Seedance Image-to-Video (fal)

Seedance is ByteDance's video model, and it's one of the more talked-about hosted options going - it's built a reputation for coherent motion and, notably, controllable camera behavior. This node animates a still image through Seedance from inside ComfyUI. Like everything ByteDance actually sells, Seedance is API-only; there's no checkpoint to download, so a bridge like this is how you touch it at all.

What it actually does

Nothing runs locally. Your image and prompt go to fal.ai, fal runs Seedance, and you get a clip back - no VRAM, no downloads, billed per generation. The one input here that's worth calling out specifically is camera_fixed: flip it on and Seedance holds the camera still, animating the subject and scene without drifting or panning. That's genuinely handy, because a lot of i2v models can't resist adding a slow push-in you didn't ask for, and a locked camera is exactly what you want for a stable product shot or a talking subject.

The inputs that matter

• image (required): the starting frame.
• prompt (required): describe the motion and action in plain language.
• resolution: 480p or 720p. 720p costs more.
• duration: 5 or 10 seconds.
• camera_fixed (default off): lock the camera for a static shot; leave it off to let Seedance move the camera as it sees fit.

There's also an optional seed (-1 for random, a fixed number to reproduce). The output is a single STRING, the video URL - download it with the pack's Fal Save Media from URL node.

Installing it

Easiest: ComfyUI Manager → search ComfyUI-fal-API → install → restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/gokayfem/ComfyUI-fal-API.git
cd ComfyUI-fal-API
pip install -r requirements.txt

Restart ComfyUI. No model file; the dependencies are fal-client and a fal API key (in config.ini under [API], or the FAL_KEY env var). No key, no video.

Where people get burned

The most common confusion here is the camera behavior - if your output keeps drifting or panning when you wanted a locked shot, you left camera_fixed off. Conversely, if your "cinematic dolly" prompt produces a stiff, static clip, you left it on. It's a single toggle but it changes the feel of the whole shot, so set it deliberately.

Otherwise it's the standard API drill: lock a seed while you dial in the prompt so you're not paying to re-roll randomness, read fal's real error text if a run fails (content-policy refusals on people are common), and on ComfyUI Windows Portable fix an import error with .\python_embeded\python.exe -m pip install fal-client.