The Create Video node takes an input - a prompt, an image, or both - and generates a video based on that input. The specific type of video created - its duration, aspect ratio, and content - depends on what inputs are fed into it and the choices made when setting up the node. Here is a breakdown of the node's configuration parameters:
The specific model you are using to generate the video. The model you choose has a significant impact on the options available to you within the node, from what durations you can choose to the aspect ratios available to you. As of January 2026, we recommend Kling 2.5 Turbo Pro and Veo 3.1 as the best all-purpose video models.
Most video models support a first frame image for their generations. Best practices are to send in a first frame in the same aspect ratio as the video you are creating, otherwise you may experience cropping or squishing of the video.
Some models also support a last frame, which can help with keeping videos consistent and smoothly transitioning from one model to another.
Not all models support last frame, so please check each model individually before using the last frame input in a workflow. Veo 3.1 and Kling 2.5 Turbo Pro both support first and last frame. You may check each video model using the inference form beforehand, or view the model details.
Please note that while not all models support last frame, you will always be able to connect an image to the last frame input of a Create Video node. If an input is connected and the video model does not support the last frame input, it will be ignored.
The model's supported durations are shown here. Some models, like Veo 3.1, only support a few durations (4s, 6s, and 8s). Other models like Seedance Pro allow any amount of seconds from 2 to 12.
A handful of models support audio generation, primarily Veo 3.1. If you are generating audio alongside a video, make certain you account for the audio in your prompt inputs; Otherwise, your audio may not match your video content.
The aspect ratio and resolution of the video are set here. Not all video models support the same aspect ratios and resolutions, so you will want to check that the video model you're using outputs in the aspect ratio you need.
Often, you will want to have multiple video aspect ratios output in a single workflow. Your video nodes may be combined with composition nodes earlier in the workflow to create flows of different aspect ratios, but you will need a different video node per aspect ratio or resolution.