sugarkworkandClaude Opus 4.7
026d537859
Port the MyPsdWriter C# implementation (https://github.com/.../mypsd) to Python. The writer uses only the standard library plus numpy/PIL, so the psd-tools dependency can be dropped entirely. - New psd_writer.py with PsdLayer + write_psd / write_psd_from_pil_layers APIs. Supports Raw / RLE / Zip / ZipPredict compression, defaulting to RLE for broad reader compatibility (Photoshop, Affinity, CSP). - Keep the C# trick of writing a 3-channel RGB header while individual layer records carry RGBA, so Photoshop does not surface a phantom unnamed alpha layer. - extract_segments.save_images_to_psd is now a thin wrapper around the new writer; the psd-tools-based implementation and the internal header.channels hack are gone. - Drop psd-tools from requirements.txt and pyproject.toml. Verified by round-tripping multi-layer PSDs (RLE and Raw) back through psd-tools: layer names match, per-layer pixel diff = 0, composite preview diff = 0. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
19 lines
452 B
TOML
19 lines
452 B
TOML
[project]
|
|
name = "comfyui-auto-mosaic"
|
|
description = "Auto detect and process sensitive areas using YOLO with mosaic, blur, white out, and PSD export."
|
|
version = "1.0.0"
|
|
readme = "README.md"
|
|
requires-python = ">=3.9"
|
|
dependencies = [
|
|
"ultralytics",
|
|
"opencv-python-headless"
|
|
]
|
|
|
|
[project.urls]
|
|
Repository = "https://github.com/sugarkwork/comfyui-auto-mosaic"
|
|
|
|
[tool.comfy]
|
|
PublisherId = "sugarknight"
|
|
DisplayName = "ComfyUI AutoMosaic"
|
|
Icon = ""
|