Dataset Card for Moonworks Lunara Aesthetic Dataset

Sample Images

Dataset Summary

paper: https://arxiv.org/abs/2601.07941

The Lunara Aesthetic Dataset is a curated collection of 2,000 high-quality image–prompt pairs designed for controlled research on prompt grounding, style conditioning, and aesthetic alignment in text-to-image generation.

All images are generated using the Moonworks Lunara, a sub-10B parameter model at inference with a novel diffusion mixture architecture and trained with Moonworks CAT method.

The images are paired with human-refined prompts and structured labels. The dataset prioritizes clarity, consistency, and licensing transparency over scale.

---

Dataset Structure

Each sample contains:

Load from the Hugging Face Hub and save to disk

from datasets import load_dataset

ds = load_dataset("moonworks/lunara-aesthetic")
ds.save_to_disk("./lunara_aesthetic")

print(ds)

Visualize Random Samples (Colab)

The following snippet shows 10 random images from the dataset, with:

• Region | Category | Topic shown above each image
• Prompt text shown below each image

This works directly in Google Colab.

from datasets import load_dataset
import matplotlib.pyplot as plt
import random
import textwrap
import math

# Parameters
n = 10
cols = 5
rows = math.ceil(n / cols)

# Load dataset
ds = load_dataset("moonworks/lunara-aesthetic")

# Randomly sample n items
indices = random.sample(range(len(ds["train"])), n)
samples = ds["train"].select(indices)

# Create figure
fig, axes = plt.subplots(rows, cols, figsize=(22, 5 * rows))
axes = axes.flatten()

for ax, sample in zip(axes, samples):
    image = sample["image"]
    region = sample.get("region", "unknown")
    category = sample.get("category", "unknown")
    topic = sample.get("topic", "unknown")
    prompt = sample.get("prompt", "")

    ax.imshow(image)
    ax.axis("off")

    # Top metadata
    ax.text(
        0.5, 1.08,
        f"{region} | {category} | {topic}",
        transform=ax.transAxes,
        ha="center",
        va="bottom",
        fontsize=9,
        weight="bold",
        clip_on=False
    )

    # Prompt underneath
    wrapped_prompt = "\n".join(textwrap.wrap(prompt, width=45))
    ax.text(
        0.5, -0.22,
        wrapped_prompt,
        transform=ax.transAxes,
        ha="center",
        va="top",
        fontsize=8,
        clip_on=False
    )

# Hide unused axes
for ax in axes[len(samples):]:
    ax.axis("off")

plt.subplots_adjust(top=0.9, bottom=0.05, hspace=0.6, wspace=0.2)
plt.show()

Citation

@misc{wang2026moonworkslunaraaestheticdataset,
      title={Moonworks Lunara Aesthetic Dataset}, 
      author={Yan Wang and M M Sayeef Abdullah and Partho Hassan and Sabit Hassan},
      year={2026},
      eprint={2601.07941},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2601.07941}, 
}

---

---