Kling 2.6 Motion ControlMotion transfer onto an image
Upload a character photo and a motion video — the AI creates a new video where your character performs those motions with high precision.
How it works
1. Upload a photo
A character image that needs motion added
2. Upload a video
Reference video with the movements to transfer
3. Get the result
AI will create a video of your character performing the movements
Sample works
Videos generated with Kling 2.6 Motion Control
Dance
Hand motions
Gestures
Walking
Turning
Sport
Detailed guide
Upload a character image
Ideal photo:
- • Clear photo of the person full-body or waist-up
- • Good lighting, no blur
- • Neutral pose (hands down or slightly bent)
- • Minimum 720×720 pixels
- • Format: JPG, PNG, WebP (up to 10 MB)
⚠️ Avoid:
- • Cropped body parts (especially hands/legs)
- • Complex poses with overlap
- • Small faces or distant shots
- • Photo with multiple people
Upload a reference video with movements
Ideal video:
- • One person in the frame with clear movement
- • Duration 3-30 seconds
- • Minimum 720×720 pixels
- • Stable camera (no shake)
- • Format: MP4, MOV, MKV (up to 100 MB)
💡 Good motion examples:
- • Dancing and dance combos
- • Hand gestures and waving
- • Walking, turning, squats
- • Sports exercises
Adjust the generation parameters
- • 720p — faster and cheaper, good for tests
- • 1080p — higher quality, for the final result
Character orientation:
• Video (up to 30s)
The character will be oriented as in the reference video.Camera composition, angle, and direction of motion are preserved from the reference.Best for complex or long animations.
✓ Exact composition match from video + motion up to 30s
• Image (up to 10s)
The character will be oriented as in the source photo.The angle and pose from the image are preserved (for example, face to camera, profile).The motion from the video adapts to the character's pose in the photo.
✓ Preserves angle/pose from photo + motion up to 10s
💡 Doesn't control the movements themselves (they come from the reference) but defines the character's composition and angle
Choose how many seconds of the reference video to use. Cost = duration × price per second.
AI-assisted prompt enhancement
The prompt is optional, but can significantly improve the result.Our service offers AI prompt enhancement —the AI will turn your idea into a detailed scene description.
How to use:
- 1. Write a short idea in your language
- 2. Pick a model for enhancement (ChatGPT, Gemini Pro or Grok)
- 3. Click "Enhance" — the AI builds a detailed prompt
- 4. Edit the result if needed
💡 Prompt idea examples:
- • "on a sunset beach, cinematic lighting, warm tones"
- • "in a futuristic interior with neon lights, cyberpunk style"
- • "in a professional studio with softboxes, 4K quality"
- • "on a city street in the evening, shop window lights, realistic style"
- • "anime style, vibrant colors, cartoony animation"
💡 Remember: motion comes from the video, the prompt describes environment, style and scene atmosphere
How to get the best results
Follow these recommendations for a stable, high-quality result
1. Keep a single shot framing
Image and reference video should match in framing for a stable result:
✓ Correct:
• Waist-up photo + waist-up video
• Full-body photo + full-body video
• Same angle (front, profile)
✗ Incorrect:
• Waist-up photo + full-body video
• Close-up + wide shot
• Different angles (front + profile)
💡 Orientation tip:
• Mode "Video" — works best when photo and video share a similar composition (both full-body, both waist-up)
• Mode "Image" — use this if you want to preserve a specific angle from the photo, even if it differs from the video
⚠️ Mismatched framing can cause motion instability or cropped actions
2. Pick smooth, clear motion
The reference video should contain natural movements at a moderate pace.
✓ Good:
• Smooth dance moves
• Calm walking or gesturing
• Continuous dynamics without jerks
✗ Avoid:
• Actions that are too fast
• Sudden jumps or moves
• Cuts inside the video
💡 The smoother the motion, the more stable and higher-quality the result
3. Leave room for movement
If the reference contains wide gestures or full-body movements, the character in the photo should have visual space around them.
⚠️ Important:
Tight framing (when the character almost fills the frame) limits the range of actions and can hurt stability. Leave at least 10-15% space around the frame edges.
4. Character image requirements
Body and head are visible fully visible and unobstructed other objects
The character takes a significant part of the frame (40-70% of the frame)
Proportions match the reference video
Supported: realistic people, anthropomorphic animals, characters with human-like proportions
The API is optimized for a single main character. Best results are achieved when the character clearly stands out from the background.
5. Reference video requirements
One character in the frame (if there are several, the movements of the one taking more space are used)
We strongly recommend videos of real people (not 3D animation)
Duration 3-30 seconds (overly complex actions may be auto-trimmed)
Stable camera, no sudden moves, zoom or shake
⚠️ Avoid:
- • Cuts between frames
- • Sudden camera moves or frequent zoom
- • Fast or chaotic character actions
- • Videos with multiple people in active motion
Key rule
The better the matching between the image and the reference video (framing, proportions, quality),and the more smoother motion in the reference — the more stable and higher quality the result will be.
Cost
720p
Standard quality
10 sec = — cr.
30 sec = — cr.
1080p
High quality
10 sec = — cr.
30 sec = — cr.
File requirements
Character image
- Minimum 720×720 pixels
- Format: JPG, PNG, WebP
- Up to 10 MB
- Clear photo of a person
Reference video
- Duration 3-30 seconds
- Minimum 720×720 pixels
- Format: MP4, MOV, MKV
- Up to 100 MB
Frequently asked questions
What is Kling 2.6 Motion Control?▼
Kling 2.6 Motion Control is a technology that transfers motion from a reference video onto a static image.
You upload a character photo and motion video, and AI creates a new video where your character performs these movements.
How is the generation cost calculated?▼
Cost depends on the of the reference video duration and resolutions:
- • 720p: — cr. per second
- • 1080p: — cr. per second
For example, a 10-second video in 720p costs — credits.
What are the character image requirements?▼
Image requirements:
- • Minimum 720×720 pixels
- • Format: JPG, PNG, WebP
- • Size up to 10 MB
Use a clear full-body or waist-up photo of a person.
What are the reference video requirements?▼
Video requirements:
- • Duration: 3-30 seconds
- • Minimum 720×720 pixels
- • Format: MP4, MOV, MKV
- • Size up to 100 MB
The video should contain clear human motion.
Что такое ориентация "image" и "video"?▼
Character orientation defines the composition and angle of the character:
🔹 Video (up to 30s)
The character will be oriented as in the reference video.The system preserves camera composition, motion, and character direction from the reference.
✅ Exact replication of the video composition
✅ Better for complex/long animations
✅ Up to ~30 seconds
🔹 Image (up to 10s)
The character will be oriented as in the source photo.The specific angle or pose from the image is preserved (face to camera, profile, etc.).
✅ Preserves angle/pose from the photo
✅ Motion adapts to the pose
✅ Up to ~10 seconds
📌 Important to understand:
Orientation Does NOT control motion (those are taken from the reference),and only defines the angle, perspective and composition of the character in the final video.
Is a text prompt required?▼
The prompt is optional, but recommended.
It helps the AI better understand the desired result and add details to the scene. Maximum of 2500 chars.