KlingAI 1.5 Pro

Kling 1.5 was where image-to-video stopped being a party trick. The motion got more physical, the subject stayed recognizably itself when it rotated, and for a lot of people this was the first Kling they'd trust in an actual shot. This Runware node is the Pro tier of that model, and notice the difference right away: frameImages is a required input. 1.5 is an I2V-first model - it wants your starting image, not a wish and a prayer.

Why you'd reach for it

Because you already have the image. The classic pipeline here is: generate a frame locally in ComfyUI (Flux, SDXL, whatever you like), then hand it to Kling to animate. That's exactly the workflow the open-source crowd kept complaining they couldn't replicate locally without "2–3 years and 128GB of VRAM" - and this node just closes that gap by sending your frame to Runware's cloud and returning a real clip. The Pro tier renders at a higher quality ceiling (1080p-class) than Standard, which is what you want when the video is the deliverable.

How it works

One of the auto-generated Runware nodes: the class carries the model ID and parameter plan, and everything else lives in the pack's shared base class. On queue, your frameImages tensor is base64'd into PNG data URIs and sent with the prompt to the Runware REST API via the runware SDK. The clip comes back and is wrapped as a native ComfyUI VIDEO object. Your GPU does nothing; your wallet does the lifting.

Inputs that matter

• frameImages (IMAGE) - required. This is your first frame; wire a Load Image (or any IMAGE output) in here.
• positivePrompt - required. Describe the motion and the scene, not the still image itself.
• duration - 5 or 10 seconds. Frames = duration × model fps.
• width / height - 1024×1024 default.
• CFGScale - 0.5 default; the model's default guidance behavior is what you'll mostly use here.
• numberResults - 1–4 variations, each with a different seed.
• outputFormat / outputQuality - MP4/WEBM/MOV and encoder quality (95 default).

The video output (VIDEO type) feeds straight into a preview or save node.

Install and the API key

This is the same pack as every other Runware node, so once it's installed you're done:

cd ComfyUI/custom_nodes
git clone https://github.com/Runware/ComfyUI-Runware
pip install -r ComfyUI-Runware/requirements.txt

Restart ComfyUI, or install "Runware" via ComfyUI Manager. For the key: Runware dashboard → paste into Settings → Runware API key, set RUNWARE_API_KEY, or run runware auth login once.

Gotchas

• The required input is a tensor, not a file path. People plug a filename into frameImages and get a cryptic error; Load Image first, then wire the IMAGE.
• Watch your composition: Kling will animate the whole frame. If you only want part of it moving, crop or edit before you send it - there's no mask input on this node.
• It's paid and it's closed: the model is Kuaishou's API-only Kling, and each run shows its dollar cost on the node title bar.
• 1.5 Pro is a few generations behind the current Kling. It's still good value for I2V, but if you're chasing the latest motion quality, the 2.x or video 3.0 nodes are where the new stuff lives.

For a first-frame → animated-clip pipeline that's genuinely painless, 1.5 Pro is a solid, battle-tested pick - not the newest, but the kind of dependable that's easy to forget is old.