{
 "format_version": 1,
 "split": "train",
 "clips_per_shard": 100,
 "num_clips": 100,
 "num_frames": 32,
 "image_size": [
  512,
  512
 ],
 "rgb_encoding": "JPEG quality 100, 4:4:4",
 "voxels_file": "voxels_train.npz",
 "num_assets": 228,
 "conventions": {
  "camera": "opencv (x right, y down, z forward)",
  "world": "right-handed, Z up",
  "T_cw": "world -> camera, 4x4, row-major",
  "pose": "x_c = R_co @ (s_co * x_o) + t_co, with R_co = R_cw @ R_wo and t_co = R_cw @ t_wo + t_cw",
  "arrays": "row-major, matrices as nested lists",
  "pixel_centers": "integer coordinates",
  "units": "arbitrary scene units, no metric scale",
  "voxels": "bool 64^3 surface occupancy, np.packbits (C order, big-endian bits) of the flattened grid",
  "visible": "uint8, 0 = background, k = object k (1-based, matches objects[k-1])"
 },
 "shards": [
  {
   "name": "train-000000.tar",
   "num_clips": 100,
   "bytes": 942592000,
   "clip_ids": [
    "01000000",
    "01000001",
    "01000002",
    "01000003",
    "01000004",
    "01000005",
    "01000006",
    "01000007",
    "01000008",
    "01000009",
    "01000010",
    "01000011",
    "01000012",
    "01000013",
    "01000014",
    "01000015",
    "01000016",
    "01000017",
    "01000018",
    "01000019",
    "01000020",
    "01000021",
    "01000022",
    "01000023",
    "01000024",
    "01000025",
    "01000026",
    "01000027",
    "01000028",
    "01000029",
    "01000030",
    "01000031",
    "01000032",
    "01000033",
    "01000034",
    "01000035",
    "01000036",
    "01000037",
    "01000038",
    "01000039",
    "01000040",
    "01000041",
    "01000042",
    "01000043",
    "01000044",
    "01000045",
    "01000046",
    "01000047",
    "01000048",
    "01000049",
    "01000050",
    "01000051",
    "01000052",
    "01000053",
    "01000054",
    "01000055",
    "01000056",
    "01000057",
    "01000058",
    "01000059",
    "01000060",
    "01000061",
    "01000062",
    "01000063",
    "01000064",
    "01000065",
    "01000066",
    "01000067",
    "01000068",
    "01000069",
    "01000070",
    "01000071",
    "01000072",
    "01000073",
    "01000074",
    "01000075",
    "01000076",
    "01000077",
    "01000078",
    "01000079",
    "01000080",
    "01000081",
    "01000082",
    "01000083",
    "01000084",
    "01000085",
    "01000086",
    "01000087",
    "01000088",
    "01000089",
    "01000090",
    "01000091",
    "01000092",
    "01000093",
    "01000094",
    "01000095",
    "01000096",
    "01000097",
    "01000098",
    "01000099"
   ]
  }
 ]
}