jia0160/opens2v-calibration

OpenS2V calibration manifest

This directory contains a small normalized calibration subset sourced from BestWishYsh/OpenS2V-5M. It is intended to prototype AutoRound diffusion T2V and I2V calibration without requiring the full 11 TB upstream dataset.

Format

output/opens2v_calibration.tsv uses AutoRound's normalized schema:

• id: stable upstream sample identifier
• caption: metadata.face_cap_qwen, falling back to metadata.cap[0]
• image: relative path to a reference frame extracted from the source video

T2V uses id and caption; I2V additionally uses image. output/provenance.json records source paths, scores, frame/crop information, selection rules, and license metadata.

Initial subset

The initial manifest contains 16 samples selected deterministically with seed 42. It follows OpenS2V's published high-quality duration, resolution, and aesthetic rules, adds a conservative motion-score guard, and limits selection to two clips per source video. Reference images are cropped from the midpoint of face_cut to avoid cut-boundary and fade frames.

This first version is intentionally a small local prototype sampled from the start of Videos/total_part1; it is not yet a statistically representative sample of all OpenS2V-5M parts. Before publishing it as AutoRound's permanent default asset, host the manifest and images at a stable public URL and review the sample composition.

Rebuild

Run build_manifest.py with the OpenS2V total_part1.json metadata and a folder containing extracted source videos:

python build_manifest.py \
  --metadata /path/to/total_part1.json \
  --videos /path/to/extracted/videos \
  --output ./output \
  --nsamples 16 \
  --seed 42

The upstream dataset card declares CC-BY-4.0. Preserve attribution and review the upstream terms before redistributing the derived manifest and frames.