ffmpeg-render-pro

  ╔══════════════════════════════════════════════════════╗
  ║                                                      ║
  ║   ████████ ████████ ██    ██ ██████  ████████  ████  ║
  ║   ██       ██       ███  ███ ██   ██ ██       ██     ║
  ║   ██████   ██████   ██ ██ ██ ██████  ██████   ██  ██ ║
  ║   ██       ██       ██    ██ ██      ██       ██  ██ ║
  ║   ██       ██       ██    ██ ██      ████████  ████  ║
  ║                                                      ║
  ║   ██████  ████████ ██    ██ ██████  ████████ ██████  ║
  ║   ██   ██ ██       ███   ██ ██   ██ ██       ██   ██ ║
  ║   ██████  ██████   ██ ██ ██ ██   ██ ██████   ██████  ║
  ║   ██   ██ ██       ██  ████ ██   ██ ██       ██   ██ ║
  ║   ██   ██ ████████ ██    ██ ██████  ████████ ██   ██ ║
  ║                                                      ║
  ║        ██████  ██████   ████                         ║
  ║        ██   ██ ██   ██ ██  ██                        ║
  ║        ██████  ██████  ██  ██                        ║
  ║        ██      ██   ██ ██  ██                        ║
  ║        ██      ██   ██  ████                         ║
  ║                                                      ║
  ║  ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░░░░ 8 WRKRS    ║
  ║  GPU: AUTO   DASHBOARD: LIVE   CONCAT: INSTANT       ║
  ╚══════════════════════════════════════════════════════╝

ffmpeg-render-pro

Parallel video rendering with live dashboard, GPU auto-detection, checkpoint system, and stream-copy concat. Includes an MCP server, a Claude Code skill, and a CLI.

Built by Beeswax Pat · Free and open source forever

Features

Parallel rendering: Split frames across N worker threads, concat with zero re-encoding
GPU auto-detection: Probes NVENC, VideoToolbox, AMF, VA-API, QSV with 1-frame validation
Live dashboard: Auto-opens in your browser with per-worker progress, FPS chart, ETA
Checkpoint system: 93% reduction in fast-forward overhead for long renders
Color grading: 5 built-in presets (noir, warm, cool, cinematic, vintage) plus custom filters
Audio merge: Combine video + audio with loudness normalization, no video re-encode
Deterministic output: Seeded RNG ensures parallel workers produce identical results to sequential
MCP server: Model Context Protocol server with 6 tools, works with Claude Code, Claude Desktop, and any MCP client
Cross-platform: Windows, macOS, Linux. Any GPU or CPU-only. Requires Node.js >= 18 plus ffmpeg.

What's new in v1.4.0

A Claude Fable 5 review pass. Fully backward-compatible: every CLI command, API signature, MCP tool name, and worker contract from 1.3.x works unchanged.

GPU detection fixed on modern ffmpeg. The validation probe frame was below NVENC's minimum resolution, so NVIDIA systems silently fell back to CPU encoding. If detect-gpu told you "NO (CPU fallback)" and you have a GPU, run it again on 1.4.0.
Example worker frame generation is 3.1x faster (13.1ms to 4.2ms per 1080p frame), with byte-identical output.
Concurrent renders into the same output directory no longer collide on temp files
mergeAudio always keeps the NEW audio track, even when the video already had one
New CLI flags: --no-dashboard, --no-open, --port, --linger-ms; --seed=0 and fractional --duration now work
colorGrade can keep the soundtrack (keepAudio: true / MCP keep_audio)
New optional MCP params: max_workers, dashboard_port, linger_ms, crf, keep_audio
New end-to-end suite renders real videos and verifies them with ffprobe + framemd5; 49 tests grew to 81

See CHANGELOG.md for the complete list.

