gemini-omni-comfyui

Google Gemini Omni ComfyUI Nodes

ComfyUI custom nodes for Google Gemini Omni — Google’s natively multimodal any-to-any video generation model. Generate, animate, and edit AI videos directly inside ComfyUI using the muapi.ai Gemini Omni API. For REST API documentation and Python examples see Gemini Omni API

License: MIT ComfyUI Google Gemini Omni Google AI

What is Google Gemini Omni?

Google Gemini Omni is Google’s natively multimodal any-to-any video generation model, capable of producing high-quality videos from text, images, or existing video clips. Accessed via the Gemini Omni API, it supports:

These ComfyUI nodes wrap the Google Gemini Omni API so you can use the model directly inside ComfyUI workflows without writing any code.

Nodes

Node Description
🔑 Gemini Omni API Key Set your muapi.ai key once — wire to all nodes
🎬 Gemini Omni Text to Video Generate video from a text prompt via Google Gemini Omni
🎬 Gemini Omni Image to Video Animate up to 5 reference images with Gemini Omni
🎬 Gemini Omni Video Edit Restyle a video clip with Gemini Omni video editing
🎤 Gemini Omni Create Audio Profile Create a custom AI voice profile for use in generation nodes
🧑 Gemini Omni Create Character Create a character from a reference image for use in generation nodes
💾 Gemini Omni Video Saver Download video URL → disk + ComfyUI IMAGE frames

Installation

  1. Open ComfyUI Manager → Install via Git URL
  2. Paste: https://github.com/Anil-matcha/gemini-omni-comfyui
  3. Restart ComfyUI

Manual

cd ComfyUI/custom_nodes
git clone https://github.com/Anil-matcha/gemini-omni-comfyui
pip install -r gemini-omni-comfyui/requirements.txt

Quick Start

  1. Sign up at muapi.ai and go to Dashboard → API Keys → Create Key
  2. Right-click the ComfyUI canvas → Add Node → MuAPI/Gemini Omni
  3. Add a 🔑 Gemini Omni API Key node, paste your key, and wire its output to any generation node
  4. Write a prompt and hit Queue Prompt

Tip: If you use the MuAPI CLI, run muapi auth configure --api-key YOUR_KEY once and all nodes will pick it up automatically — no need to paste the key anywhere.

Node Reference

🔑 Gemini Omni API Key

Set your muapi.ai API key once and wire the output to all Gemini Omni generation nodes. Alternatively, leave every api_key field blank — nodes automatically read from ~/.muapi/config.json if you’ve authenticated via the CLI.

🎬 Gemini Omni Text to Video

Generate a video from a text description using the Google Gemini Omni text-to-video API.

Field Values Default
api_key Wire from API Key node or leave blank for CLI config —
prompt Text describing the video —
duration 4 / 6 / 8 / 10 seconds 8
aspect_ratio 16:9 / 9:16 16:9
resolution 720p / 1080p / 4k 1080p
audio_id_1 … audio_id_3 (none) or one of 30 Google Gemini AI voice names — up to 3 voices (none)
character_id_1 … character_id_3 Optional — character IDs from Create Character node — up to 3 —
seed -1 (random) or 0–2147483647 -1

Outputs: video_url (STRING) · request_id (STRING)

🎬 Gemini Omni Image to Video

Animate up to 5 reference images into a video using the Google Gemini Omni image-to-video API.

Field Values Default
api_key Wire from API Key node —
prompt Text describing the animation —
image_1 Required — ComfyUI IMAGE tensor —
image_2 … image_5 Optional — additional reference images —
duration 4 / 6 / 8 / 10 seconds 8
aspect_ratio 16:9 / 9:16 16:9
resolution 720p / 1080p / 4k 1080p
audio_id_1 … audio_id_3 (none) or one of 30 Google Gemini AI voice names — up to 3 voices (none)
character_id_1 … character_id_3 Optional — character IDs from Create Character node — up to 3 —
seed -1 (random) or 0–2147483647 -1

Outputs: video_url (STRING) · request_id (STRING)

🎬 Gemini Omni Video Edit

Restyle or transform a video clip using the Google Gemini Omni video editing API. Optionally supply up to 5 reference images alongside the video (7 total slots — video uses 2, each image uses 1). At least one of video_url or image_1 must be connected.

Field Values Default
api_key Wire from API Key node —
prompt Editing instruction —
duration 4 / 6 / 8 / 10 seconds 8
aspect_ratio 16:9 / 9:16 16:9
resolution 720p / 1080p / 4k 1080p
trim_start 0.0 – 29.0 (seconds) 0.0
trim_end 0.5 – 30.0 (seconds, max window 10s) 8.0
video_url Optional — HTTPS URL or local file path —
image_1 … image_5 Optional — reference images (max 5 with video) —
audio_id_1 … audio_id_3 (none) or one of 30 Google Gemini AI voice names — up to 3 voices (none)
character_id_1 … character_id_3 Optional — character IDs from Create Character node — up to 3 —
seed -1 (random) or 0–2147483647 -1

Outputs: video_url (STRING) · request_id (STRING)

🎤 Gemini Omni Create Audio Profile

Create a custom Gemini Omni AI voice profile. The resulting kie_audio_id can be passed into the audio_id_1 … audio_id_3 fields of the generation nodes.

Field Values Default
api_key Wire from API Key node —
audio_id One of 30 Google Gemini AI voice names (base voice to customise) —
name Profile display name (max 210 characters) —
voice_description Optional — text description of the voice style —
example_dialogue Optional — example speech for the voice —

Outputs: kie_audio_id (STRING) · profile_name (STRING)

🧑 Gemini Omni Create Character

Create a Gemini Omni character from a reference image. The resulting character_id can be passed into the character_id_1 … character_id_3 fields of the generation nodes.

Field Values Default
api_key Wire from API Key node —
image ComfyUI IMAGE tensor — reference image for the character —
descriptions Text description of the character —
character_name Optional — display name for the character —
audio_id_1 … audio_id_3 Optional — voice IDs to associate with this character —

Outputs: character_id (STRING) · character_name (STRING) · character_image_url (STRING)

💾 Gemini Omni Video Saver

Download a Gemini Omni output video URL to disk and decode frames as a ComfyUI IMAGE tensor for downstream processing.

Field Values Default
video_url Wire from any Gemini Omni generation node —
prefix Output filename prefix gemini_omni
save_subfolder Subfolder under ComfyUI/output/ gemini_omni
frame_load_cap Max frames to load (0 = all) 0
skip_first_frames Skip N frames from the start 0
select_every_nth Load every Nth frame 1

Outputs: frames (IMAGE) · filepath (STRING) · frame_count (INT)

Audio Voices

When audio_id is set, a Google Gemini AI voice narrates or accompanies the generated video. Available voices:

achernar · achird · algenib · algieba · alnilam · aoede · autonoe · callirrhoe · charon · despina · enceladus · erinome · fenrir · gacrux · iapetus · kore · laomedeia · leda · orus · puck · pulcherrima · rasalgethi · sadachbia · sadaltager · schedar · sulafat · umbriel · vindemiatrix · zephyr · zubenelgenubi

Example Workflows

Import any of these into ComfyUI via Load or drag-and-drop:

License

MIT — see LICENSE

This site is open source. Improve this page.