Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
video
video

Wan2.2-S2V ODE clips (10k, processed)

10,000 talking-head clips processed for Wan2.2-S2V ODE-pair sampling.

Each clip is 81 frames @ 16 fps, centre-cropped to 832×480, with 16 kHz mono audio (= 21 latent frames / 5.0625 s, the native S2V window). Derived from huangmq_5s_30k (clips longer than 5 s).

Files

  • clips_shard_00.tarclips_shard_04.tar — 2,000 clips each (clips/NNNNNN.mp4).
  • clips_manifest.jsonl — one record per clip ({clip, frames, fps, width, height, dur}).

Use

See the one-click sampler at https://github.com/showmeon/s2v (oneclick_ode_sample_native.sh): it downloads these clips + the native Wan-AI/Wan2.2-S2V-14B, captions each clip with a VLM, builds the S2V manifest, and samples ODE pairs with VAE-decoded previews.

Unpack: for t in clips_shard_*.tar; do tar -xf "$t"; done

Downloads last month
217