类库
› see-through
shitagaki-lab/see-through
这是一个基于深度学习的动画角色图层分解工具，能够将单张动漫人物图片自动分解为多个可编辑的语义图层（如头发、面部、服装等），最多支持23层，并推断绘制顺序，从而将静态插图转换为可操控的2.5D模型。
技术栈
测试
pytest
unknown
查看全部依赖 (42)
依赖
-e
unknown
Cython
unknown
NumPy
unknown
Pandas
unknown
PyQt6
unknown
PyQt6-Qt6
unknown
Pydantic
unknown
accelerate
unknown
click
unknown
colorama
unknown
convnext_perceptual_loss
unknown
diffusers
unknown
einops
unknown
grad-cam
unknown
huggingface-hub
unknown
ipykernel
unknown
kornia
unknown
lpips
unknown
matplotlib
unknown
natsort
unknown
omegaconf
unknown
opencv-python
unknown
pillow
unknown
pillow-jxl-plugin
unknown
psd-tools
unknown
py7zr
unknown
pycocotools
unknown
pyyaml
unknown
qtpy
unknown
safetensors
unknown
sam-2
unknown
scikit-image
unknown
scikit-learn
unknown
scipy
unknown
simple-parsing
unknown
termcolor
unknown
timm
unknown
tokenizers
unknown
tqdm
unknown
transformers
unknown
uvloop
unknown
wheel
unknown
