> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rainfrog.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Rainfrog Node Canvas: Connect Nodes, Generate Campaigns
> Learn how Rainfrog's visual node canvas replaces text prompts with a drag-and-connect workflow that drives AI image and video generation.
The Rainfrog node canvas is your creative workspace. Instead of starting from a blank text prompt, you build a visual graph by connecting asset nodes to generator nodes — and Rainfrog's graph engine handles the rest. Every decision you make on the canvas, from the product you're featuring to the lighting style you want, becomes part of the generation automatically.
## The canvas at a glance
When you open a project in the [Rainfrog studio](https://app.rainfrog.ai), you start on a blank canvas. Add nodes from the floating dock on the left (or from the canvas context menu), wire them together, and click **Generate**. The canvas auto-saves your work as you go, so you never lose progress mid-session.
Think of the canvas as a visual recipe. Each node you place represents one ingredient — a product, a character, a style, a setting — and the connections between them tell Rainfrog how those ingredients relate in the final output.
You can also drop a **template** from the dock (Fashion Shooting, Product Still Life, or UGC Content) to place a pre-wired graph and start faster.
## Node types
Rainfrog nodes fall into four groups.
### Asset and input nodes
| Node | Role |
| --------------- | -------------------------------------------------------------------------------------- |
| **Product** | Holds the product you want to feature — a garment, bag, packshot, or any physical item |
| **Character** | Defines the person in the scene, from Rainfrog's stock library or your own reference |
| **Style** | Encodes the visual treatment: color palette, lighting, and mood |
| **Environment** | Sets the scene — the background or setting your product and character inhabit |
| **Text** | Optional description text you can enhance from graph context |
| **Image** | An uploaded still image you can wire into generators as a reference |
| **Video** | An uploaded video clip used as a reference for supported video models |
### Specialized generators
| Node | Role |
| -------------------------- | ----------------------------------------------------------------------- |
| **Full Body Generator** | Builds front, side, and back views of a character wearing your products |
| **Outfit Sheet Generator** | Produces a flat-lay / outfit sheet from a character and products |
### Output nodes
| Node | Role |
| ------------------- | ------------------------------------------------------------------------------------------------------ |
| **Image Generator** | Main still-image output. Accepts description, characters, products, references, style, and environment |
| **Video Generator** | Short-form video output. Inputs change based on the video model you select |
### Utility
| Node | Role |
| -------- | ------------------------------------------------------------------------------ |
| **Note** | A sticky note on the canvas — for your own comments, not wired into generation |
Rainfrog is designed for creative direction, not prompt engineering. Most workflows run without writing prompts. The optional **Text** node is there when you want extra direction — not as the primary way to drive the studio.
## How the graph becomes a generation
When you click **Generate** on an Image Generator, Full Body Generator, Outfit Sheet Generator, or Video Generator, Rainfrog resolves every connected input and assembles the generation automatically. You stay focused on which assets are wired — not on crafting a prompt from scratch.
## Connecting nodes
To connect two nodes, drag from the **output handle** (the dot on the right edge) of a source node to the **input handle** (the dot on the left edge) of a generator or other target node. A wire appears between them, confirming the connection.
A few things to know:
* You can connect multiple products or characters to a single Image Generator when the handles allow it.
* Input labels on the generator tell you what each handle expects (for example **Products**, **Character**, **References**, **Style**, **Environment**).
* To remove a connection, click the wire and press **Delete**, or drag the handle away from its target.
Connecting an existing image output into a new Image Generator's **References** input is a common way to iterate or edit a previous result without rebuilding the whole graph.
## Undo, redo, and auto-save
Every change you make on the canvas — adding a node, connecting wires, choosing an asset — is auto-saved to your project. You don't need to manually save.
If you make a mistake or want to backtrack, use the standard keyboard shortcuts:
* **Undo:** `Cmd + Z` (Mac) / `Ctrl + Z` (Windows)
* **Redo:** `Cmd + Shift + Z` (Mac) / `Ctrl + Shift + Z` (Windows)
Your node configuration stays inside the project. Come back later, swap a node, and regenerate anytime.