Every Clip Pipeline release, in one place.

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

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/