KlingAI 1.5 Standard

Kling 1.5 Standard is the discount aisle of a genuinely good image-to-video model. Same architecture as the Pro node, same required-input shape, but a lower resolution ceiling and a smaller credit bill per clip. If your workflow is "generate a frame, animate it, keep it moving through the pipeline" rather than "this clip is the final deliverable," Standard is the version you should actually leave in your graph.

What's different from Pro

The workflow is identical - both 1.5 nodes demand a frameImages tensor plus a positivePrompt - so this isn't a different node, it's a different price tier of the same one. Standard resolves at a 720p-class ceiling where Pro gets the 1080p-class look, and the per-run cost drops accordingly. For anything that's going to be a motion test, a storyboard beat, or a segment that gets re-rendered anyway, Standard is where you'd rather burn credits. Upgrade to Pro only for the shots that matter.

How it works

Generated Runware node, shared pack logic. Your frameImages IMAGE tensor gets base64'd to PNG data URIs, bundled with the prompt, and POSTed to the Runware REST API over the runware SDK. The finished clip is downloaded and returned as a native VIDEO object. No local GPU, no model files, no per-model setup - that last part is the pack's whole reason for existing.

Inputs that matter

• frameImages (IMAGE) - required. Your first frame from a Load Image node or any IMAGE output.
• positivePrompt - required. Describe the motion; the image already tells the model what it looks like.
• duration - 5 or 10 seconds. Longer = more frames = more credits.
• 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) - a cheap-ish way to roll the dice on motion, though each variation still costs.
• outputFormat (MP4/WEBM/MOV) and outputQuality (95).

Output is one video socket (VIDEO type) into a preview/save node.

Install and the API key

One install covers every node in the pack:

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

Restart ComfyUI (or use ComfyUI Manager → search Runware). Add a key from the Runware dashboard via Settings → Runware API key, the RUNWARE_API_KEY env var, or runware auth login.

Gotchas

• frameImages wants a real IMAGE tensor - a file path or a URL string will just fail.
• Kling is closed, Kuaishou's, and API-only. You're renting inference, and the node's title bar will tell you exactly what each run cost.
• This is an old generation now. The motion quality of 1.5 Standard is fine for drafts, but the 1.6, 2.x, and 3.0-era Kling nodes are meaningfully better if your budget allows.

Think of it as the workhorse tier: unglamorous, cheaper, and perfectly adequate for everything that isn't the money shot.