A batch post-production toolkit for AI artists who ship hundreds of images a day — watermark, strip generation metadata, run a configurable filter pipeline, and auto-name files from booru character tags, all in one pass.
It started as a tool to slap a watermark on images quickly. It has since grown into a small Swiss-army knife for the last mile of an AI image workflow: the repetitive, easy-to-forget steps.
Point it at a folder, set your defaults once, and every export comes out watermarked, metadata-free, optionally filtered, and sensibly named — without touching the originals.
AI artists and creators who generate images and videos in batches and need to:
- Apply one or two consistent watermarks before publishing
- Strip generation metadata (EXIF, ComfyUI prompts/workflow, XMP, ICC)
- Optionally run post-processing filters to alter or finish the output
- Auto-rename files from the character tags embedded in ComfyUI metadata
- Do all of the above to a whole batch at once, not file by file
- Drag & drop — drop files or folders directly onto the list; no file pickers required
- Dual watermarks — a primary watermark plus an optional second one, each with independent size, opacity, and corner
- Corner placement — pick a fixed corner or randomize per image (consecutive repeats are avoided)
- Live preview — size, opacity, and position update in real time as you adjust them
- Images & video — PNG, JPG/JPEG, and MP4, AVI, MOV, MKV
- Metadata stripping — output files are saved without EXIF, PNG text chunks, XMP, or ICC profiles
- Configurable post-processing pipeline (disabled by default) — see below
- Auto-naming from ComfyUI metadata — extracts character tag candidates from the positive prompt and renames output files; a character library remembers your past choices (images only)
- Batch processing — progress bar with mid-batch cancel; failures are logged and summarized instead of crashing the run
When enabled, filters run in a fixed order. Each step is a no-op at its neutral value, so you only pay for what you turn on:
- JPEG artifact removal — Non-Local Means denoising
- Upscale — relative resize (Lanczos, bicubic, Hamming, bilinear, box, nearest)
- Kuwahara blur — edge-preserving smoothing (mean or gaussian)
- Median filter
- Downscale — relative resize (same sampler options)
- Sharpen — unsharp mask with radius and threshold
- Hue / Saturation / Brightness
- Chromatic aberration
- Vignette
- JPEG compression simulation — quality and chroma subsampling
- Film grain — structured grain, optionally monochromatic
- Gaussian noise — per-channel, optionally monochromatic/inverted
Grab the latest AImgKit.exe from the Releases page. No Python required — double-click and run. ffmpeg is bundled, so video audio is preserved automatically.
SmartScreen / antivirus warning — The binary is not code-signed and is shipped as a single-file build (UPX disabled to reduce false positives). Windows SmartScreen or your AV may still flag it as unknown. Click "More info → Run anyway" to proceed, or run from source instead.
pip install -r requirements.txt
python AImgKit.pyOn Windows, ffmpeg is bundled with the .exe; when running from source, video audio is preserved only if ffmpeg is available on your PATH (or next to the script). Without it, videos are still produced but without audio.
- (Optional) Set an output folder — leave empty to save to
<source>/watermarked_clean - Add a watermark image via Browse or drag & drop; optionally enable Watermark 2
- Drag images/videos into the list
- Adjust settings across the Watermark, Post-Processing, and Auto-Name tabs (preview updates live)
- Click Start Batch Processing
Settings, the character library, and logs are stored per user:
- Windows:
%LOCALAPPDATA%\AImgKit\ - Other OS:
~/AImgKit/
See watermark_config.example.json for the full set of defaults.
| Path | Purpose |
|---|---|
AImgKit.py |
Main GUI application (CustomTkinter) |
post_filters.py |
Post-processing filter implementations and pipeline |
comfy_metadata.py |
Reads ComfyUI PNG metadata and extracts character tag candidates |
core/ |
GUI-independent logic: formats, watermark sizing/placement, corner selection, naming, metadata-free saving |
tests/ |
pytest suite |
Pre-built binaries are on the Releases page. To build from source:
pip install -r requirements.txt
pip install pyinstaller
pyinstaller AImgKit.specThe spec file bundles the icon, CustomTkinter/tkinterdnd2 assets, and ffmpeg.exe (place it in the project root before building). Output: dist/AImgKit.exe.
pip install -r requirements.txt -r requirements-dev.txt
pytest -q- Python 3.10+
- Pillow — image manipulation
- OpenCV (
cv2) — video processing and filters - NumPy — array operations
- tkinterdnd2 — drag & drop support
- customtkinter — modern themed (dark/light) UI
Exact pinned versions are in requirements.txt.
MIT — see LICENSE
The Windows build bundles an unmodified static LGPL build of ffmpeg.exe
(used only to remux audio, as a separate executable). FFmpeg and the other
third-party components retain their own licenses — see
THIRD_PARTY_NOTICES.