Infinity Star Text-to-Video (fal)

InfinityStarTextToVideo is one of the less-famous entries in gokayfem's ComfyUI-fal-API pack - a text-to-video node that calls the Infinity Star model hosted on fal and hands you back a URL to the finished clip. It's not a household name the way Kling or Luma are, so treat it as something to try when you want a different look than the usual suspects, rather than a default. Like everything in this pack, it runs on fal's servers: no local model, no VRAM, you pay per generation.

How it works

You give it a text prompt (plus a few controls), fal runs the generation remotely, and the node outputs a video_url string pointing at the result. Because the video comes back as a URL rather than a native ComfyUI VIDEO, you'll want to download or save it before that link expires.

The inputs and outputs that matter

• prompt - describe the scene and the motion. This is the main lever.
• negative_prompt - defaults to low quality.; add things you specifically don't want.
• aspect_ratio - 16:9, 9:6, or 1:1.
• guidance_scale (0–10, default 7.5) - how tightly it sticks to your prompt. Higher = more literal, lower = more creative drift.

Two optional controls are worth a light touch. enhance_prompt (on by default) lets the service rewrite your prompt before generating - handy for short prompts, but turn it off if you wrote something careful and want it respected verbatim. tau_video (0–2, default 0.4) is a model-specific sampling knob; leave it at the default unless you're deliberately experimenting, since there's no reliable community guidance on it. There's also a seed (default 42) for reproducibility. The single output is video_url (STRING).

How to install it

Via ComfyUI Manager: search ComfyUI-fal-API, install, restart. Manually:

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

No models download. Add a fal API key to config.ini under [API] (from config.ini.example) or export FAL_KEY=your_key, restart, and look under FAL/VideoGeneration.

Common issues & troubleshooting

• Auth error. FAL_KEY not set. Fix and restart.
• The video link is dead when you come back to it. fal result URLs expire. Feed video_url into the pack's Save-Media-from-URL utility, or download it right away.
• A run fails. The pack raises fal's actual error - usually a validation issue or a content-policy rejection. Read the text; it names the problem.
• Prompt seems ignored. Nudge guidance_scale up, and try turning enhance_prompt off if the auto-rewrite is steering somewhere you didn't ask for.
• Cost. Video generation is a paid API call and not a cheap one. The pack's result cache makes repeat identical calls free (force_rerun bypasses), and the spend guard caps a session.
• Windows Portable. Missing fal-client? python_embeded\python.exe -m pip install fal-client.

Bottom line: this is an experimentation node. If you need a dependable go-to for text-to-video, the Kling and MiniMax nodes in the same pack are the better-known picks - but Infinity Star is right there when you want to see what a different model does with the same prompt.