Background Remove Skill

Remove backgrounds from images using AI (rembg/U2-Net) or built-in methods.

Output: PNG or WebP with transparent background.

Quick Examples

Prerequisites

• •

  rembg - AI-based background removal (recommended)

  bash

  pip install rembg
  # Or with GPU acceleration (faster, requires CUDA)
  pip install rembg[gpu]
  

• •

  Pillow - Required for image processing

  bash

  pip install Pillow
  

The first run will download the U2-Net model (~170MB) which is cached for future use.

Methods

Workflow

Step 1: Gather Requirements (REQUIRED)

Use the AskUserQuestion tool for each question. Ask ONE question at a time.

Q1: Image Source

"Which image(s) should I remove the background from?

Please provide the file path or paste the image."

Wait for response.

Q2: Method (Optional)

"Which removal method?

• •AI (rembg) - Best quality, works on any image (default)
• •Built-in - Faster, best for white backgrounds"

Wait for response. Default to AI if user doesn't specify.

Q3: Output Location (Optional)

"Where should I save the result?

• •Same location with _nobg suffix (default)
• •Custom path"

Wait for response.

Step 2: Execute Background Removal

Single image:

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i "/path/to/image.jpg" \
  -o "/path/to/output.png"

Batch processing:

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i "/path/to/img1.jpg" "/path/to/img2.png" "/path/to/img3.webp" \
  -o "/path/to/output_folder"

Using built-in method (faster for white backgrounds):

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i "/path/to/icon.png" \
  -m builtin

Step 3: Deliver Result

1. •Show the result to the user
2. •Confirm the background was removed successfully
3. •Offer to:
   • •Process additional images
   • •Try a different method if quality isn't satisfactory
   • •Adjust output format (PNG vs WebP)

Script Parameters

Output Formats

The output format is determined by the file extension:

Integration with Other Skills

This skill can be called by other skills that need background removal:

From Python (import)

import sys
sys.path.insert(0, "${SKILL_PATH}/skills/background-remove/scripts")
from background_remove import remove_background

result = remove_background("/path/to/image.png", "/path/to/output.png", method="rembg")
if result.get("success"):
    print(f"Saved to: {result['file']}")
else:
    print(f"Error: {result['error']}")

From Command Line

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i "/path/to/image.png" \
  -o "/path/to/output.png" \
  -m rembg

Error Handling

rembg not installed:

rembg not installed. Install with: pip install rembg[gpu] (or pip install rembg for CPU-only)

The script will automatically fall back to the built-in method.

Image not found:

Image not found: /path/to/image.png

Processing failed:

• •Try a different method
• •Check if the image file is corrupted
• •Ensure sufficient memory for large images

Tips for Best Results

1. •Use rembg for photos - AI handles complex edges (hair, fur, transparent objects)
2. •Use builtin for graphics - Faster for icons/logos with clean white backgrounds
3. •Check edges - If edges are rough, the AI method usually gives better results
4. •Batch process - Process multiple images at once for efficiency
5. •GPU acceleration - Install rembg[gpu] for faster processing on NVIDIA GPUs

Examples

Remove background from a photo

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i "product_photo.jpg" \
  -o "product_transparent.png"

Batch process a folder

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i photos/*.jpg \
  -o "transparent_photos/"

Fast removal for icons (white background)

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i "icon.png" \
  -m builtin

Output as WebP (smaller file size)

python3 ${SKILL_PATH}/skills/background-remove/scripts/background_remove.py \
  -i "photo.jpg" \
  -o "result.webp"