ComfyUI Node

KlingAI 2.1 Master

The flagship that fixed the details

By Runware·Created 2 years ago·Updated about a month ago· 140
KlingAI 2.1 Master
  • frameImages
  • video
positivePrompt
negativePrompt
size1080p (1:1)
CFGScale0.50
duration5
numberResults1
safetyfalse
safety.checkContentfalse
safety.modefast
ttlfalse
ttl_value60
outputFormatMP4
outputQuality95

Kling 2.1 Master is what happens when the model that was already the ceiling gets another pass. The 2.1 line tightened the things 2.0 fumbled - hands, physics, the small details that separate "impressive" from "usable" - and Master is the top tier of it. This node puts that model on your ComfyUI canvas with a single required input and a lot of cloud GPU doing the work.

Why you'd reach for it

Same pitch as 2.0 Master, sharper execution. Kling is Kuaishou's closed API-only model, so the only way it appears in a node graph at all is through a service like Runware. You get the flagship quality without owning the hardware - your positivePrompt goes up, a finished MP4 comes back as a native ComfyUI VIDEO. If 2.0 Master is the ceiling, 2.1 Master is the ceiling with the light switches fixed.

How it works

Another generated Runware node with the pack's shared base class doing the real work: prompt (plus optional frameImages, base64'd to PNG data URIs) → Runware REST API via the runware SDK → clip downloaded and wrapped as VIDEO. Each run's cost and NSFW check show up on the node title bar. Zero local rendering.

Inputs that matter

  • positivePrompt - required. Text-to-video by default.
  • frameImages (IMAGE) - optional; wire in a first frame for image-to-video.
  • duration - 5 or 10 seconds. Flagship per-frame pricing means longer hurts.
  • width / height - 1024×1024 default; Master handles high res, and you'll pay for it per frame.
  • CFGScale - 0.5 default; skip negativePrompt unless you raise guidance above 1.
  • numberResults (1–4) - separate cost per variation.
  • outputFormat (MP4/WEBM/MOV), outputQuality (95).

Output: one video socket (VIDEO type).

Install and the API key

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

Restart ComfyUI (or Runware via ComfyUI Manager). Key from the Runware dashboardSettings → Runware API key, RUNWARE_API_KEY, or runware auth login.

Gotchas

  • This is the premium tier of a premium model - treat numberResults like it costs money (it does). Iterate at 1, expand only when the prompt is locked.
  • frameImages expects a real IMAGE tensor; a filename will just error.
  • The VIDEO output type needs modern ComfyUI; old builds degrade to a saved file path.
  • Closed weights, per-request pricing, and it's the middle child: 2.1 Master is excellent, but the video 2.6 and 3.0 nodes have since pushed Kling further.

Reach for 2.1 Master when you need the flagship's polish and 2.0's flaws bug you. It's a "the details matter" kind of upgrade - and on a Master tier, the details are the product.

CategoryRunware/Video/klingai

Inputs (14)

NameTypeDefaultDescription
positivePromptSTRINGText prompt describing elements to include in the generated output.
frameImagesoptIMAGE
negativePromptoptSTRINGPrompt to guide what to exclude from generation. Ignored when guidance is disabled (CFGScale ≤ 1).
sizeoptCOMBO1080p (1:1)Output size. Only one of resolution or dimensions is sent.
CFGScaleoptFLOAT0.500–1Guidance scale representing how closely the output will resemble the prompt. Higher values produce results more aligned with the prompt.
durationoptCOMBO5Length of the generated video in seconds. The total number of frames produced is determined by duration multiplied by the model's frame rate (fps).
numberResultsoptINT11–4Number of results to generate. Each result uses a different seed, producing variations of the same parameters.
safetyoptBOOLEANfalseEnable to set safety. Off uses the model's default.
safety.checkContentoptBOOLEANfalseEnable or disable content safety checking.
safety.modeoptCOMBOfastSafety checking mode for video generation.
ttloptBOOLEANfalseEnable to set ttl. Off uses the model's default.
ttl_valueoptINT60Time-to-live (TTL) in seconds for generated content. Only applies when `outputType` is `URL`.
outputFormatoptCOMBOMP4File format for the generated video.
outputQualityoptINT9520–99Compression quality of the output. Higher values preserve quality but increase file size.

Outputs (1)

NameTypeDescription
videoVIDEO