image-and-face-apps
Use PaddleGAN ppgan.apps image and face application predictors for single-image generation, restoration, denoising, super-resolution, style transfer, face parsing, enhancement, cartoonization, makeup transfer, and StyleGANv2 latent workflows.
- License 许可
- Apache-2.0
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- image-and-face-apps
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Use PaddleGAN ppgan.apps image and face application predictors for single-image generation, restoration, denoising, super-resolution, style transfer, face parsing, enhancement, cartoonization, makeup transfer, and StyleGANv2 latent workflows.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/image-workflows.md6.9 KB
- 参考文档references/predictor-api.md8.4 KB
- 参考文档references/troubleshooting.md4.5 KB
03 SKILL.md 正文
image-and-face-apps
Use this sub-skill when the request stays in one image, one face, or one latent code at a time.
Route here
- photo-to-anime and portrait cartoonization
- face parsing, face enhancement, and makeup transfer
- single-image denoising, deblurring, inpainting, super-resolution, and depth
- StyleGANv2 sampling, fitting, mixing, and editing
- Pixel2Style2Pixel inversion and SinGAN-style single-image generation
Do not route here
- video restoration, motion driving, or lip-sync ->
video-and-audio-apps - exported or static inference model workflows ->
deployment-export - dataset download or preprocessing work ->
data-preparation - full runs without explicit weights, media, and backend approval
Start with
- Run
python scripts/check_image_app_deps.pyfor a safe import/dependency check. - Open
references/image-workflows.mdto pick the right model family. - Open
references/predictor-api.mdfor constructor andrun()shapes. - Open
references/troubleshooting.mdwhen weights, dlib, CLIP, or face detection are missing.
Output rule
Use a caller-owned output directory and keep file names predictable. Prefer explicit weight_path values when reproducibility or offline use matters.
Notes
- Use installed
ppganpackage imports and bundled helpers, not source-checkout scripts. - If a predictor can auto-download weights, say so before relying on that behavior.
- CLIP-guided StyleGAN editing is optional and not the default route here.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.