Datasets:
metadata
license: other
license_name: fair-1.0.0
license_link: LICENSE
task_categories:
- text-to-image
tags:
- pixel-art
- sprites
- game-assets
- 2d-sprites
- text-to-image
size_categories:
- 100K<n<1M
Alucard Sprites
A curated dataset of pixel art sprites at 128x128 RGBA resolution with text captions, built for training the Alucard text-to-sprite generative model.
Each sprite has been individually normalized (multi-sprite grids split into individual sprites, tight-cropped and centered) and captioned using CLIP zero-shot classification describing sprite type, color, size, and view angle.
Dataset Details
| Property | Value |
|---|---|
| Resolution | 128x128 RGBA |
| Format | Parquet with embedded images |
| Columns | image (PNG bytes), text (caption) |
Usage
With Alucard (training)
pip install git+https://github.com/evilsocket/alucard.git
With HuggingFace datasets
from datasets import load_dataset
ds = load_dataset("evilsocket/alucard-sprites")
sample = ds["train"][0]
print(sample["text"]) # caption
sample["image"].show() # PIL Image
License
Released under the FAIR License (Free for Attribution and Individual Rights) v1.0.0.
- Non-commercial use (personal, educational, research, non-profit) is freely permitted under the terms of the license.
- Commercial use (SaaS, paid apps, any monetization) requires visible attribution to the project and its author. See the license for details.
- Business use (any use by or on behalf of a business entity) requires a signed commercial agreement with the author. Contact
evilsocket@gmail.comfor inquiries.