1080p animation22.61% less time
Stock prior preset9.7344 s
FFmpeg-v2 native profile7.5339 s
1920×1080 / 30fps · 60s source range · 9 trials
FFmpeg + workflow control / macOS
FFmpeg workflows.
Up to 28.48%
less elapsed time.
Turn media commands into repeatable jobs. Use tuned profiles, a typed CLI and local MCP tools to encode, verify and deliver with a clear result.
Recorded M4 Max native-profile result vs the prior Stock workflow preset, 4/2/2/4 → 8/2/2/8. Quality gates passed. Native engine is an optional source build.
Measured before → afterQuality gates passed
Live action / VFX28.48% less time
Stock prior preset7.5278 s
FFmpeg-v2 native profile5.3840 s
1920×800 / 24fps · 60s source range · 9 trials
22.61% less timeRecorded HD workflow / prior Stock preset
28.48% less timeRecorded live action / prior Stock preset
Eight MCP toolsLocal media, explicit execution
Verified outputsFull decode, hash and safe publication
Why use FFmpeg-v2?
Get the engine.
Keep control of the work.
FFmpeg already gives you powerful codecs and controls. FFmpeg-v2 turns them into a reusable workflow for applications, batches and MCP clients.
Use a measured profile
Eligible long M4 Max software jobs get coordinated encoder, decoder, filter and validation pools. Keep full output checking while reducing workflow time.
Inspect the measured gains →Run jobs with history
Typed plans, a durable queue, idempotency, cancellation, retry and retained partials give your app an execution contract it can inspect.
Build around the CLI →Connect MCP and Python
Expose local media work through eight typed MCP tools or the Python interface. The same planner, worker and verification receipt stay behind each result.
Connect your media tools →From Stock FFmpeg to a managed workflow
More than a command wrapper.
Keep FFmpeg's media engine and add the job control, resource supervision and output acceptance your application needs.
Plan before execution
Typed JSON, exact argv, source/backend identity and a resource estimate.
Recover a batch
Persistent queue, attempt history and explicit cancellation or retry.
Verify before delivery
Metadata, full decode and SHA-256 before staged no-clobber publication.
Supervise owned work
Independent pressure, footprint, disk, deadline and parent checks.
Quality remains part of the job
Save time.
Keep the checks.
Same configured quality target1,800 HD frames and 1,440 live-action frames evaluated against aligned source frames.
VMAF, PSNR-Y, SSIM and output-size gates passed. Full decode, audio/metadata checks and hashing remain in the workflow.
One M4 Max, two source ranges. The gain compares different workflow tuning; Stock supports the same thread controls. Same-tuning tests do not establish a general native-encoder speed advantage. Output sizes increased slightly.
Read quality values and the matched Stock controls →Built for your project
Put it to work.
Prepare audio for transcription, sample frames, automate exports or connect a local media tool through MCP.
Audio and searchPCM extraction, frames
and contact sheets.
and contact sheets.
Editing and exportTranscode, concat, subtitles
and audio mixing.
and audio mixing.
Application jobsQueue, progress, cancel,
retry and receipts.
retry and receipts.
MCP workflowsPlan → submit → start
worker → poll result.
worker → poll result.
git clone https://github.com/developerjillur/ffmpeg-v2.git
cd ffmpeg-v2
python3 -m venv .venv
.venv/bin/python -m pip install --upgrade pip
.venv/bin/python -m pip install -e .
.venv/bin/ffv2 --versionPython 3.9+ · macOS / Apple Silicon · FFmpeg and FFprobe installed separately · GPL-3.0-or-later. Normal jobs can use your existing backend; the quoted native-profile measurements require that tested native setup.
Make media work part of your application.
Start with a local job, inspect its plan and accept a verified result.