MuxMaster

MuxMaster

MuxMaster (muxm) — a single-command video repacking and encoding utility that handles Dolby Vision, HDR10, audio track selection, subtitle processing, and container muxing so you don’t have to. Built for home media enthusiasts managing Plex, Jellyfin, or Emby libraries from disc rips and downloads. Pick a profile, point it at a file, and get a properly encoded output without memorizing ffmpeg flags.

# Install via Homebrew (macOS)
brew install TheBluWiz/taps/muxm

# Encode for Apple TV Direct Play — that's it
muxm --profile atv-directplay-hq movie.mkv

For Dolby Vision, AV1, or subtitle OCR support, run muxm --install-dependencies first (see Setup Helpers).

💡 Why MuxMaster?

Getting a Blu-ray rip to direct-play correctly on an Apple TV, a Roku, or through Plex — without the server transcoding on the fly — is a surprisingly deep problem. The video might be Dolby Vision Profile 7 that needs conversion to Profile 8.1. The audio might be TrueHD, which your player can’t direct-play, so you need E-AC-3 — or AAC stereo if the target is a phone or an old Roku. PGS subtitle bitmaps are OCR’d to soft text (SRT or mov_text) when the container requires it — only the universal profile burns forced subs, and any profile can opt in via --sub-burn-forced. And the color space metadata needs to survive the whole process.

You can solve all of this with raw ffmpeg, but the command will be 15+ flags long, different for every source file, and you’ll need to inspect the source with ffprobe first to figure out what half those flags should even be. Every new file is a new puzzle.

HandBrake is the go-to GUI for video encoding, and it’s excellent for what it does. But its preset system doesn’t adapt to what’s actually in the file. It can’t detect that your source is already Apple TV-compliant and skip the encode. It doesn’t extract Dolby Vision RPUs, convert between DV profiles, or inject them back into re-encoded video. It won’t selectively OCR your PGS subtitles to SRT when the output container can’t carry bitmaps. And it won’t generate a JSON report of everything it did for your records. HandBrake gives you a good encode; MuxMaster gives you an opinionated pipeline that understands the relationship between your source, your target device, and every stream in the file.

Tdarr solves the batch-processing and automation problem well, especially at library scale. But it requires a server, a database, a web UI, and Node.js — it’s infrastructure. If you want to process a single file, or a handful of files, with precise control over DV handling, audio track selection, and subtitle policy, Tdarr’s plugin system means writing JavaScript to configure what muxm handles with a single --profile flag. Tdarr is a media library manager; MuxMaster is a per-file encoding tool that aims to make every decision correctly so you don’t have to inspect the output.

MuxMaster sits in the gap between “I know ffmpeg well enough to do this manually” and “I need a server-based automation platform.” It’s a single Bash script with only three required packages (ffmpeg, jq, and bc) and optional tooling for Dolby Vision and subtitle OCR. It understands Dolby Vision at the RPU level, and its profile system encodes the tribal knowledge of what actually works on real hardware into repeatable, overridable presets.

Configuration is where the design philosophy comes together. Most CLI tools expect you to read the source code to learn which variables exist, then hand-build a dotfile from scratch. muxm --create-config generates a complete, commented config file pre-seeded with a real profile’s values — you start from a working baseline and customize, not from a blank page. Configs cascade through three tiers (system, user, project) so an encoding team can lock organization defaults in /etc/.muxmrc while individuals override their preferred CRF or audio settings in ~/.muxmrc and specific project directories can pin a streaming profile. And --print-effective-config shows you the fully resolved result of all those layers before you commit to an encode, so you always know exactly what’s about to happen.

