byted-tos-image-process

Volcengine TOS Image Process

Inspect and transform images stored in Volcengine TOS — metadata, format conversion, resize, watermark, blind watermark, and AI-powered image understanding.

Setup (once per environment)

Install dependencies on first use:

cd {baseDir}
pip install -r {baseDir}/requirements.txt

Then run scripts with Python 3.7+:

python3 {baseDir}/scripts/<script>.py <args>

If you see a ModuleNotFoundError for tos, reinstall dependencies.

Environment Variables

This skill relies on the TOS identity declared in the metadata block. Common runtime variables are:

Environment Variable	Required	Description
TOS_ACCESS_KEY	Yes	TOS access key ID
TOS_SECRET_KEY	Yes	TOS secret access key
TOS_ENDPOINT	Yes	TOS endpoint URL
TOS_REGION	Yes	TOS region
TOS_BUCKET	Yes	Source bucket that stores the image
TOS_OBJECT_KEY	No	Source object key of the image. Can be overridden with --key
TOS_SECURITY_TOKEN	No	STS session token when using temporary credentials
TOS_SAVEAS_BUCKET	No	Default target bucket for saving processed results
TOS_SAVEAS_OBJECT_PREFIX	No	Default key prefix for saving processed results

Quick start (common tasks)

# Read image metadata
python3 {baseDir}/scripts/image_info.py --key photo.jpg

# Convert to WebP
python3 {baseDir}/scripts/image_format.py --key photo.jpg --f webp --output converted.webp

# Resize to width 500
python3 {baseDir}/scripts/image_resize.py --key photo.jpg --width 500 --output resized.jpg

# Draw points and connecting lines
python3 {baseDir}/scripts/image_draw.py --key photo.jpg \
  --points 50x50-200x120-320x220 --line --color FF0000 --output draw.jpg

# Zoom by resize + crop
python3 {baseDir}/scripts/image_zoom.py --key photo.jpg \
  --resize-w 1200 --crop-w 500 --crop-h 400 --gravity center --output zoom.jpg

# Add visible text watermark
python3 {baseDir}/scripts/image_watermark.py --key photo.jpg \
  --text "My Brand" --font fangzhengshusong --color FF0000 --size 72 \
  --gravity center --output watermarked.jpg

# Embed blind watermark (requires ≥512×512 image and account permission)
python3 {baseDir}/scripts/image_blindwatermark.py --key photo.jpg \
  --kv text=HelloBlind --output blind.jpg

# Run a custom process string
python3 {baseDir}/scripts/image_process.py --key photo.jpg \
  --process "image/resize,w_300,h_300,m_fill" --output filled.jpg

# Preview the resolved request without calling TOS
python3 {baseDir}/scripts/image_resize.py --key photo.jpg \
  --width 500 --dry-run --json

# AI-powered understanding for a TOS image object (requires whitelist)
python3 {baseDir}/scripts/image_understanding.py --key photo.jpg \
  --prompt "Describe this image in detail"
python3 {baseDir}/scripts/image_understanding.py --key document.png \
  --prompt "识别图片中的所有文字内容"