Add controlled camera movements to text-to-video and image-to-video generation using the camera_control parameter. Supports pan, tilt, zoom, roll, and dolly. Available on v1.6+ models.
Supports: POST /v1/videos/text2video and POST /v1/videos/image2video
"camera_control": {
"type": "simple", # "simple" = one movement axis
"config": {
"horizontal": 0, # Pan: -10 (left) to 10 (right)
"vertical": 0, # Tilt: -10 (down) to 10 (up)
"zoom": 0, # Zoom: -10 (out) to 10 (in)
"roll": 0, # Roll: -10 (CCW) to 10 (CW)
"pan": 0, # Dolly: -10 (left) to 10 (right)
"tilt": 0, # Dolly: -10 (down) to 10 (up)
}
}
Rule: For type: "simple", only ONE field in config can be non-zero.
response = requests.post(f"{BASE}/videos/text2video", headers=get_headers(), json={
"model_name": "kling-v2-6",
"prompt": "A medieval castle on a cliff at sunrise, fog in the valley below",
"duration": "5",
"mode": "professional",
"camera_control": {
"type": "simple",
"config": {"horizontal": 5, "vertical": 0, "zoom": 0, "roll": 0, "pan": 0, "tilt": 0}
},
})
response = requests.post(f"{BASE}/videos/text2video", headers=get_headers(), json={
"model_name": "kling-v2-6",
"prompt": "Close-up of a tiger's face, intense eyes, jungle background",
"duration": "5",
"mode": "professional",
"camera_control": {
"type": "simple",
"config": {"horizontal": 0, "vertical": 0, "zoom": 7, "roll": 0, "pan": 0, "tilt": 0}
},
})
response = requests.post(f"{BASE}/videos/text2video", headers=get_headers(), json={
"model_name": "kling-v2-6",
"prompt": "A skyscraper, starting from the base looking up, glass and steel",
"duration": "5",
"mode": "standard",
"camera_control": {
"type": "simple",
"config": {"horizontal": 0, "vertical": 8, "zoom": 0, "roll": 0, "pan": 0, "tilt": 0}
},
})
response = requests.post(f"{BASE}/videos/text2video", headers=get_headers(), json={
"model_name": "kling-v2-master",
"prompt": "Walking through a dark forest, trees on both sides, moonlight filtering through",
"duration": "5",
"mode": "standard",
"camera_control": {
"type": "simple",
"config": {"horizontal": 0, "vertical": 0, "zoom": 0, "roll": 0, "pan": 0, "tilt": 5}
},
})
# Animate a landscape photo with a slow zoom out
response = requests.post(f"{BASE}/videos/image2video", headers=get_headers(), json={
"model_name": "kling-v2-1",
"image": "https://example.com/landscape.jpg",
"prompt": "Gentle wind, clouds moving slowly",
"duration": "5",
"camera_control": {
"type": "simple",
"config": {"horizontal": 0, "vertical": 0, "zoom": -5, "roll": 0, "pan": 0, "tilt": 0}
},
})
Important: For image-to-video, camera_control is mutually exclusive with image_tail, dynamic_masks, and static_mask.
| Shot Type | Config Field | Value | Description |
|---|---|---|---|
| Pan right | horizontal |
5-8 | Slow pan, good for landscapes |
| Pan left | horizontal |
-5 to -8 | Reveal from right to left |
| Tilt up | vertical |
5-8 | Upward reveal (buildings, trees) |
| Tilt down | vertical |
-5 to -8 | Downward motion |
| Zoom in | zoom |
5-10 | Focus on subject, dramatic |
| Zoom out | zoom |
-5 to -10 | Reveal, establishing shot |
| Dutch tilt | roll |
3-5 | Unsettling, artistic angle |
| Push in | tilt |
3-5 | Dolly forward (depth effect) |
| Pull back | tilt |
-3 to -5 | Dolly backward (reveal) |
| Value Range | Effect |
|---|---|
| 1-3 | Subtle, barely perceptible |
| 4-6 | Moderate, natural-feeling |
| 7-8 | Strong, dramatic |
| 9-10 | Maximum, may feel unnatural |
Recommendation: Start at 4-6 for most cinematic shots. Go higher for dramatic reveals or action sequences.
| Error | Cause | Fix |
|---|---|---|
400 invalid camera config |
Multiple non-zero values in simple mode | Set only one field to non-zero |
| Jerky motion | Value too high | Reduce to 4-6 range |
| No visible movement | Value too low (1-2) | Increase to 3+ |
| Mutual exclusivity error | Combined with masks/image_tail | Use only camera_control OR masks |