ComfyUI Node

Kling VIDEO 2.6 Standard

Reference-driven video, trimmed for cost

By Runware·Created 2 years ago·Updated about a month ago· 140
Kling VIDEO 2.6 Standard
  • referenceImages
  • referenceVideos
  • video
positivePrompt
numberResults1
providerSettings.klingai.characterOrientation(default)
providerSettings.klingai.keepOriginalSoundfalse
safetyfalse
safety.checkContentfalse
safety.modefast
ttlfalse
ttl_value60
outputFormatMP4
outputQuality95

Kling VIDEO 2.6 Standard is the value tier of the 2.6 generation, and it's worth reading its input list like a spec sheet - because what's missing tells you what this tier is for. It keeps the reference-driven strength of 2.6 (referenceImages, referenceVideos, character orientation) but trims the resolution knobs and the noise. No width/height, no CFGScale, no frameImages. You point it at references, you prompt it, and it renders - cheaper than Pro, and deliberately less fiddly.

Why you'd reach for it

Reference-driven video at a price you can afford to loop. The 2.6 line's strength is using what you give it - an image, a source clip - to hold character and motion together. Standard keeps that core and drops the parts most people don't touch anyway. If your workflow is "reference in, draft video out" and the drafts outnumber the finals, Standard is the tier that makes the ratio work.

How it works

The pack's generated node, shared base class: referenceImages tensors are base64'd to PNG data URIs, referenceVideos goes as a typed URL/path/UUID reference, and the request rides to the Runware REST API via the runware SDK. The clip returns as a native ComfyUI VIDEO object with cost and NSFW flag on the title bar.

Inputs that matter

  • positivePrompt - optional but expected. Describe what should happen; the model leans on it once references are set.
  • referenceImages (IMAGE) - feed stills to steer style and character.
  • referenceVideos - feed a source clip to steer motion and identity.
  • providerSettings.klingai.characterOrientation - image or video: which reference defines the character's look and facing.
  • providerSettings.klingai.keepOriginalSound - keep the reference clip's audio in the output.
  • numberResults (1–4), outputFormat (MP4/WEBM/MOV), outputQuality (95), plus the safety group.

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

  • No width/height on this node means you don't control resolution directly - the tier decides. If exact framing matters, you're on the wrong node; Pro is the one with the knobs.
  • The media sockets are different kinds: images are IMAGE tensors (Load Image first), videos are URL/path/UUID strings. Cross them and you get a confusing error.
  • keepOriginalSound only does anything when there's a reference video to keep sound from - wire one or the toggle is dead weight.
  • Closed Kuaishou model, per-run pricing, cost shown on the node.

Standard's pitch is honest: references in, drafts out, cheaper than Pro, fewer knobs to break. For the "prove the motion before you pay for polish" phase of a project, it's hard to argue with.

CategoryRunware/Video/klingai

Inputs (13)

NameTypeDefaultDescription
referenceImagesoptIMAGE
referenceVideosoptRUNWARE_REFERENCEVIDEOS
positivePromptoptSTRINGText prompt describing elements to include in the generated output.
numberResultsoptINT11–4Number of results to generate. Each result uses a different seed, producing variations of the same parameters.
providerSettings.klingai.characterOrientationoptCOMBO(default)Source for character orientation reference.
providerSettings.klingai.keepOriginalSoundoptBOOLEANfalseMaintain the original sound from the reference video.
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