Spaces:
Running
Running
Case Zero - initial public release (fully local: Qwen2.5-1.5B via llama.cpp + Supertonic, custom pixel-noir SPA via gradio.Server)
414dc55 Download src/case_zero/schemas/visual.py from build-small-hackathon/case0: direct link, hf CLI and curl.
- Browser
- Download file 914 Bytes
-
https://huggingface.co/spaces/build-small-hackathon/case0/resolve/refs%2Fpr%2F2/src/case_zero/schemas/visual.py
- Command line
-
hf download hf://spaces/build-small-hackathon/case0@refs/pr/2/src/case_zero/schemas/visual.py
-
curl -L -o visual.py https://huggingface.co/spaces/build-small-hackathon/case0/resolve/refs%2Fpr%2F2/src/case_zero/schemas/visual.py
914 Bytes
| """The visual descriptor the model emits per character / scene / prop. | |
| It drives the offline PIL compositor: the model creatively specifies appearance and | |
| the code only renders that specification into pixel art. | |
| """ | |
| from __future__ import annotations | |
| from pydantic import BaseModel, ConfigDict | |
| from .enums import SubjectType | |
| class VisualDescriptor(BaseModel): | |
| model_config = ConfigDict(frozen=True) | |
| subject_type: SubjectType | |
| # Free-form, model-authored tags the compositor maps onto its layer library. | |
| palette: str = "noir" | |
| gender: str = "" # "male" / "female" -> gendered sprite features + TTS voice | |
| age_band: str | None = None | |
| build: str | None = None | |
| hair: str | None = None | |
| attire: str | None = None | |
| mood: str | None = None | |
| accent_color: str | None = None | |
| location_tags: tuple[str, ...] = () | |
| prop_tags: tuple[str, ...] = () | |
| prompt_hint: str = "" | |