Kling AI 影片生成

Kling is Kuaishou's video model family and one of the strongest options on FlyAIgh for reference-driven generation — clips guided by one or more input images rather than pure text. Three variants are available.

Kling V4 is the newest. It renders 3 to 30 seconds in a single pass at 720p or 1080p, always with stereo sound, and is built for larger motion, complex camera moves, steadier emotion on faces and legible on-screen text. It covers text-to-video, image-to-video with an optional last frame, and a reference mode that takes up to 10 images and up to 5 reference videos (30 seconds in total) to lock a look, a movement, a camera move or a rhythm. It costs 12.7 credits a second at 720p (64 for 5 seconds) and about 19.4 at 1080p. With a reference video attached the supplier bills differently — the output length plus the longer of the reference and the output, at 7.4 credits a second at 720p or 11.3 at 1080p — and the generator shows the exact total first.

Kling V3 is the classic variant: text-to-video, image-to-video (one first frame), and first/last frame mode (two reference images), at 8 credits per second (40 for a 5-second clip). Kling V3 Omni has the same price but accepts up to 7 reference images with semantic roles inside the prompt, and adds a 4K tier. Both V3 variants generate 3–15 second clips in 16:9, 9:16 or 1:1 at 720p or 1080p, with sound as a switch that does not change the price.

Kling is particularly strong at character motion (walks, gestures, facial expression) and reference fidelity. Pick Kling V4 for long single takes, reference videos or richer sound; pick V3 Omni for 4K or many reference images on a budget; pick V3 for the lowest Kling price.