Kling O1 Reference to Video
kling-o1-reference-to-video: turn references into realistic videos
A tradução deste card ainda não está disponível; parte do conteúdo aparece em inglês.
kling-o1-reference-to-video is a generative video model that takes a reference image or video clip along with a text prompt and produces a new video that preserves key visual elements from the reference while following your description. It excels at creating consistent, realistic motion and style transfer from a single source.
Use it when you have a specific look, character, or setting you want to build upon — for example, turning a product photo into a short promotional clip or animating a character concept. The model is designed for rapid iteration, making it easy to explore variations.
What makes kling-o1-reference-to-video distinctive is its ability to maintain high fidelity to the reference while generating smooth, coherent video sequences. It handles complex scenes and subtle motion, giving creators a powerful tool for storytelling and prototyping without needing advanced video editing skills.
- Animate a product photo into a short ad
- Create a video from a character concept art
- Extend a short clip with new action
- Visualize a scene from a storyboard frame
- Generate background footage from a location reference
Preserves key visual details from your input image or video.
Produces smooth, natural-looking movement and transitions.
Maintains the aesthetic and mood of your reference across the generated video.
- Use a well-lit, high-resolution reference image for best fidelity.
- Keep your text prompt simple and focused on the action you want.
- Experiment with different reference angles to see how the model adapts.
- Use the seed parameter to reproduce or vary a specific result.
- May struggle with very complex or crowded scenes.
- Generated videos are short (typically a few seconds).
- Requires a clear, high-quality reference for best results.
Close-up of a woman with flowing hair turning to camera, soft golden hour light, shallow depth of field
A sports car drifting on a mountain road, dust and smoke, dramatic lighting
A floating island with ancient ruins, clouds drifting below, epic cinematic shot
A number that sets the starting point for generation — same seed + same inputs gives the same result.
Number of refinement steps — more steps can improve quality but take longer.
Nucleus sampling: limits the pool of possible next steps to a cumulative probability — lower values make output more focused.
Length of the generated video in seconds.
Controls randomness — lower values give more predictable results, higher values more creative variety.
Shape of the video (e.g., 16:9 for widescreen, 9:16 for vertical).
How closely the video follows your text prompt — higher values stick to the prompt, lower values allow more freedom.
Os preços exibidos são o que você paga na Muitto. Sem surpresas.
What kind of reference can I use?
You can use an image or a short video clip as the reference.
How long is the generated video?
Output videos are typically a few seconds long, depending on settings.
Can I control the camera movement?
Yes, you can suggest camera motion in your text prompt.