---

Table of Contents

• Why MuxMaster?
• Format Profiles
• How It Works
• Installation
  • Upgrading
• Usage
• Configuration
• Additional Features
• FAQ
• License
• Bug Reports
• Contact
• Author
• Changelog

---

🎯 Format Profiles

Profiles are named presets that configure muxm for a specific use case in a single flag. Every setting a profile changes can be individually overridden with CLI flags.

muxm --profile <name> input.mkv

Profile	Goal	Video	Audio	Subtitles	DV
archive	Lossless preservation	Copy (no re-encode)	All tracks, lossless passthrough	Multi-track stream-copy (all types, up to 99)	Preserve
hdr10-hq	Max HDR10 quality	HEVC CRF 17	Lossless passthrough	Single-track per type, soft subs	Strip
atv-directplay-hq	Apple TV Direct Play	HEVC Main10 (copy if compliant)	E-AC-3	Soft forced (mov_text); others as mov_text; PGS→OCR	Convert to P8.1 (auto, if possible)
av1-hq	High-quality AV1 archive	AV1 CRF 28/24 (res-aware), preset 6	Lossless passthrough	Single-track per type, soft subs	Disabled (AV1 pipeline)
streaming-hevc	Modern HEVC streaming	HEVC CRF 20	E-AC-3 448k	Soft forced + full (no SDH); PGS→OCR	Strip
streaming-av1	AV1 streaming	AV1 CRF 30 (≤1080p SDR) / 28 (≥4K or HDR), preset 6	Opus 256k	Soft forced + full (no SDH); PGS→OCR	Strip
animation	Anime/cartoon optimized	HEVC CRF 16, 10-bit	Lossless passthrough	Multi-track stream-copy (up to 6); preserve ASS/SSA	Strip
atv-directplay-animation	Anime for Apple TV Direct Play	HEVC CRF 16, slower, animation psy params	E-AC-3 (lossless transcoded for ATV)	Multi-track; native ASS/SSA; soft forced	Strip
universal	Play anywhere	H.264 SDR (tone-map HDR)	AAC stereo (native preferred, else downmix)	Burn forced; export others as external SRT	Strip
youtube-upload	YouTube upload prep	H.264 High CRF 16, slow	Best-scored track (surround allowed)	Soft forced; export full subs as SRT	Disabled (HDR10 preserved)

archive — Lossless Archival

For collectors who want bit-perfect preservation. Copies video without re-encoding, keeps all matching audio and subtitle tracks via lossless stream-copy, and generates a JSON report with a checksum (SHA-256 by default; use --checksum-algo blake2b for BLAKE2b via b2sum). Always outputs MKV. Commentary and descriptive audio tracks are dropped by default. Language filtering is controlled by AUDIO_LANG_PREF and SUB_LANG_PREF — when empty (the default), all languages pass. Skips processing entirely if the source already matches.

Formerly dv-archival, which is still accepted as a deprecated alias.

muxm --profile archive movie.mkv

hdr10-hq — High Quality HDR10

Strips Dolby Vision layers and re-encodes to clean HDR10 HEVC at CRF 17. Preserves lossless audio (TrueHD, DTS-HD MA, FLAC) via passthrough. MKV output.

muxm --profile hdr10-hq movie.mkv

atv-directplay-hq — Apple TV Direct Play

Targets true Direct Play on Apple TV 4K via Plex: MP4 or MKV container (follows the source container — both Direct Play via Plex/Infuse), HEVC Main10 with DV Profile 8.1 when possible, E-AC-3 surround (with Atmos JOC when present), and soft forced subtitles (mov_text). Copies compliant video without re-encoding. Skips processing if source is already ATV-compliant.

muxm --profile atv-directplay-hq movie.mkv

av1-hq — High-Quality AV1

High-fidelity AV1 encode via SVT-AV1 at preset 6 with a resolution-aware CRF: 28 for ≤1080p SDR (the transparency point) and 24 for ≥4K or HDR sources (the calibrated default; pass --crf 20 for measured HEVC-CRF-18 parity at the cost of larger files). Lossless audio passthrough, MKV container, and SHA-256 checksum enabled by default. Dolby Vision is automatically disabled — the AV1 encode pipeline does not support DV muxing. Intended as a modern space-efficient alternative to hdr10-hq for clients with AV1 decode support.

