Skip to content

Releases: bensblueprints/clip-pipeline

Release list

Clip Pipeline v1.0.1

Clip Pipeline v1.0.1 Latest
Latest

Choose a tag to compare

View all tags
@bensblueprints bensblueprints released this 21 Oct 05:58
v1.0.1
5f4c56c

Clip Pipeline 0.1.1 — GPU caption-engine path fix.

Fixes:

  • VSR (GPU inpaint, Rebuild, Still clipper): the engine no longer looks for a hardcoded developer path. It now honours CLIP_PIPELINE_VSR_DIR, or resolves the install from the engine's own virtualenv, so a user-installed video-subtitle-remover works on Windows and Apple Silicon macOS.
  • Documentation: full user guide with screenshots at https://onetimesuite.com/clip-pipeline/guide/ and a release history synced from this feed at https://onetimesuite.com/clip-pipeline/changelog/.

Same self-contained build as 1.0.0:

  • Bundled ffmpeg (with libass), Node, yt-dlp and a portable faster-whisper engine. No separate Python, ffmpeg or Node install.
  • Local Whisper runs on CPU or an NVIDIA GPU. The model downloads once on first use (default small ~466 MB; tiny ~75 MB, medium ~1.5 GB, large-v3 ~3 GB).
  • Per-step Skip buttons: exclude any step from Run All; manual Run still works.
  • Exports: choose a local export folder in Settings; finished clips are copied there automatically. Google Drive export is coming soon.
  • YouTube Clipper: 1–30 clips per run, 15s / 30s / 60s clip length, captions on/off, caption colour picker and vertical position slider. Qwen-first Groq moment scoring with model fallback and fixed-length fallback.
  • No-AI path: transcript-as-script, local caption timing, built-in system voice, and heuristic clip selection.
  • Heavy optional engines (VSR rebuild/inpaint, ComfyUI image regeneration) stay outside the installer and are skipped automatically when not installed.

Groq API key: required for the full workflow — script rewriting and YouTube Clipper moment scoring. It is free (no card): create one at https://console.groq.com/keys and paste it into Settings → Groq API key. Without it the app falls back to local Whisper, the transcript as the script, the built-in system voice, and fixed-length clips. ElevenLabs (cloud voices) and the YouTube Data API (channel discovery) remain optional.

Bundled ffmpeg is GPL; source and build information at ffmpeg.org.

The macOS build is unsigned. Right-click the app and choose Open the first time.

SHA-256:

  • Windows: b4e296bcd16df0e3f9d202fdd7539571ef9a6fbaaedb836fc625ebf32a7bb08e
  • macOS: d8737f044a8d58881e75a81dc0e17db31a10f2a9ba519ea012f9bf101f179e02
Assets 4
Loading

Clip Pipeline v1.0.0

Clip Pipeline v1.0.0

Choose a tag to compare

View all tags
@bensblueprints bensblueprints released this 19 Jul 11:10
v1.0.0
fb11923

Windows installer and macOS Apple Silicon (arm64) DMG for Clip Pipeline.

Self-contained build:

  • Bundled ffmpeg (with libass), Node, yt-dlp and a portable faster-whisper engine. No separate Python, ffmpeg or Node install.
  • Local Whisper runs on CPU or an NVIDIA GPU. The model downloads once on first use (default small ~466 MB; tiny ~75 MB, medium ~1.5 GB, large-v3 ~3 GB).
  • Per-step Skip buttons: exclude any step from Run All; manual Run still works.
  • Exports: choose a local export folder in Settings; finished clips are copied there automatically. Google Drive export is coming soon.
  • YouTube Clipper:
    • 1–30 clips per run and 15s / 30s / 60s clip length.
    • Burn captions on/off, caption colour picker, and vertical caption position slider.
    • Qwen-first Groq moment scoring with automatic model fallback; fixed-length fallback if AI finds nothing.
  • No-AI path: transcript-as-script, local caption timing, built-in system voice, and heuristic clip selection.
  • Heavy optional engines (VSR rebuild/inpaint, ComfyUI image regeneration) are skipped automatically when not installed.

Groq API key: required for the full workflow — script rewriting and YouTube Clipper moment scoring. It is free (no card): create one at https://console.groq.com/keys and paste it into Settings. Without it the app falls back to local Whisper, the transcript as the script, the built-in system voice, and fixed-length clips. ElevenLabs (cloud voices) and the YouTube Data API (channel discovery) remain optional.

Bundled ffmpeg is GPL; source and build information at ffmpeg.org.

The macOS build is unsigned. Right-click the app and choose Open the first time.

SHA-256:

  • Windows: be3e727d3d00c992feaa5e9c97205ac9f078c13524a556c414000a66fc528c01
  • macOS: 2e3172a5907377f0d954bddf75fe43b34f2d9d5d763e4661346f5dc4764d4289

User guide with screenshots: https://onetimesuite.com/clip-pipeline/guide/
Release history: https://onetimesuite.com/clip-pipeline/changelog/

Loading