Requirements

Node.js >= 18
ffmpeg installed and on PATH

Install

# Global install gives you the ffmpeg-render-pro + ffmpeg-render-pro-mcp binaries
npm install -g ffmpeg-render-pro

# Or clone the repo directly
git clone https://github.com/beeswaxpat/ffmpeg-render-pro.git
cd ffmpeg-render-pro

Quick Start

# System info (workers, RAM, CPU, ffmpeg version)
ffmpeg-render-pro info

# Probe hardware encoders
ffmpeg-render-pro detect-gpu

# 5s benchmark render (dashboard auto-opens at http://127.0.0.1:8080)
ffmpeg-render-pro benchmark

# Longer render, custom resolution
ffmpeg-render-pro benchmark --duration=30 --width=1080 --height=1920 --fps=30

# Force CPU / GPU encoding
ffmpeg-render-pro detect-gpu --cpu
ffmpeg-render-pro detect-gpu --gpu

CLI

ffmpeg-render-pro info                # System snapshot
ffmpeg-render-pro detect-gpu          # Probe hardware encoders
ffmpeg-render-pro render <worker.js>  # Render with your worker script
ffmpeg-render-pro benchmark           # Quick 5s test render
ffmpeg-render-pro version             # Print the installed version

Dashboard control flags for render and benchmark: --no-dashboard (disable entirely), --no-open (serve but don't open a browser), --port=8080, and --linger-ms=30000 (how long the dashboard stays up after completion; 0 exits immediately). Run ffmpeg-render-pro with no arguments for the full flag reference.

API

const {
  renderParallel,    // Core: parallel rendering engine
  createEncoder,     // Pipe raw frames to ffmpeg
  detectGPU,         // Cross-platform GPU detection
  getConfig,         // Auto-tune workers, codec selection
  concatSegments,    // Stream-copy segment joining
  colorGrade,        // Apply color grades (presets or custom)
  mergeAudio,        // Combine video + audio
  startDashboard,    // Live progress dashboard
  saveCheckpoint,    // Checkpoint serialization
  loadCheckpoint,    // Checkpoint restoration
} = require('ffmpeg-render-pro');

renderParallel(options)

The main entry point. Splits a render across workers, shows a live dashboard, and produces a final MP4.

await renderParallel({
  workerScript: './my-worker.js',  // Your frame generator
  outputPath: './output.mp4',
  width: 1920,
  height: 1080,
  fps: 60,
  duration: 60,         // seconds
  title: 'My Render',
  autoOpen: true,       // auto-open dashboard in browser
  maxWorkers: 8,        // cap for auto worker count (override with workerCount)
  dashboardLingerMs: 0, // 0 = resolve immediately; CLI default keeps it up 30s
});

Width and height must be even (the pipeline encodes yuv420p). For library use, set dashboardLingerMs: 0 so the call resolves without holding the process open. renderParallel resolves with { outputPath, elapsed, totalFrames, avgFps }. Set FFMPEG_RENDER_PRO_DEBUG=1 in the environment for full stack traces on error.

Writing a Worker

Workers receive frame ranges via workerData and pipe raw BGRA frames to ffmpeg:

const { workerData, parentPort } = require('worker_threads');
const { spawn } = require('child_process');

const { width, height, fps, startFrame, endFrame, segmentPath, workerId } = workerData;

// Spawn ffmpeg encoder
const ffmpeg = spawn('ffmpeg', [
  '-y', '-f', 'rawvideo', '-pixel_format', 'bgra',
  '-video_size', `${width}x${height}`, '-framerate', String(fps),
  '-i', 'pipe:0',
  '-c:v', 'libx264', '-preset', 'fast', '-crf', '20',
  '-pix_fmt', 'yuv420p', '-movflags', '+faststart',
  segmentPath,
], { stdio: ['pipe', 'pipe', 'pipe'] });

const buffer = Buffer.alloc(width * height * 4);

for (let f = startFrame; f < endFrame; f++) {
  // Fill buffer with your frame data (BGRA format)
  renderMyFrame(f, buffer);

  // Write with backpressure
  const ok = ffmpeg.stdin.write(buffer);
  if (!ok) await new Promise(r => ffmpeg.stdin.once('drain', r));

  // Report progress
  parentPort.postMessage({ type: 'progress', workerId, pct: ..., fps: ..., frame: ..., eta: ... });
}

ffmpeg.stdin.end();
ffmpeg.on('close', () => parentPort.postMessage({ type: 'done', workerId }));

See examples/basic-worker.js for a complete working example.

Worker data (injected via worker_threads workerData): width, height, fps, seed, startFrame, endFrame, segmentPath, workerId, totalFrames, duration, plus anything you pass in renderParallel({ workerData }).

