You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/api-references/video-models/kling-ai/video-v2.6-pro-motion-control.md
+32Lines changed: 32 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -71,6 +71,30 @@ If the video generation task status is `completed`, the response will include th
71
71
72
72
## Code Example
73
73
74
+
1. Provide the URL of the image containing the character you want to animate.
75
+
2. Provide the URL of the video where another character performs the movements you want to transfer to the animated character.
76
+
3. If needed, describe minor background details or additional objects in the frame using the `prompt` parameter. Example: `"At 00:03, a brightly colored parrot flies in from the left, briefly circles above the character once, and then hurries off to the right."`
77
+
4. Set the `character_orientation` parameter to `image` or `video`, depending on whether you want to use the character’s orientation from the image reference or from the video reference.
78
+
5. By default, the model uses the audio track from the reference video. You can disable this behavior by setting the `keep_audio` parameter to `false`.
79
+
80
+
<details>
81
+
82
+
<summary> Input Reference Guidelines</summary>
83
+
84
+
* Ensure the character’s entire body and head are clearly visible and not obstructed in both the image and the motion reference.
85
+
* Keep the character’s proportions consistent between the image and the reference video. Avoid pairing a full-body motion reference with a half-body or cropped image.
86
+
* Upload a motion reference featuring a single character whenever possible.\
87
+
If the motion reference contains two or more characters, the motion of the character occupying the largest portion of the frame will be used for generation.
88
+
* Real human actions are recommended. Certain stylised characters, humanoid animals, and characters with partial humanoid body proportions are supported and can be recognised.
89
+
* Avoid cuts, camera movements, and rapid scene changes in the motion reference video.
90
+
* Avoid overly fast or complex motions. Steady, moderate movements generally produce better results.
91
+
* The short edge of the input media must be at least 300–340 px (depending on the input type), and the long edge must not exceed 3850–65536 px.
92
+
* The supported duration of the motion reference video ranges from 3 to 30 seconds.\
93
+
The generated video length will generally match the duration of the uploaded reference video.
94
+
* If the motion is too complex or fast-paced, the generated output may be shorter than the original video, as the model can only extract valid continuous action segments.
95
+
96
+
</details>
97
+
74
98
The code below creates a video generation task, then automatically polls the server every **15** seconds until it finally receives the video URL.
0 commit comments