FLUX 3 Video: Image to Video
Turn a single image into a full video with Flux.3. Describe the motion in a text prompt, and the model animates your image into a complete scene with synchronized audio. Outputs 1 video in 720p at 24fps, 5 to 20 seconds long, supporting any aspect ratio from 9:16 to 21:9 with built-in sound design. Ideal for quick social media clips, ad mockups, and concept films requiring realistic motion and cinematic style.
例を試す
このワークフローについて
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.
よくある質問
No. With SaveVideo, the workflow exports a silent video. Add music, dialogue, or sound design later in a video editor or a separate audio pipeline.
Yes. Load a reference image with LoadImage and connect it to Flux3ImageToVideoNode. Then adjust the node’s image influence/strength (if available) so the model follows your frame without overconstraining motion.
Set fps and the number of frames. Duration ≈ frames ÷ fps (e.g., 240 frames at 24 fps ≈ 10 s). SaveVideo will encode at the fps you specify.
Any aspect ratio is supported by setting width/height. Taller or wider frames and higher resolutions use more memory; start at 720p, keep batch size to 1, and increase resolution gradually.