"""Per-frame vertices of a converted scene, in either layout written by data_generation/movi_to_sim_data.py: per-file /frame_/obj__pos.npy one array per object and frame (T x N files) packed /frames.npy (T, sum_i Nv_i, 3): the same arrays, concatenated per frame /frames_counts.npy (N,): the vertex count of every object Both hold the same float32 values (the stored, unscaled vertices R(q) v_asset + p). The packed layout needs two files per scene instead of ~5,000 (MOVi-B: 1,200 x 2 files instead of 3.8 million), so it fits on any local disk or a network file system. The frame paths /frame_ keep naming the frames in both layouts; with the packed layout they are names only (no folders exist). """ from collections import OrderedDict from pathlib import Path from typing import List import numpy as np _OPEN_LIMIT = 128 # packed scenes kept memory-mapped per process _open = OrderedDict() # scene dir -> (frames memmap, offsets) def is_packed(scene_dir: Path) -> bool: return (Path(scene_dir) / 'frames.npy').is_file() def _packed(scene_dir: Path): key = str(scene_dir) if key in _open: _open.move_to_end(key) return _open[key] frames = np.load(Path(scene_dir) / 'frames.npy', mmap_mode='r') counts = np.load(Path(scene_dir) / 'frames_counts.npy').astype(np.int64) if frames.ndim != 3 or frames.shape[2] != 3 or frames.shape[1] != int(counts.sum()) or (counts <= 0).any(): raise ValueError(f'{scene_dir}: frames.npy {frames.shape} does not hold the vertex counts {counts.tolist()}') entry = (frames, np.concatenate([[0], np.cumsum(counts)])) _open[key] = entry while len(_open) > _OPEN_LIMIT: _open.popitem(last=False) return entry def num_frames(scene_dir: Path) -> int: if is_packed(scene_dir): return int(_packed(scene_dir)[0].shape[0]) return len(frame_dirs(scene_dir)) def frame_dirs(scene_dir: Path) -> List[Path]: """frame_ paths of a scene, t = 0 .. T-1 (per-file layout: the existing folders, sorted).""" scene_dir = Path(scene_dir) if is_packed(scene_dir): return [scene_dir / f'frame_{t:05d}' for t in range(_packed(scene_dir)[0].shape[0])] return sorted(p for p in scene_dir.glob('frame_*') if p.is_dir()) def load_frame(scene_dir: Path, t: int) -> List[np.ndarray]: """The stored vertices (Nv_i, 3) of every object at frame t, float32 (copies for the packed layout).""" scene_dir = Path(scene_dir) if is_packed(scene_dir): frames, off = _packed(scene_dir) if not 0 <= t < frames.shape[0]: raise IndexError(f'{scene_dir}: frame {t} outside 0 .. {frames.shape[0] - 1}') row = np.asarray(frames[t]) return [row[off[i]:off[i + 1]].copy() for i in range(len(off) - 1)] frame_dir = scene_dir / f'frame_{t:05d}' out = [] while (frame_dir / f'obj_{len(out)}_pos.npy').exists(): out.append(np.load(frame_dir / f'obj_{len(out)}_pos.npy')) if not out: raise FileNotFoundError(f'{frame_dir}: no obj__pos.npy') return out