Kling 3.0 Omni Text to Video

The Kling 3.0 Omni Text to Video node is the flagship text-to-video door into Kling's latest models - kling-v3-omni and the reasoning-style kling-video-o1. If you want the current generation of Kling quality straight from a text prompt, this is it. The Omni model is Kling's multi-modal headliner: character consistency, native text rendering, and audio that comes along with the picture.

Mechanically it's the standard partner-node pattern: your prompt goes to Kling's servers through your Comfy account, the job runs in the cloud, and the finished VIDEO comes back to your canvas when it's done. No local model, no GPU work on your side, billed per generation to Comfy account credits - the node shows a live price badge that factors in resolution, duration, and audio.

The inputs that matter:

• prompt - describes the video content. The tooltip's fun catch: the Omni model accepts both positive and negative descriptions in one field, so you can write "a lighthouse at dusk, no people" instead of juggling a separate negative prompt.
• model_name - kling-v3-omni or kling-video-o1. O1 is the slower, more deliberative variant; among other differences, storyboards are ignored for it.
• duration - 3 to 15 seconds, slider by default. That 15s ceiling is the current Kling max, and it's the biggest single reason to move up from the old 5/10s nodes.
• resolution - 720p, 1080p, or 4k. Optional; 1080p default.
• storyboards - the interesting one. Flip it from "disabled" to "1 storyboard" through "6 storyboards" and the node expands per-segment prompt and duration fields, so you can generate a whole multi-shot sequence in one call. Each storyboard prompt caps at 512 characters. Ignored on the o1 model.
• generate_audio - off by default here (unlike the plain Kling 3.0 node), so flip it on when you want sound.
• seed - a stub. The tooltip says it plainly: the seed only controls whether the node re-runs; results are non-deterministic regardless of what you set.

Single VIDEO output, straight into Save Video or Preview Video.

The gotcha most people hit is the storyboard sum. With multi-shot enabled, you're paying for the total generated duration, and the price badge reflects it - a 6-shot sequence is a real credit commitment, so start with one or two storyboards to taste the style before you spend on the full sequence. And note the model you pick changes the feature set: if you specifically want storyboards, don't reach for o1.

Usual partner-node housekeeping: log into your Comfy account or the node refuses to run, and if the whole API-node family clogs your node search, --disable-api-nodes hides them.