Live Build
Operator Visible
LOCAL-FIRST | WINDOWS NATIVE | BETA 2
The local-first media pipeline for power users.
Automate intake, encoding, OCR, muxing, diagnostics, and deployment without surrendering your workflow to a cloud service. Vetrra is the deterministic orchestrator for serious Plex libraries, home labs, and operator-owned media systems.
- No spinners. Visible pipeline state at every stage.
- Network optional. Local files, local compute, local control.
- Hardware accelerated. NVENC, QSV, and operator-defined policies.
Active queue
SABNZBD INTAKE
Dune.Part.Two.2160p.Remux
Step 4 / Encode
QSV 78%
Blade.Runner.2049.4K
Step 6 / Mux
Awaiting subtitles
Alien.1979.Directors.Cut
Step 7 / Poster OCR
OCR matched
The.Matrix.1999
Step 8 / Deploy
QC hold
Acceleration
ENCODERS
NVENC
QSV
CPU fallback
OCR vision model
OLLAMA
qwen2.5vl extracting title blocks, subtitle overlays, and poster metadata for final classification.
THE LOCAL-FIRST DOCTRINE
Built for operators who want determinism, not a black box.
Vetrra runs natively on Windows, respects local storage, and treats your hardware as the execution environment. The product is designed for people who want visible state transitions, reproducible output, and control over every dependency in the chain.
No spinners
Queue positions, transcode progress, OCR activity, and final deployment are surfaced as explicit states rather than vague promises.
Network optional
Vetrra is local-first by design. Core processing runs on your machine with your preferred storage layout and your own dependencies.
Ultimate control
Hardware acceleration, subtitle policy, muxing behavior, and deployment rules stay operator-defined instead of platform dictated.
THE 8-STEP PIPELINE
Every stage is visible, constrained, and built around real media tooling.
Vetrra is not generic automation. It is a staged media pipeline that coordinates ingestion, file structure, encode policy, OCR, and final deployment with the same interface language used by the desktop application.
STEP 1
SABNZBD
Intake and search
Capture source media in a structured queue, normalize job metadata, and keep incoming work attached to the rest of the pipeline.
STEP 2
FILESYSTEM
Organize
Build deterministic paths, naming rules, and library structure before any expensive compute begins.
STEP 3
MKVTOOLNIX
MKV processing
Inspect streams, prep track policy, and establish the container state required for clean encode and mux decisions.
STEP 4
FFMPEG
Video encode
Apply hardware-aware transcode rules with explicit NVENC, QSV, and CPU fallback visibility. ETA, bitrate, and queue state remain operator-readable.
NVENC
QSV
CPU
STEP 5
SUBTITLES
Subtitle extraction
Surface language tracks, forced flags, and extraction results instead of hiding subtitle policy in opaque automation.
STEP 6
MUX
Muxing
Merge video, audio, subtitle, and metadata decisions into a deterministic final container with visible health checks.
STEP 7
OCR
Poster handling
Use Ollama vision models to parse poster text, classify assets, and keep gallery metadata attached to the content library.
STEP 8
QC + DEPLOY
Quality control and deploy
Finish with explicit output structure, deploy confirmation, and operator-readable logs instead of a silent file drop.
deploy\movies\Alien (1979)\Alien (1979) [4K HEVC HDR].mkv
deploy\movies\Alien (1979)\poster.jpg
deploy\movies\Alien (1979)\subtitles\eng.forced.srt
qc\report\Alien-1979.jsonOPERATIONAL CLARITY
Stop guessing what your automation is doing.
Vetrra exposes transcode state, dependency failures, and queue health directly in the operator surface. The target user is a systems-minded media curator who wants exact information, not abstract reassurance.
DIAGNOSTICS DRAWER
Track the exact stage, the exact tool, and the exact failure domain.
When a mux job stalls, a subtitle track fails validation, or a transcode falls back to CPU, the interface makes the reason visible enough to act on.
ffmpeg -hwaccel qsv
Healthy
mkvmerge subtitle policy
Review required
poster OCR model
Running
BETA 2 TRANSPARENCY
Radically honest staging beats fake polish.
Vetrra is currently in Beta 2. The core PySide6 engine and Linux headless paths are actively undergoing end-to-end testing. Expect rapid iteration and operational refinement.
Tk to Qt transition complete
PyArmor licensing and Lemon Squeezy planned
Beta 2 rollout preparation