Requires AV1 encoder support in ffmpeg. Unlike Dolby Vision tools, a missing AV1 encoder is a fatal error — ffmpeg will exit immediately. Homebrew’s standard ffmpeg includes SVT-AV1 (libsvtav1, the default for the AV1 profiles) but not libaom (libaom-av1) or libass. For the complete set, brew install ffmpeg-full (bundles both AV1 encoders + libass + tesseract), or run muxm --install-dependencies, which installs it for you. ffmpeg-full is keg-only, but muxm detects and uses it automatically at runtime.

muxm --profile av1-hq movie.mkv

streaming-hevc — Modern HEVC Streaming

Optimized for Plex, Jellyfin, and Emby on modern clients: Shield, Fire TV, Roku Ultra, smart TVs, and web browsers. HEVC CRF 20 with E-AC-3 surround at streaming-friendly bitrates and soft subtitles. Always outputs MP4. Strips DV and keeps HDR10. Balances quality with file size.

Formerly streaming, which is still accepted as a deprecated alias.

muxm --profile streaming-hevc movie.mkv

streaming-av1 — AV1 Streaming

AV1 streaming encode for modern clients with AV1 decode support (Fire TV Stick 4K Max, Chromecast with Google TV, web browsers). Preset 6, Opus audio at 256k (AUDIO_FORCE_BITRATE="256k"), soft subtitles. Always outputs MP4. Strips DV; HDR10 preserved. The CRF is resolution/HDR-aware: CRF 30 for ≤1080p SDR (visually transparent), dropping to CRF 28 for ≥4K or HDR sources where CRF 30 falls below the perceptual-transparency line (see docs/AV1_CALIBRATION.md); an explicit --crf overrides this. Delivers smaller files than streaming-hevc at equivalent perceptual quality on supported hardware.

Requires AV1 encoder support in ffmpeg — see the note under av1-hq above.

muxm --profile streaming-av1 movie.mkv

animation — Anime & Cartoon Optimized

Tuned for animation content: lower psy-rd/psy-rdoq to avoid ringing on hard cel edges, 10-bit even for SDR to eliminate banding in gradients, and lossless audio passthrough. MKV container preserves styled ASS/SSA subtitles. All matching subtitle tracks — including PGS bitmap streams — are stream-copied from source (up to 6). Chapter markers are preserved.

muxm --profile animation movie.mkv

atv-directplay-animation — Anime for Apple TV Direct Play

Combines the animation-tuned encoder settings of the animation profile with the Apple TV Direct Play constraints of atv-directplay-hq. Encodes at CRF 16 with the slower preset and animation-optimized psychovisual parameters (psy-rd=1.0, psy-rdoq=0.5, aq-mode=3) to suppress ringing on hard cel edges and eliminate banding in gradients. Audio is transcoded to E-AC-3 surround for ATV compatibility — lossless codecs (TrueHD, DTS-HD MA, FLAC) that cannot direct-play on Apple TV are re-encoded rather than passed through. All matching subtitle tracks are stream-copied in multi-track mode, with native ASS/SSA preserved when the output is MKV. Forced subtitles are embedded as soft subs in any container — no burn-in. Container follows the source: MKV in, MKV out; MP4 in, MP4 out.

muxm --profile atv-directplay-animation show.mkv

universal — Universal Compatibility

Plays on everything: old Rokus, mobile devices, web browsers, and non-HDR TVs. Tone-maps HDR to SDR, encodes to H.264, produces a single AAC stereo track (selects the best native stereo track if present; otherwise downmixes the best-scored surround track to stereo), burns forced subtitles, exports others as external SRT, and strips chapters and non-essential metadata. Always outputs MP4.

muxm --profile universal movie.mkv

youtube-upload — YouTube Upload Prep

Optimized for uploading to YouTube. Encodes to H.264 High profile at CRF 16 with the slow preset in an MP4 container. Audio is kept as the best-scored track — surround is allowed and passed through, giving YouTube’s encoder the cleanest source. Forced subtitles are embedded as soft subs (mov_text); all other dialogue subtitles (excluding SDH) are exported as an external .srt file for upload to YouTube Studio. Dolby Vision is disabled; HDR10 metadata is preserved so YouTube can use it during its own re-encode. Strips non-essential metadata; chapter markers are preserved.