Messages a worker posts to the parent via parentPort.postMessage(...):

Message	When	Fields
`{ type: 'progress' }`	periodically while encoding	`workerId`, `pct`, `fps`, `frame`, `eta`
`{ type: 'fast-forward-start' }`	before replaying state up to `startFrame` (optional)	`workerId`, `frames`
`{ type: 'done' }`	after the segment is fully written (required)	`workerId`
`{ type: 'error' }`	on failure	`workerId`, `error`

Each worker writes its frame range to segmentPath; the renderer stream-copy concats the segments in order.

Post-processing API

Use these directly, or via the CLI and MCP tools. Video is stream-copied where possible, so there is no quality loss.

const { colorGrade, mergeAudio, concatSegments } = require('ffmpeg-render-pro');

// Color grade with a built-in preset (noir, warm, cool, cinematic, vintage)
await colorGrade({ inputPath: 'raw.mp4', outputPath: 'graded.mp4', preset: 'cinematic' });

// ...or a custom ffmpeg -vf filter chain
await colorGrade({ inputPath: 'raw.mp4', outputPath: 'graded.mp4', filter: 'eq=contrast=1.08:saturation=0.9', crf: 18 });

// Keep the soundtrack through the grade (default strips audio)
await colorGrade({ inputPath: 'final.mp4', outputPath: 'graded.mp4', preset: 'noir', keepAudio: true });

// Merge audio: video is stream-copied, audio encoded to AAC. loop + loudnorm are optional.
await mergeAudio({ videoPath: 'graded.mp4', audioPath: 'track.mp3', outputPath: 'final.mp4', bitrate: 320, loop: true, normalize: true });

// Concatenate same-codec, same-resolution segments with stream copy (instant)
await concatSegments(['part-000.mp4', 'part-001.mp4'], 'joined.mp4');

Checkpoints (long renders)

For multi-hour renders, pre-generate state snapshots so each worker replays only the frames since the nearest checkpoint instead of from frame 0.

const { generateCheckpoints, loadCheckpoint, restoreCheckpoint } = require('ffmpeg-render-pro');

// One-time update-only pass: advance your systems and snapshot every `interval` frames
generateCheckpoints({ systems, totalFrames: 432000, fps: 60, checkpointDir: './.checkpoints', interval: 60000 });

// Inside a worker: jump to the nearest snapshot at or below startFrame
const cp = loadCheckpoint('./.checkpoints', startFrame);
if (cp) {
  const resumeFrame = restoreCheckpoint(cp, systems); // returns the snapshot's frame number
  // fast-forward systems from resumeFrame to startFrame, then render
}

systems is an object of named modules, each implementing getState() and setState() (plus update(dt) for generateCheckpoints).

Modules

Module	Purpose
`parallel-renderer`	N-worker thread pool with progress tracking
`encoder`	Raw frame pipe to ffmpeg with backpressure
`gpu-detect`	Cross-platform hardware encoder discovery + validation
`config`	Auto-tune workers based on resolution, RAM, CPU
`concat`	Stream-copy segment joining (instant)
`color-grade`	ffmpeg video filter presets + custom chains
`audio-merge`	Video + audio merge with loudnorm support
`dashboard-server`	Zero-dep HTTP server with auto-open browser
`progress`	Per-worker terminal + JSON progress tracking
`checkpoint`	State serialization for long renders

Benchmarks

Run your own:

node examples/render-test.js --duration=5
node examples/render-test.js --duration=30
node examples/render-test.js --duration=60 --width=1080 --height=1920

Tests

npm test           # smoke suite + MCP stdio handshake + end-to-end renders
npm run test:smoke # smoke suite only
npm run test:mcp   # MCP server handshake only
npm run test:e2e   # real renders verified with ffprobe + framemd5

A zero-dependency suite (81 tests) covering module exports, input validation (including odd-dimension and worker-count math), dashboard path-safety (traversal + null-byte + double-encoding vectors), checkpoint round-trip and corruption fallback, and the MCP server stdio handshake. The e2e suite renders real videos and checks codec, dimensions, exact frame counts, same-seed determinism (framemd5), concat, color grades, and audio merges; it skips itself cleanly on machines without ffmpeg.

MCP Server

ffmpeg-render-pro includes a Model Context Protocol (MCP) server with 6 tools. Works with Claude Code, Claude Desktop, and any MCP client.

