BluePixel Editor

Open-source AI image/video editing agent by vivo Image Group.

Why BluePixel Is Different

• Human-like Natural Language Interactions — The agent provides personalized recommendations for each input image, delivering better editing effects.
• 4K Native Output — Process and output at full 4K resolution with no downscaling.
• Pixel Preservation — Untouched areas stay byte-identical. Zero degradation to unedited regions.

Quick Start

# Install the BluePixel CLI
pip install bluepixel-cli

# Authenticate (uses Google OAuth)
bluepixel auth login

# Edit an image with a natural language prompt
bluepixel edit input.jpg --prompt "Remove the person in the background" -o output.jpg

# Batch edit
bluepixel edit images/ --prompt "Enhance colors and upscale to 4K" -o results/

Core Capabilities

1. Natural Language Editing

Describe edits in plain text. No brushes or masks required (though supported).

bluepixel edit photo.jpg --prompt "Change the sky to a golden sunset"
bluepixel edit product.png --prompt "Remove all text from the packaging"

2. 4K Native Output

Process and output at full 4K resolution — no downscaling during AI processing.

bluepixel edit input.jpg --prompt "Enhance details" --resolution 4k -o output.jpg

3. Pixel Preservation

Untouched areas stay byte-identical. Zero degradation to unedited regions.

bluepixel edit photo.jpg --prompt "Replace only the background" --preserve-pixels -o output.jpg

4. Supported Tasks

Task	Example Prompt
Background removal	"Remove the background and make it transparent"
Style transfer	"Transform this photo into a watercolor painting"
Object editing	"Add a red hat to the person"
Text removal	"Remove all text while keeping the background"
Super resolution	"Upscale to 4K and enhance details"
Image restoration	"Recolor and restore, remove scratches and damage"
Virtual try-on	"This person is wearing the jacket from image2.jpg"

5. API Integration

from bluepixel import BluePixelClient

client = BluePixelClient()
result = client.edit(
    image="input.jpg",
    prompt="Remove the person in the background",
    resolution="4k",
    preserve_pixels=True,
)
result.save("output.jpg")

6. OpenClaw Integration

BluePixel works as an OpenClaw skill. When a user asks to edit an image:

1. Identify the input image path
2. Determine the edit instruction from the user's request
3. Run bluepixel edit with the appropriate prompt
4. Return the output path to the user

# From within an OpenClaw agent
bluepixel edit "$INPUT_PATH" --prompt "$USER_PROMPT" -o "$OUTPUT_PATH"

Configuration

Config file: ~/.bluepixel/config.yaml

default_resolution: 4k
preserve_pixels: true
output_format: png
api_endpoint: https://api.bluepixel.vivo.com/v1

References

• API documentation — full REST API reference
• Supported models — available AI models and their capabilities