Fal API Kling Video

Fal API Kling Video is the shortest-path way to get Kling Video 2.1 to animate your image: feed it one picture and a motion prompt, get an MP4 back. Kling is Kuaishou's closed video model - there are no weights to download, no local install, no VRAM math. It's API-only, and this node is the ComfyUI door into it.

The inputs are minimal on purpose - this is a "single image to video" node with none of the frills the bigger video nodes carry. You need api_token (FAL key, node input or FAL_KEY env var), image (the still you want to move), and prompt (the default "A cinematic video transformation" is a fine starting point - describe the motion you want, not the scene, since the scene is already your image). The node submits to fal-ai/kling-video/v2.1/standard/image-to-video, waits for the job, and hands back:

• video_path - where the MP4 landed on disk.
• generation_info - the API's metadata as JSON.
• generation_time - how long the job took, which is useful because Kling jobs can run a while and it's good to know what "normal" looks like.

That's the whole API surface. No aspect ratio, no duration, no camera controls - if you want that control, this pack's Seedance and LTX-2 nodes are the siblings that carry it.

The context worth knowing: Kling sits in the crowded API-only video tier that the KB's labs and closed-source coverage maps well - Kling, Seedance, Veo, Hailuo are all closed and API-gated, with Kling historically holding a strong reputation in the video arena rankings. The flip side, which the community says bluntly, is that closed API video is where you pay per second and have zero control over the pipeline. This node inherits all of that: it's a convenience wrapper, not an unlock.

Real-world notes. Every run costs FAL credits and takes a non-trivial chunk of wall-clock time - this is not a "hit queue and instantly see motion" node; it's a "go make tea" node. The default prompt is deliberately vague because image-to-video works best when the prompt describes the movement - camera push-in, water rippling, a subject turning - rather than restating the content of your still. And because it's a paid third-party call, the failure modes are API failures: bad key, exhausted credits, a timeout. There's no local troubleshooting to do; when it fails, it's usually the bill or the network.

Install is the pack standard: ComfyUI Manager search "Node Fun", or clone https://github.com/nofunstudio/Node_Fun_ComfyUI into custom_nodes, pip install -r requirements.txt (the fal-client dependency is what powers this), restart. For anyone who just wants "my image, moving, now" without standing up Wan or LTX locally, this is genuinely the lowest-effort route in the pack. Just don't expect to steer it much - it's a single-image, single-prompt tool, and it knows it.