tongflow-api-gemini

Official TongFlow plugin. Text and image generation via Google Gemini.

Capabilities

Implements these ABI slots (runs locally as a Python process, no GPU):

• Generate / rewrite text (gen-text) — create or edit copy from a prompt.
• Combine text (combine-text) — merge multiple text nodes into one.
• Split long text (split-text) — break a long passage into chunks.
• Generate image (image-gen) — text → image via Nano Banana.
• Edit image (image-edit) — image + text → image.
• Fuse images (image-fusion) — combine multiple reference images into one.
• Arrange & batch groups (arrange-group) — group and arrange text/clip batches for downstream processing.
• Filter or drop clips (drop-video) — drop unwanted clips by rule.

Image slots default to Nano Banana Lite (gemini-3.1-flash-lite-image) — Google's low-latency, 1K image model — via the same generateContent API and key as the text slots.

Credentials

Add in TongFlow Settings (gear icon, top-right):

Key	Required	Notes
GEMINI_API_KEY	✅	Create one at aistudio.google.com/apikey. GOOGLE_API_KEY is accepted as a fallback.
GEMINI_MODEL	optional	Override the default text model (e.g. gemini-2.0-flash).
GEMINI_IMAGE_MODEL	optional	Override the default image model (e.g. gemini-3.1-flash-image, gemini-3-pro-image).
GEMINI_IMAGE_SIZE	optional	Output resolution (1K default; 2K/4K require a Flash/Pro image model).
GEMINI_IMAGE_ASPECT_RATIO	optional	Force an aspect ratio (e.g. 16:9). Defaults to the node's width/height, snapped to the nearest supported ratio.

Values are stored locally and take effect without a restart.