Awesome-omni-skill doubao-watermark-remover
Remove the visible Doubao (豆包) AI watermark from images. Use when asked to remove Doubao watermarks, clean Doubao-generated images, or process images with the "豆包AI生成" watermark.
install
source · Clone the upstream repo
git clone https://github.com/diegosouzapw/awesome-omni-skill
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/diegosouzapw/awesome-omni-skill "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/tools/doubao-watermark-remover" ~/.claude/skills/diegosouzapw-awesome-omni-skill-doubao-watermark-remover && rm -rf "$T"
manifest:
skills/tools/doubao-watermark-remover/SKILL.mdsource content
Doubao Watermark Remover
Dependencies
- Python 3.9+
- Pillow (install with
)pip install -r requirements.txt
Quick start
-
Install dependencies in the scripts folder:
cd doubao-watermark-remover/scripts && pip install -r requirements.txt
-
Run the CLI:
python remove_watermark.py <input-image> <output-image>
CLI usage
- Parameters:
: path to the Doubao watermarked imageinput-image
: path for the cleaned image (format inferred from extension)output-image
Example:
python remove_watermark.py ./in.png ./out.png
What this skill provides
: CLI entry point and core algorithm.scripts/remove_watermark.py
: Tool to capture Doubao watermark alpha map.scripts/capture_alpha_map.py
: Math, detection rules, and usage guide.references/algorithm.md
Workflow
- First time: Generate alpha map using
capture_alpha_map.py - Use
for batch processing.remove_watermark.py
Capturing Alpha Map (First Time Required)
Since Doubao watermark pattern may vary, you need to capture the alpha map first:
-
Generate a pure black image (0,0,0) and have Doubao add watermark to it
-
Run the capture tool:
python capture_alpha_map.py <black-with-watermark.png> <width> <height>Size options: 140x35 (small images) or 180x40 (large images)
-
The tool will save the alpha map to
folderassets/
Notes
- Doubao watermark "豆包AI生成" appears in the bottom-right corner
- Script uses Pillow for image IO and per-pixel edits
- Output format is inferred from the output file extension