> ## Knowledge Base Index
> Fetch the complete knowledge base index at: https://help.seelab.ai/sitemap.xml
> Use this file to discover available pages before exploring further.
> Pure-Markdown content can be obtained by appending a '.md' suffix to the content URLs listed in the sitemap (without the trailing slash).

# Component reference

Each component in the library is a step of your workflow. They are grouped into six families: what comes in (**Source**), what prepares the text (**Tools**), what generates (**Create**), what edits (**Edit**), what delivers (**Export**) and the shared settings (**Variables**).

# Source

| Component | What it does |
| --- | --- |
| **Prompt** | The text passed to connected nodes, which use it as long as they have no local prompt of their own. |
| **CSV Prompts** | A list of prompts imported from a CSV file or typed row by row. **Limit** sets how many rows to run; "Will run x / y prompts" confirms it. |
| **Reference Image** | Up to 10 images from the Resource Hub. **Reference delivery**: **All together** sends them together with each generation, **One by one** sends them one at a time, to roll out a series. |
| **Collection** | A project collection as the starting point, with a limit (20 by default) and the same All together / One by one option. Can watch for new entries. |

| Combine two Reference Image nodes, one set to **All together** (the setting) and the other to **One by one** (your products): each product is placed in turn in the same setting.

# Tools

| Component | What it does |
| --- | --- |
| **LLM** | Writes a text from instructions and up to 5 context images, then injects it as the prompt of connected nodes. Choice of language model, **Instructions** and **Presets**. |

# Create

| Component | What it does |
| --- | --- |
| **Image** | Generates images from the connected prompts, references and variables: model, prompt order, count, format (21:9 to 8:1), quality, 1k, 2k or 4k resolution. |
| **Video** | Generates videos: model, Start / End Frames or Multi Refs mode, format, resolution, duration, sound. Start images can come from an Image node. |

# Edit

| Component | What it does |
| --- | --- |
| **Reframe** | Reframes each incoming image to a target format: 1:1, 16:9, 9:16, 4:3, 3:4 or custom. |
| **Upscale** | Upscales to 1k, 2k or 4k with a preset, an optional prompt and a model choice. |
| **Replace Background** | Changes the setting while keeping the subject, according to the prompt. |
| **Remove BG** | Cuts out incoming images; the **Crop** option trims to the subject. |
| **Edit video** | Edits each incoming video according to the prompt, with reference images. Source video mp4 or webm, 50 MB max. |
| **Ads Generator** | Adapts an image to IAB display formats: Amazon, Billboard 970 x 250, Skyscraper 160 x 600, Medium rectangle 300 x 250, mobile interstitials, or custom formats. |

# Export

| Component | What it does |
| --- | --- |
| **Output** | Gathers the final results for review and download as a .zip, videos included if needed. |
| **Add to Collection** | Automatically files the results in a collection, created if none is chosen. |

# Variables

| Component | What it does |
| --- | --- |
| **Model** | A model chosen once, image or video, then connected to several nodes: you change it in one place. |
| **Number of generations** | The number of results, set once for every connected node. |

# Learn more

* [Getting started with workflows](https://help.seelab.ai/en/article/workflow-etdrlw/)
* [The canvas, Inspector and sections](https://help.seelab.ai/en/article/the-canvas-inspector-and-sections-1qyfkgc/)