v1.6.5
What's changed
Internal / performance
- Cached average-frame decode for the space-time analyses (#347).
stroboscope,silhouette_waterfallandspacetime_volumeall reconstruct the same average background frame by decoding the whole video. It is now computed once and cached perMgVideo(keyed by filename, invalidated when it changes), so chained space-time calls reuse it instead of re-decoding. Joins the existing derived-signal caches (quantity-of-motion, audio envelope). - Raw-frame shared decoding is intentionally avoided — caching the decoded frame stack is too memory-heavy for typical videos (>1 GB), so only small derived signals are cached.
Full changelog: https://github.com/fourMs/MGT-python/blob/master/CHANGELOG.md