Add to Claude Code

# After `npm install -g ffmpeg-render-pro` the MCP binary is on your PATH:
claude mcp add --transport stdio ffmpeg-render-pro -- ffmpeg-render-pro-mcp

# Or without global install (uses npx):
claude mcp add --transport stdio ffmpeg-render-pro -- npx --yes --package=ffmpeg-render-pro ffmpeg-render-pro-mcp

Add to Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "ffmpeg-render-pro": {
      "command": "ffmpeg-render-pro-mcp"
    }
  }
}

Or, if you prefer not to install globally:

{
  "mcpServers": {
    "ffmpeg-render-pro": {
      "command": "npx",
      "args": ["--yes", "--package=ffmpeg-render-pro", "ffmpeg-render-pro-mcp"]
    }
  }
}

MCP Tools

Tool	Description
`detect_gpu`	Probe hardware encoders (NVENC, VideoToolbox, AMF, VA-API, QSV)
`system_info`	Show CPU cores, RAM, recommended workers, ffmpeg version
`render_video`	Parallel render with live dashboard
`color_grade`	Apply presets (noir, warm, cool, cinematic, vintage) or custom filters
`merge_audio`	Combine video + audio with loudness normalization
`concat_videos`	Stream-copy join multiple videos (instant, no re-encode)

Each tool's full input schema (parameter names, types, defaults) is advertised by the server at runtime via the MCP tools/list method, so an agent can introspect it directly. render_video also accepts dashboard, auto_open, max_workers, dashboard_port, and linger_ms for headless or tuned use; color_grade accepts crf and keep_audio.

Claude Code Skill

This repo includes a ready-to-use Claude Code skill. To install it, copy the skill folder into your Claude skills directory:

# macOS / Linux
cp -r .claude/skills/ffmpeg-render-pipeline ~/.claude/skills/

# Windows
xcopy .claude\skills\ffmpeg-render-pipeline %USERPROFILE%\.claude\skills\ffmpeg-render-pipeline\ /E /I

Once installed, Claude Code will automatically use the skill when you ask it to render video or audio with ffmpeg.

Security Notes

Dashboard server binds to 127.0.0.1 only. It is never reachable from other machines on your network.
No telemetry, no phone-home, no CDN loads. Dashboard runs entirely from local files using system fonts.
MCP server is a local-filesystem tool. When wired into an AI agent, it will render, read, and write files anywhere the current user has access. Treat it like any other filesystem-enabled tool: only run it with a trusted agent, and consider restricting the process's working directory if you use it with untrusted prompts.
Stream-copy concat uses temp files under os.tmpdir(). Output paths you pass are still written as-is, so make sure your output path is where you want it.

Changelog

See CHANGELOG.md for the full release history.

License

MIT

Author

Beeswax Pat

Name		Name	Last commit message	Last commit date
Latest commit History 22 Commits
.claude/skills/ffmpeg-render-pipeline		.claude/skills/ffmpeg-render-pipeline
.github/workflows		.github/workflows
bin		bin
examples		examples
src		src
test		test
.gitignore		.gitignore
CHANGELOG.md		CHANGELOG.md
LICENSE		LICENSE
README.md		README.md
glama.json		glama.json
package-lock.json		package-lock.json
package.json		package.json
server.json		server.json

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

ffmpeg-render-pro

Features

What's new in v1.4.0

Requirements

Install

Quick Start

CLI

API

renderParallel(options)

Writing a Worker

Post-processing API

Checkpoints (long renders)

Modules

Benchmarks

Tests

MCP Server

Add to Claude Code

Add to Claude Desktop

MCP Tools

Claude Code Skill

Security Notes

Changelog

License

Author

About

Uh oh!

Releases 1

Packages

Uh oh!

Contributors

Uh oh!

Languages

Folders and files

Latest commit

History

Repository files navigation

ffmpeg-render-pro

Features

What's new in v1.4.0

Requirements

Install

Quick Start

CLI

API

renderParallel(options)

Writing a Worker

Post-processing API

Checkpoints (long renders)

Modules

Benchmarks

Tests

MCP Server

Add to Claude Code

Add to Claude Desktop

MCP Tools

Claude Code Skill

Security Notes

Changelog

License

Author

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 1

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages