Kling AI VideoCreate videos from a text prompt, animate a start image, or combine image and video references with Kling 3.0 Omni. Direct the action, camera and sound in one workspace.
Create from text, animate an image or direct a scene with visual references.
Kling 3.0 Omni is Kuaishou’s multimodal AI video model for creating and editing video with text, image and video references. Visual inputs help guide who appears, what the scene looks like and how the action unfolds, while the prompt describes the performance and camera direction.
Use Text to Video for a written idea, Image to Video for a start frame, or Reference when several images or an existing clip should guide the result. You can also compare the Kling 3.0 generator for text and start-frame creation. Motion Control focuses on transferring a specific performance to a character image.
Character, product and location images provide a visual starting point for a new scene. The prompt adds movement, interactions and camera direction.
Existing footage provides context for a new version of a scene. Text instructions describe the changes and identify details to preserve, such as the subject, setting or camera movement.
Guide characters, environments and creative edits with your own visual references.
Give the model a clear view of your subject before you describe the performance. Reference images help anchor appearance while your prompt supplies the action and framing. For a recurring character, reuse the same clean references and check the face, costume and silhouette in each result.
Combine a character or object reference with a location image to make the intended relationship clear. Describe where the subject stands, what it touches and how the camera moves. A simple action gives you room to evaluate scale, lighting and contact before attempting a longer sequence.
Start with a video when its movement or composition is already useful. Describe the change you want, then name what should stay: the camera path, action timing, subject position or background. Keeping the request specific makes it easier to judge whether the edit followed your intention.
Use a source clip to explore an illustrated or animated treatment. Specify the line work, materials, lighting and color palette instead of listing several competing styles. Review the result for flicker and changing details, especially where hands, hair and small objects move.
Text, image and reference inputs, with the output settings available in this workspace.
Kling 3.0 Omni
Create from text, a start image, or image and video references with Kling O3.
Images + video
Use up to 7 images, or 1 video with up to 4 images.
3–15 seconds
Choose a clip length in the editor. Source footage and the operation can impose additional limits.
720p · 1080p
Select the output resolution before generating.
Generate or retain
Generate sound from text or image inputs. When Reference includes a video, choose whether to keep its original sound.
Auto or custom
Plan multiple shots with a combined duration of up to 15 seconds.
16:9 · 9:16 · 1:1
Choose these ratios for text and reference generation. Image-to-video follows the uploaded image’s framing.
100 MB per file
Keep each image or video reference within the upload limit.
Name what changes
Describe the edit and identify the visual details that should stay.
Six ways to turn your references into a new video.
Start from a product, character or location reference, then describe the action and camera movement.
Use product references to plan a reveal, a close-up or a new campaign setting while keeping the item recognizable.
Reuse character references across short scenes, with a consistent face, wardrobe and visual direction.
Pair a location reference with clear action and camera direction to explore the next moment in a story.
Create portrait-format ideas around a product, outfit or recurring subject, then adapt the framing for each channel.
Break an idea into a short visual sequence with a clear subject, a simple action and carefully planned shots.
Bring a character design and an environment reference together to explore movement, scale and scene composition.
Concept images illustrating the workflows above.
Use existing visual material to give the next scene a clear starting point.
Plan short scenes around the same character references. Keep a consistent wardrobe and lighting brief when continuity matters.
Explore how a designed character moves through a location before committing to a finished animation or cutscene.
Use a clip as the starting point for a new treatment. Specify the change while keeping useful camera movement, action timing and composition.
Try a different scene treatment or visual style while using an existing clip to guide the movement and composition.
Start with the inputs you already have and the task you want to complete.
Review identity, movement and sound before using a take in your project.
Faces, clothing, logos and fine details can drift. Review the result at normal speed and check key frames before using it in a finished project.
Several characters, complex contact and fast camera moves make results harder to judge. Begin with one subject and a short action, then add complexity.
The duration selector describes the requested output. Source clips and the selected operation may introduce further requirements or limits.
Listen for timing, clarity and changes to the source track. A successful visual edit does not establish that the audio is ready to publish.
Choose a creation mode, describe the scene and review the finished take.
Select Text to Video for a written idea, Image to Video to animate a start image, or Reference to combine up to seven images or one video with up to four images. Each uploaded file can be up to 100 MB.
Describe the action or edit, then choose resolution, duration and sound. Set framing for text or reference generation; image-to-video follows the start image. Keep the combined shot duration within 15 seconds.
Check the credit estimate before generating. Inspect identity, movement, contact and sound in the result, then download the video or refine one instruction for another take.
Inputs, settings and what to check before you generate.
Kling 3.0 Omni is a multimodal video model focused on reference-based creation and editing. It uses visual references alongside text direction to guide subjects, scenes and shot sequences. This workspace supports Kling O3 text-to-video, image-to-video and reference-to-video generation.
Both workspaces can create clips from text or a start image. Omni also provides a reference workflow for combining character, product and scene images or using existing footage to guide an edit. Choose the mode that matches your starting material.
This editor accepts image and video files. You can use up to seven images, or one video plus up to four images, with a 100 MB limit per file. Use images that clearly show the subject rather than several nearly identical or conflicting views.
All three creation modes offer 720p and 1080p output with a selectable duration from 3 to 15 seconds. Choose 16:9, 9:16 or 1:1 for text and reference generation; image-to-video follows the uploaded image’s framing. Source footage and the chosen operation can impose additional limits.
Use Sound in Text to Video, Image to Video or Reference mode with image inputs to request generated audio. When Reference includes a video, choose Keep original sound instead. Listen to the finished result to check timing and clarity.
Yes. Enable Multi-Shot for automatic shot planning, or use Custom Multi-Shot to write individual shot descriptions and durations. The editor limits the combined duration to 15 seconds. Give each shot a clear action and avoid asking for too many events at once.
No. References guide the generation, but details can change as the subject turns, moves quickly or becomes obscured. Inspect faces, hands, labels and object boundaries. Simplify the action or improve the references if a detail drifts.
The Generate video button displays the estimated credit requirement for the current settings. Check that estimate before submitting. You can review available credit packages on the Pricing page.
Create with Kling 3.0 Omni.
Start with text, an image or visual references. Describe a clear action and make your first Kling 3.0 Omni video.