EmotionPlayer 3.1.1 - Stable

EmotionPlayer v3.1.1 Release Notes ✨

What’s New in v3.1.1

• ✅ More robust inference pipeline

  • Fixed handling of short / low‑FPS / “weird” videos (e.g. many clips from stock sites).
  • Stable frame sampling based on real duration, avoiding zero‑frame and division‑by‑zero edge cases.
  • Safer error handling around native DLL calls.

• 🧠 Correct alignment of emotion timeline

  • Timeline emojis (happy/sad) now fully respect the dynamic frame sampling interval (1–6 seconds).
  • UI and models (positiveness, filter, samp) use the same temporal grid, so visual feedback matches real inference points.

• 📊 Redesigned progress window

  • Shows the current video name (with tooltip for full path if needed).
  • Displays the current stage: Positiveness or Filter.
  • Clearly visible percentage over a dark progress bar — no more white text on white background.
  • Progress is driven by a shared native counter and updated from the UI thread for smooth display.

• ⚙️ Safer multithreaded inference

  • Parallel chunk processing over frames with clean separation of indices (no data races).
  • Threads/tasks are tracked via Task completion instead of manual flags.
  • Still uses ~80% of available CPU cores by default for inference 🔥

• 🧩 Stability & crash fixes

  • Fixed MediaElement.NaturalDuration == Automatic crash by checking HasTimeSpan.
  • Defensive checks around missing frames, invalid durations, and native progress edge cases.

---

EmotionPlayer v3.1.1 focuses on making the inference pipeline and UI feedback significantly more reliable, especially on short and non‑standard videos, while keeping the multithreaded performance gains from v3.1.