01
Text-to-video generation on both Kling 2.5 Turbo and Kling 3.0
02
Image-to-video conditioning with a start frame, end frame, or both
03
Video editing and motion-control modes available on Kling 3.0
04
Motion-control variant transfers movement from a reference video to any character image
05
Over 2 million runs completed for Kling 2.5 Turbo Pro on Replicate alone, confirming production-scale reliability
06
5-second or 10-second clips on Kling 2.5 Turbo
07
Up to 15 seconds per generation on Kling 3.0
08
Up to 1080p on Kling 2.5 Turbo Pro
09
Up to 4K resolution on Kling 3.0
10
16:9, 9:16, and 1:1 supported
11
Note: when a start image is provided, image dimensions override the aspect ratio setting
12
Native dialogue generation with lip sync
13
Sound effects and ambient audio in a single generation pass
14
Audio quality is strongest in English and Chinese
15
Text-timing engine parses multi-step, causal instructions for coherent pacing
16
Multi-shot mode accepts a JSON array of up to 6 shot definitions, each with its own prompt and duration
17
Color, lighting, brushwork, and mood held consistent through complex motion