datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
krea2-wildcards
Krea 2 Turbo Prompt and Output Catalog
This dataset contains 514 usable English prompts across 63 active categories. Every row keeps the exact prompt, the URL of its generated Krea 2 output, and generation provenance for filtering, retrieval, and prompt reuse.
It is the row-oriented export of the open Krea 2 Wildcards catalog. This is a community dataset, not an official Krea release.
Start here
Browse prompts beside their generated outputs.
Open the native… See the full description on the dataset page: https://huggingface.co/datasets/sjh9714/krea2-wildcards.krea2-hdr-pass-exr
Curated scene-linear / display-linear HDR frames (human-passed, v1)
1170 EXR frames that passed a multi-round human quality review (watermarks,
grain, jitter, degenerate/duplicate content removed) over a 41k-frame pool.
Format
OpenEXR, RGB HALF, scanline, ZIP; Rec.709/D65 chromaticities embedded.
Pixels are relative linear light: each frame group was normalized so the
group's positive-luminance median maps to 0.18 (shared per clip / per movie;
a group-level scale… See the full description on the dataset page: https://huggingface.co/datasets/LAXMAYDAY/krea2-hdr-pass-exr.