Working with Control Layers
Control Layers allow you to provide strong visual guidance (like outlines or depth information) to certain AI models, giving you more precise control over the structure and form of your generated images. This guide explains how to set up and use Control Layers in Fater's Image Editor.
(Before proceeding, ensure you understand the fundamentals by reading the Control Layers (Edges/Depth) Core Concept page.)
Step 1: Prepare or Create Your Control Image
First, you need an image layer that will serve as your control input. This can be an Edges map or a Depth map.
Edges Map: A black and white image where white lines define the desired outlines and structures.
Generate in Fater:
Ensure your desired visual content (from which to extract edges) is visible on the canvas and within the Generation Area Bounding Box.
On the Inpaint Toolbar, click the Generate Edges Map (CPU icon).
To adjust edge detection sensitivity, right-click the CPU icon before clicking to open its options. Set the Low and High Thresholds as needed, then click the icon to generate.
A new layer with the Edges map will be created.
Draw Manually: Create a new empty layer, set your Pencil Tool color to white, and draw your desired edges.
Upload: Import an externally prepared Edges map as a new layer.
Depth Map: A grayscale image where shades represent distance (lighter = closer, darker = further).
Generate in Fater:
Ensure your desired visual content is visible on the canvas and within the Generation Area Bounding Box.
On the Inpaint Toolbar, click the Generate Depth Map (Mountain icon).
To change generation type or tile splits, right-click the Mountain icon before clicking. Select "Monotile" (faster, single pass) or "Multitiles" (slower, potentially more detail, can take up to 5 minutes to compute). Adjust X/Y Split numbers for Multitiles if needed. Then click the icon to generate.
A new Depth map layer will be created.
Upload: Import a Depth map generated from 3D software or other tools as a new layer.
Step 2: Designate the Layer as a Control Layer
Once you have your Edges or Depth map image on a layer:
Go to the Layers Tab in the Right Sidebar.
Select an AI model in the Left Sidebar that is designed for "Controlled" generation (e.g., a model from "Generate" or "Edit" category, with "Controlled" subcategory). This will make the "Control Layers" section visible at the bottom of the Layers Tab.
Drag your Edges/Depth map layer from the main layer list and drop it into the "Control Layers" section. Note : Control Layers generated in Fater are already correctly setted here. (Placeholder: Add visual)
Step 3: Set Preprocessed Type
Within the "Control Layers" section in the Layers Tab, each control layer item will have a dropdown:
Preprocessed Type: Ensure this is correctly set to match your layer's content:
Select
edgesif it's an Edges map.Select
depthif it's a Depth map.
Step 4: Configure the "Controlled" AI Model
In the Left Sidebar, ensure your chosen "Controlled" AI model is selected.
Its control panel will show specific parameters:
Prompt: Write a text prompt describing the content and style you want.
Control Strength / Adherence: (Parameter name varies) Adjust how strongly the AI should adhere to your Control Layer.
Other Parameters: Steps, Guidance, Seed, etc.
Step 5: Generate
Ensure your Control Layer (Edges or Depth map) is visible in the Layers Tab (Eye icon is on).
Verify your Generation Area Bounding Box is set correctly to encompass the area you want the AI to process.
Click the Generate button (✨) in the Floating Prompt Area.
How Control Layers are Used by the AI:
When you generate, the main visible canvas content (excluding layers marked as "Control") is sent to the AI as the base image context (if applicable for the model).
Your active, visible Control Layers are sent separately to the AI model if it supports that type of input (Edges or Depth). They are not rendered as part of the main scene image sent to the model; they act as an additional guiding input.
The AI will then attempt to generate an image that respects your text prompt, the main scene context (if any), and the structural guidance from your Control Layer(s). The result will appear as a new layer.
Tips for Working with Control Layers:
Bounding Box is Key: The AI only "sees" and processes what's inside the Generation Area Bounding Box for both the main scene and when interpreting Control Layers relative to that area.
Control Layer Clarity: Ensure your Edges maps have clean lines and Depth maps have good contrast.
Experiment: The interplay between prompt, Control Layer, and Control Strength matters.
Last updated