Because YouTube re-encodes every upload regardless, the goal is to give YouTube’s encoder the cleanest, highest-quality source possible — not to minimize file size. CRF 16 with slow delivers near-transparent quality that survives YouTube’s compression with noticeably less artifacting than a smaller source would.

muxm --profile youtube-upload movie.mkv

Overriding Profile Defaults

Use CLI flags to customize any profile on the fly:

# Use archive but keep only English and Japanese audio (drops all other languages)
muxm --profile archive --audio-lang-pref eng,jpn movie.mkv

# Use hdr10-hq but with a different CRF and an added AAC stereo fallback track
muxm --profile hdr10-hq --crf 20 --stereo-fallback movie.mkv

# Use universal but keep chapters
muxm --profile universal --keep-chapters movie.mkv

---

🔧 How It Works

Expand to see the full encoding pipeline

When you run muxm, the script executes a multi-stage pipeline that inspects the source file, makes codec and container decisions based on the active profile, processes each stream type independently, and assembles the final output. Here’s what happens under the hood:

1. Source Inspection. muxm calls ffprobe once and caches the full JSON metadata for the source file. Every subsequent decision — codec detection, color space identification, audio channel layout, subtitle format — reads from this cache rather than re-probing.

2. Profile Resolution. Settings are resolved through a layered precedence chain: hardcoded defaults → system config (/etc/.muxmrc) → user config (~/.muxmrc) → project config (./.muxmrc) → profile → CLI flags. Contradictory combinations (like --profile archive --no-dv) trigger warnings but never errors — CLI flags always win.

3. Video Pipeline. The video stage handles the most complexity. It detects Dolby Vision by probing both stream metadata and frame-level side data, then identifies the DV profile (5, 7, or 8) and compatibility ID. For profiles that preserve DV, it extracts the RPU (Reference Processing Unit) via dovi_tool, converts between DV profiles when necessary (e.g., Profile 7 dual-layer → Profile 8.1 single-layer), encodes the base layer with x265, and injects the RPU back into the encoded stream. For non-DV profiles, it detects the source color space (BT.2020 PQ, BT.2020 HLG, or BT.709 SDR), sets the matching x265 or x264 color parameters and pixel format, and applies tone-mapping when the profile targets SDR output.

Profile 5 caveat (best-effort). Profile 5 sources carry an IPT-PQ-c2 (“ICtCp”) base layer rather than a standard BT.2020 image. muxm re-encodes that base layer as-is and converts only the RPU to Profile 8.1 — it does not color-convert the IPT base-layer pixels (no general-purpose tool does this losslessly). The result is therefore best-effort: on Dolby-Vision-capable players it generally maps correctly, but on non-DV players (which ignore the RPU) the base layer may render with incorrect colors. muxm emits a warning when it processes a Profile 5 source; if exact color fidelity matters, keep the original. Profile 7 is unaffected — its base layer is already standard BT.2020.

4. Audio Pipeline. Audio track selection operates in one of two modes. In single-track mode (the default for most profiles), muxm scores each track based on language match, channel count (plus a surround bonus at ≥6ch), codec preference, and bitrate — with configurable weights in .muxmrc (see man muxm for details). Scoring is count-based only, not channel_layout-aware: a discrete 7.1 track and a 5.1.4 Atmos-in-TrueHD track with the same channel count score identically — object-based/height channels aren’t distinguished from a same-count surface layout. Commentary and descriptive tracks are demoted (by AUDIO_SCORE_COMMENTARY_PENALTY) when their title flags them or they carry a comment / visual_impaired / hearing_impaired disposition, so a flagged-but-untitled commentary can’t win the primary slot. The pipeline picks the best available track (honoring --audio-lang-pref) and decides whether to copy it through or transcode it based on the profile’s codec requirements. The universal profile additionally prefers a native stereo track when one exists (AUDIO_PREFER_STEREO) — honoring AUDIO_LANG_PREF, so a wrong-language stereo is skipped in favor of downmixing the preferred-language track — falling back to a downmix of the best-scored surround track otherwise. A second AAC stereo fallback track can be added alongside a surround primary with --stereo-fallback (off by default in every profile). In multi-track mode (AUDIO_MULTI_TRACK=1, enabled by archive), all tracks that pass the language and commentary filters are stream-copied directly from the source — no scoring, no transcoding, no intermediate files. Commentary and descriptive audio tracks can be excluded via AUDIO_KEEP_COMMENTARY. Lossless codecs (TrueHD, DTS-HD MA, FLAC) are passed through untouched when the profile and container support it.

