AI Lab

How to access: Click the AI Lab icon in the top-left corner of the Edit interface.

AI Lab – People Removal

Passerby Selection

• Auto Detection: The system automatically identifies passerby areas. Click again to re-detect the mask. After auto detection, manual tuning brush can be used to refine the mask.
• Manual Tuning: Manually paint over passerby areas to add or remove mask regions. Brush size, softness, and opacity are adjustable.

Apply Effect: Click Continue Generate to apply. Each application consumes 0.3 credits.

AI Lab – Smart Removal

Removal Mode

• Magic Removal: Best suited for removing large or complex distractions, delivering more natural results. 0.5 credits per use.
• Quick Removal: Best suited for removing small distractions with faster processing. 0.1 credits per use.
• Manual Tuning: Manually paint areas to add or remove mask regions. Brush size and softness are adjustable.

Supporting Tools

• Subject Protection: Protects detected subjects or associated items; brushing is limited to areas outside the subject.
• Edge Optimization: Automatically refines the transition edges around the subject.

Apply Effect: Click Remove to execute. Each application consumes the corresponding credits.

Choose how you would like the effects to be applied after using AI Lab (People Removal/Smart Removal) and returning to the Edit interface.

• Apply to Current Image (Keep one image copy)
  The effect will be applied directly to the current photo in the editing interface. The original source file will not be modified.

• Generate Copy (Create a new copy)
  A new copy with the applied effect will be created. You can change the storage location for generated copies in Settings.

AI Lab – AI Set Fusion

AI Set Fusion helps you create more complete and atmospheric scene-based portraits through intelligent cutout, background replacement, and layered foreground composition.

Feature details

• Automatically cuts out the subject, with manual brush refinement available when needed.
• Supports both Subject Only and Subject + Connected Objects recognition modes.
• Lets users choose from official scene packs or upload custom backgrounds.
• Supports layered foreground assets for additional composition control, including foreground order adjustment.
• Allows repositioning, scaling, rotation, and horizontal / vertical flipping of the subject.
• After generation, Character Lighting can be used to improve the visual blend between the subject and the background.

Asset and upload limits

• Official packs can be filtered by scenarios such as General, Student, Sports, Children, Infant, Outdoor, and Indoor.
• Supports up to 50 foreground elements in one composition.
• Custom backgrounds support JPG / PNG / WebP; custom foregrounds currently support PNG only.
• Custom uploaded assets are limited to a maximum long side of 4000 px and 20 MB per file.

Notes

• Generated results appear in the Results list and can be saved individually or in batches.
• Saved results return to the gallery as new images.
• Unsaved AI Scene results are lost when leaving the feature.
• Generation requires a background on the canvas, and the subject must remain within the visible canvas area.