About Auto Focus

Auto Focus is an AI-powered tool that ensures your subjects are always perfectly framed, regardless of the original image size or orientation. By automatically detecting key elements, it adjusts the crop and zoom to keep the most important parts of your image front and center.

Accessing Auto Focus feature

Select an Image Layer in your design.
In the right-hand property panel, click on Modify with AI.
Select Auto Focus from the available AI tools.

Select Your AI Detection Model

To begin, choose the model that best suits your image content:

People Model: Optimized specifically for human subjects. This model allows for advanced portrait framing (e.g., focusing only on the face, head, shoulders or the full body).
General Model: Best for a wide range of subjects including objects, animals, vehicles, or complex scenes with multiple different elements.

Using the General Model (Tag Selection)

When you select the General Model, the AI identifies specific elements within your image as individual "Tags."

Manual Selection: After clicking Launch Detection, a list of detected tags will appear (e.g., Glasses, Trousers, Woman). You must select one or more tags to tell the AI what to focus on.
Merging Multiple Tags: If you select multiple tags, the AI creates a Merged Selection. This ensures the final framing includes all selected elements in a single cohesive view.
Automatic Merging in Generations: During automated variations (API/Quick generation/..), if the AI detects several of your chosen tags in a new image, it will automatically merge them to maintain your composition.

Using the People Model (Framing & Priority)

If you select the People model, you can customize how the AI frames the person:

Portrait Framing

Choose exactly which part of the subject should be the focal point:

Face / Head: Tight crops for portraits.
Shoulders: Ideal for "talking head" style layouts.
Full Body: Keeps the entire person within the frame.

Zoom Level

Control how much the AI should "push in" on the detection:

Max / Medium / Low: Adjust the intensity of the crop.
Off: Matches the detection area exactly without adding extra zoom.

Multi-Subject Priority

If your image contains more than one person, tell the AI who matters most:

Largest subject: Focuses on the person closest to the camera.
Centered / Leftmost / Rightmost: Focuses based on their position in the frame.
All subjects: Merges the selection to keep everyone in view.

💎 Enterprise Feature: Celebrity Matching For Enterprise subscriptions, you can prioritize subjects based on specific names. If your image contains a known public figure, you can input a name (e.g., "Michael Jordan" or simply "Jordan"). The AI will identify the celebrity and automatically center the focus on them.

Launch Detection & Apply

Once you have configured your settings:

Click Launch Detection. The AI will scan the image and highlight detected areas.
If using the General model, we ask you to pick specific tags (e.g., "Glasses," "Woman") to refine the focus area.
Review the Merged Selection dashed line on your image.
Click Apply AI Focus to confirm.

Automatic Processing for Generations

To automate your workflow, toggle "Enable Auto-focus for generation".

All future variations whether created via API, Spreadsheet, or Quick generation will automatically apply these focus rules to any new image uploaded to this layer. This ensures perfectly framed headshots or product shots every time, at scale.

Quick Edits from the Layer Panel

You don't always need to open the full Auto Focus modal to make changes. From the AI Image tab in the right-hand layer properties panel, you can modify your settings at any time without launching a new detection.

Instant Adjustments: Quickly toggle between People or General models or change your Portrait Framing (e.g., switching from "Face" to "Full body") directly from the sidebar.
Manual Tag Entry: For the General model, this is especially useful as you can directly type in your desired tags instead of waiting for a full image scan.