Kling Video Effects

Sometimes you don't want to describe a video from scratch - you want Kling to run a known, packaged motion: a hug, a kiss, a heart drawn in the air. That's the entire genre of KlingDirect_VideoEffects: you hand it one or two images and a effect_scene name like hug, and Kling returns a short animated clip of the action, not the prompt. It's the API-flavored version of the effect templates you see in Kling's own app, exposed as a node.

The design is deliberately dumb-simple, and that's its strength. No prompt field, no camera control, no negative prompt - the effect is the instruction. You pick the scene, the duration, and which images participate, and Kling handles the choreography. For "two characters hug" or "person makes a heart with their hands," this gets you a working result in one call where hand-prompting a generic video model would take several attempts.

What you set

• image_1 (IMAGE) - required, your first subject.
• image_2 (IMAGE) - optional, the second subject. Effects like hug and kiss want both; single-subject effects just need image_1.
• effect_scene - the action, as a string. hug is the default, but the pack's Effect Templates node exists precisely because the list is server-maintained and grows: run it to fetch the current catalog as JSON rather than guessing names.
• model_name - kling-v1 (default) or kling-v1-5.
• duration - 5 or 10 seconds.
• mode - pro or std, default std. For a fun-effect node, std is usually the right call - you're paying per render and these are meant to be quick.

Outputs

The standard video quintet: video (frames as IMAGE), video_file (downloaded path), audio (Kling can add sound to the clip), url (hosted result), and task_id for chaining into Task Status or Video Extend.

Install and what it needs

Part of ComfyUI-Kling-Direct: ComfyUI Manager → search "Kling Direct" → install, or git clone https://github.com/IxMxAMAR/ComfyUI-Kling-Direct into ComfyUI/custom_nodes, then restart. No model downloads; needs your Kling access key + secret key from https://app.klingai.com/global/dev (KYC required) via the Kling AI Authentication node or env vars, with the right region. Same shared setup as every node here.

Where people get burned

• Guessing effect names. The tooltip says "e.g., hug, kiss, heart," and the catalog isn't frozen - new effects land server-side. If a name 404s, run Effect Templates to see what's actually valid before you blame the node.
• Two-image effects with one image. Feed a hug or kiss scene only image_1 and you may get a one-sided render or a failure. Add image_2 when the effect is inherently two-person.
• Forgetting it's a paid render. Each effect is a metered video job. The std mode is there precisely so you can preview cheaply - use it, then decide if the 10-second pro version is worth the credits.

If you find yourself wanting custom choreography rather than templates, you've outgrown this node - Motion Control or plain Text to Video is your next step up. For the packaged stuff, this is the one.