Add universal video-type clipping flow with type-aware caption UX and non-gaming AI-first selection

[Copilot]

This change executes the roadmap item to support non-gaming source content (podcasts, sports, education, news, etc.) end-to-end. It adds backend and dashboard support for VIDEO_TYPE, introduces type-specific caption style handling, and skips gaming-centric heuristic prefiltering for non-gaming AI providers.

• Problem scope

  • Extend clipping + caption workflows beyond gaming.
  • For non-gaming types, avoid heuristic-first scene selection and start from AI/deep-analysis paths where those heuristics are not meaningful.

• Backend: universal content routing

  • Added VIDEO_TYPE as first-class config/env input.
  • Introduced get_video_type() and should_bypass_heuristic_prefilter() in shorts.py.
  • Updated rank_scenes_with_ai():
    • gaming: preserves existing heuristic + AI candidate path.
    • non-gaming + openai|gemini: bypasses heuristic prefilter, uses timeline/deep-analysis candidates directly.
    • deep-analysis path no longer injects heuristic backups for non-gaming.

• Captioning/TTS behavior for new types

  • Added VIDEO_TYPE → default auto caption style mapping in subtitle_generator.py (e.g. podcast_quote, sports_playbyplay, educational_explainer).
  • Expanded AI caption style guides in ai_providers.py for non-gaming domains (podcast/sports/news/interview/comedy/music/documentary/cooking/fitness/etc.).
  • Extended semantic-tagging context normalization for these categories.
  • Added TTS style-prefix fallback mapping in tts_generator.py so new caption styles resolve to existing voice presets.

• Frontend: Generate page UX

  • Added Video Type selector with persisted env state.
  • Kept gaming experience unchanged (Auto / Gaming / GenZ / Story).
  • For non-gaming types, added Auto / Curated caption mode with curated style sets per video type.

• Docs/config updates

  • Marked roadmap “Universal Video Type Support” as complete.
  • Updated .env.example (VIDEO_TYPE=gaming, CAPTION_STYLE=auto) and README config docs for new controls and style behavior.

# src/shorts.py
def should_bypass_heuristic_prefilter(ai_provider=None, video_type=None) -> bool:
    provider = (ai_provider or os.getenv("AI_PROVIDER", "openai")).strip().lower()
    vtype = (video_type or get_video_type()).strip().lower()
    return provider in {"gemini", "openai"} and vtype != "gaming"

Original prompt

Execute Universal video type support from the @ROADMAP.md file. Make sure you make both backend and frontend. Figure out how to design the frontend for the various video types, as for gamine we have caption styles. What could be caption styles for the various video styles ? Also, for the video types other than gaming, you will have to bypass heuristics mode entirely as it does not make sense. You will have to start from the step after local heuristics analysis for gemini, openai, deep analysis.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.