The Generation Area (Bounding Box)

In Fater's Image Editor, the Generation Area defines the specific rectangular region on the canvas where AI models will perform their operations (like generating content, inpainting, or upscaling). It's visualized as a dashed bounding box.

Understanding and controlling this area is crucial for achieving desired results, managing detail, efficient generation, and ensuring proper lighting integration.


What the AI "Sees"

  • Confined Operation: When you initiate an AI task, the model only processes the pixels inside the current Generation Area Bounding Box. Anything outside this box is effectively invisible to the AI during that specific generation instance – it's treated as if it doesn't exist for that operation.

  • Masking Constraint: Similarly, you can only draw or apply masks within the boundaries of the current Generation Area. Tools like the mask brush will stop at the edge of the box.


Tip: Include Relevant Light Sources for Realism

  • Lighting Context: The AI analyzes the content within the bounding box to understand the scene's lighting environment. It uses this context to render generated objects with appropriate shadows, highlights, and reflections, making them look integrated.

  • Impact of Exclusion: If significant light sources influencing the area you're editing (e.g., a bright window, a lamp casting strong light, the sun's direction) are outside the bounding box, the AI won't "see" them. This can result in generated elements having flat, generic, or inconsistent lighting and reflections that don't match the overall scene.

  • Recommendation: Whenever practical, try to adjust your Generation Area to include the key light sources that directly affect the region you are generating or inpainting. You might need to make the bounding box slightly larger than the absolute minimum required for your subject to capture this essential lighting information. Balancing this with the need for detail (see below) is part of mastering the tool.


Controlling the Generation Area

You primarily control the Generation Area using tools from the Inpaint Toolbar:

  • Resizing Handles: When no tool is active, handles appear around the bounding box, allowing you to click and drag to resize the area manually.

  • Resize Canvas to Layer (Crop icon): Automatically adjusts the bounding box to fit the exact dimensions and position of a selected layer. This is essential after adding a new base image.

  • Resize Canvas to Masked Area (Crop icon variant/context): Adjusts the bounding box to tightly enclose the currently masked area. When no pixel is masked, it crops to all visible layers instead.


Why Control the Size? Pixel Density & Detail

AI image models generally work with a fixed processing capacity related to the number of pixels they can handle effectively in one go. The size of your Generation Area directly impacts the pixel density the AI works with:

  • Large Generation Area: The AI spreads its processing power across a wider area. Good for overall scenes, but details might be less refined. May capture broader lighting context.

  • Smaller Generation Area: Focusing the AI's capacity increases pixel density within that region.

Benefits of a Smaller Generation Area:

  • Higher Detail Quality: Allows significantly higher quality details, crucial for:

    • Faces and Hands

    • Small Objects

    • Background Elements

    • Specific Textures

  • Efficient Inpainting: Ensures the AI focuses only on fixing the masked spot with maximum detail.

Balancing Detail and Lighting: You may need to find a balance. For maximum detail on a small feature, use a tight bounding box. For best lighting integration when adding a new object, ensure the box includes relevant light sources, even if it means slightly lower detail on the object itself compared to a hyper-focused box.

Workflow Example (Adding Detail to a Face):

  1. Generate your main scene (potentially with a larger Generation Area to capture lighting).

  2. Select the resulting layer.

  3. Use Masking Tools to mask the face.

  4. Use Resize Canvas to Masked Area to shrink the Generation Area Bounding Box tightly around the face mask.

  5. Refine your prompt (e.g., detailed portrait, sharp focus, realistic skin texture) and AI parameters.

  6. Click Generate. The AI now focuses its power on the face for finer detail.

Mastering the Generation Area involves understanding its impact on both detail and lighting to achieve the best overall result.

Last updated