5. Subtitle Pipeline. Subtitles are categorized into forced, full, and SDH tracks. In single-track mode (the default for most profiles), muxm selects one track per type — the first source-order match passing the SUB_LANG_PREF language filter (an inclusion set, not a ranking) — and can exclude SDH tracks. PGS bitmap subtitles are OCR’d to SRT (via pgsrip or sub2srt) when the output container can’t carry them natively. In multi-track mode (SUB_MULTI_TRACK=1, enabled by archive and animation), all tracks that pass the language and type filters are stream-copied from the source — no OCR (plain-text subs are converted to mov_text for MP4/MOV). Because multi-track mode never silently drops a track, a kept stream the target container can’t hold losslessly — a bitmap subtitle (PGS, VobSub), styled ASS/SSA into MP4/MOV, or lossless HD audio (TrueHD, DTS-HD MA) into MP4/MOV — is a hard error before the encode starts, recommending --output-ext mkv; MKV handles all formats. SUB_MAX_TRACKS caps the total number of subtitle streams kept. Text-based subtitles (ASS/SSA) can be preserved in their native format — retaining positioning, fonts, and typesetting — when the output container is MKV (--sub-preserve-format), or converted to plain-text SRT for broader compatibility; this applies in both single-track and multi-track modes. Forced subtitles can be burned into the video stream; other tracks can be embedded or exported as external .srt files. The two modes differ only in selection (single-track ranks and picks one per type, OCR’ing bitmaps when needed; multi-track keeps everything stream-copyable); from there a single shared emitter handles muxing, so disposition, language, title, and codec tagging are identical across both. Embedded subtitles kept as-is — a preserved PGS bitmap, a native ASS/SSA track, or a directly-mappable text stream — are mapped straight from the source stream, which preserves their original timing; OCR’d, format-converted, and external sidecar subtitles are added as separate inputs.

6. Final Mux. All processed streams are assembled into the target container (MP4, MKV, M4V, or MOV) with correct codec tagging, chapter markers, subtitle disposition flags, and metadata. Subtitle inputs, stream maps, and per-stream tagging are produced in one pass by a single emitter (_emit_sub_tracks) shared by both single- and multi-track modes — embedded tracks are mapped from the source input (-map $src:s:N), file tracks (OCR’d, converted, or external) are added as deduplicated -i inputs. For Dolby Vision in MP4, muxm verifies that the dvcC/dvvC container signaling record is present via MP4Box.

7. Verification. The output file is validated with ffprobe to confirm it’s non-empty and parseable. Optionally, a checksum is written (SHA-256 via .sha256, or BLAKE2b via .b2 when --checksum-algo blake2b is used) and a JSON report is generated documenting every decision, warning, and stream mapping from the run.

If any stage fails, muxm logs the failure, cleans up incomplete temp files, and exits with a descriptive error code. The --dry-run flag executes the entire decision pipeline without writing real output, so you can preview exactly what muxm would do before committing to an encode.

---

📦 Installation

Compatibility

muxm requires Bash 4.3+ and runs on macOS (10.15 Catalina or later) and modern Linux distributions (Ubuntu 20.04+, Fedora 33+, Debian 11+, Arch). It is tested primarily on macOS with Homebrew-installed ffmpeg builds. Windows is supported via Docker — the container bundles muxm with its entire toolchain, driven by double-clickable batch helpers.

Hardware acceleration is supported on macOS (Apple VideoToolbox) only. On Linux, --hw-accel is accepted but encoding always runs in software. See docs/HW_ACCEL.md for the full architecture and backend matrix.

Homebrew (recommended — macOS)

brew install TheBluWiz/taps/muxm
muxm --install-completions      # bash/zsh tab completion

This installs muxm with its required Homebrew-managed dependencies (bash 4.3+, ffmpeg, jq) and the man page automatically. bc isn’t part of the formula — it ships with macOS (and most Linux distros) already, which is why it isn’t listed as a Homebrew dependency.

