single scene or action for best coherence
• Use the negative prompt to exclude common artifacts like blur and distortion
• Lower the Prompt Adherence slider for more creative freedom, raise it for tighter control
• For image-to-video, describe the motion you want rather than repeating what's in the image
• For dialogue, write speech in your prompt naturally and Kling will generate matching voices

Multi-Shot Storyboard

The Storyboard mode lets you build a video sequence from multiple shots. Each shot gets its own prompt and duration, and Kling renders them together into a single continuous video.

You can add up to 6 shots. Two split modes are available: Custom durations lets you set each shot's length independently (3-15s each), while Auto split lets the model decide how to distribute time across your shots.

Think of each shot like a cut in a film. Describe what happens in each one: the framing, action, and camera movement. The model handles transitions between them.

Start and End Frames

Use the Reference Frames & Characters section to upload a Start Frame and End Frame.

• Start Frame is required in Image-to-Video
• Start Frame is also required if you use an End Frame or Reference Images
• End Frame is optional in every mode

Reference Images

Upload up to 4 reference images in the Reference Frames & Characters section. Each gets an @Element tag (@Element1, @Element2, etc.) that you use in your prompt or shot prompts to tell the model which character or object to place.

Use a clear frontal photo of the character with good lighting and no obstructions. The model will maintain that character's appearance throughout the video, including across multi-shot storyboards.

Reference image requirements

For Kling v3 Pro image inputs, keep every start/end/reference image at minimum 300x300, with aspect ratio between 0.40 and 2.50, and file size under 10MB. PNG and JPEG work well.

If an image is smaller than 300px on either side (for example 591x295), generation can fail even when the prompt is valid.

Example Prompts

Text-to-Video:

• "A cat stretching lazily on a sun-drenched windowsill, dust motes floating in warm afternoon light, cozy indoor atmosphere"
• "Drone flyover of a winding river through autumn forest, vibrant red and gold foliage, golden hour lighting"
• "Close-up of espresso being poured into a ceramic cup, steam rising, shallow depth of field, cafe ambiance"

Image-to-Video:

• "The character slowly turns to face the camera and smiles, gentle wind blowing through their hair"
• "Camera slowly zooms out while leaves fall from the tree, soft natural lighting"
• "The landscape transitions from day to golden sunset, clouds moving across the sky"

Storyboard (3 shots):

1. "Wide establishing shot of a misty mountain valley at dawn, camera slowly drifting forward"
2. "Medium shot of a lone hiker walking along the ridge, wind blowing their jacket"
3. "Close-up of their face as they reach the summit and smile, golden light breaking through clouds"

With Reference Images:

• Upload a photo of a person as @Element1, then prompt: "@Element1 walks through a bustling city street, camera tracking from the side, golden hour lighting"