TAID-AtmosEdit

Atmospheric editing dataset with 10,000 rows. Each seed identifies a scene and p_idx identifies one of its atmospheric conditions.

Columns and numeric spaces

• S, V: 8-bit RGB PNG. Byte values contain linearly quantized values using round(clamp(x, 0, 1) * 255). No sRGB/gamma transfer function is applied.
• D: NumPy .npy bytes (float32, HWC RGB) in linear HDR space [0, 5].
• s_density, s_aerosol, s_ozone: atmospheric parameters as float32.
• seed, p_idx: scene and condition identifiers as int32.

Load diffuse shading with:

import io
import numpy as np
from datasets import load_dataset

row = load_dataset("ShunTatsukawa/TAID-AtmosEdit", split="train")[0]
D = np.load(io.BytesIO(row["D"]), allow_pickle=False)

Stored values are linear, not log-space. Models may compute log differences between atmospheric conditions during training.

License

This dataset is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).