This ComfyUI workflow turns a single text prompt into a short video using the Flux3ImageToVideoNode, then encodes the result to a standard video file with SaveVideo. By default, the template is set up to render at 720p and 24 fps, but you can change width/height and fps to match vertical (9:16), square (1:1), or cinematic (21:9) formats. The workflow can run prompt-only for pure text-to-video, or optionally take in a starting frame through LoadImage to guide composition, subject placement, or style from the first frame.

Technically, Flux3ImageToVideoNode expands your prompt into a sequence of frames with temporally consistent motion. Core controls you’ll use include the prompt text, seed (for repeatability), width/height (aspect and resolution), number of frames (duration), and fps. If you provide a reference image via LoadImage, the node can use it as an initialization frame to better lock framing and identity; the influence of this can be tuned inside the Flux3ImageToVideoNode. Once the frames are generated, SaveVideo assembles them into a single clip. Note that this workflow outputs silent video; if you need audio, add it later in a video editor.