Optional dependencies — for Dolby Vision, subtitle OCR, and subtitle burn-in:

muxm --install-dependencies       # installs everything missing in one pass

--install-dependencies checks whether the ffmpeg in use is a complete build — libass (for --sub-burn-forced) plus both AV1 encoders (libsvtav1 and libaom-av1). If anything is missing, it installs Homebrew’s bottled ffmpeg-full, which bundles all of them. This is non-destructive: ffmpeg-full is keg-only, so it is installed alongside your existing ffmpeg — nothing is uninstalled or replaced. Because keg-only formulae are not put on PATH, muxm detects ffmpeg-full and uses it automatically at runtime (it prints a one-line note when it does); to make ffmpeg resolve to it in your own shell, add the export PATH=…/opt/ffmpeg-full/bin:$PATH line the installer prints.

Or install individually as needed:

brew install dovi_tool            # Dolby Vision RPU handling
brew install gpac                 # DV container signaling (MP4Box)
brew install tesseract            # PGS subtitle OCR engine
brew install ffmpeg-full          # complete ffmpeg: libass (--sub-burn-forced) + both AV1 encoders + tesseract (keg-only; muxm auto-detects it)

Manual Install (macOS/Linux)

git clone https://github.com/TheBluWiz/MuxMaster.git
cd MuxMaster
chmod +x muxm
sudo cp muxm /usr/local/bin/muxm
muxm --setup                     # installs dependencies, man page, and tab completion

Linux (from source)

There’s no Linux package yet — install the dependencies (native packages or Homebrew, both shown below), then follow Manual Install above to put muxm on your PATH.

Native packages (apt / dnf / pacman):

# Core (required) — ffprobe ships inside the ffmpeg package on all three
sudo apt install ffmpeg jq bc            # Debian / Ubuntu
sudo dnf install ffmpeg jq bc            # Fedora / RHEL
sudo pacman -S ffmpeg jq bc              # Arch

# Optional — Dolby Vision muxing, subtitle OCR, subtitle burn-in (MP4Box ships in gpac)
sudo apt install gpac tesseract-ocr               # Debian / Ubuntu
sudo dnf install gpac tesseract                   # Fedora / RHEL
sudo pacman -S gpac tesseract tesseract-data-eng  # Arch
pipx install pgsrip                               # PGS → SRT OCR (all distros)

• dovi_tool (Dolby Vision RPU handling) is not in the Debian, Fedora, or Arch repositories — install the prebuilt binary from its GitHub releases (or the dovi_tool AUR package on Arch). DV handling auto-disables if it is missing.
• libass + AV1: --sub-burn-forced needs an ffmpeg built with libass, and the av1-hq / streaming-av1 profiles need an AV1 encoder (libsvtav1 or libaom-av1). Most distro ffmpeg builds include both — verify with ffmpeg -version (look for --enable-libass and --enable-libsvtav1), or run muxm --install-dependencies, which prints a per-tool advisory (with package names) on any Linux distro it can’t auto-install for.

Homebrew (Linux) — convenience option:

Homebrew runs on Linux too and is the simplest way to get dovi_tool — the one optional dependency the native package managers above don’t carry (homebrew-core ships prebuilt dovi_tool bottles for x86-64 and ARM64), so there’s no GitHub-release or AUR step. It is a heavier footprint than native packages, so use it for convenience, not as a replacement for your distro’s packages.

# 1. Install Homebrew on Linux — see https://brew.sh
# 2. Add brew to your PATH — its installer prints the exact line; also add it to ~/.profile:
eval "$(/home/linuxbrew/.linuxbrew/bin/brew shellenv)"
# 3. Install muxm's dependencies (gpac provides MP4Box; ffmpeg-full bundles ffprobe + libass + AV1):
brew install ffmpeg-full jq bc gpac tesseract dovi_tool   # or: muxm --install-dependencies

pgsrip (the optional PGS-subtitle OCR helper) is a Python tool, so install it with pipx install pgsrip regardless of which route above you choose.

Docker — Windows, or any OS without the toolchain

The docker/ directory packages muxm and every dependency (ffmpeg, dovi_tool, MP4Box, tesseract with full language data, pgsrip) into a single Ubuntu-based image. This is the supported way to run MuxMaster on Windows — no WSL setup, no manual installs, and a double-clickable encode.bat that walks you through each encode:

• Windows (no command-line experience needed): follow the step-by-step Windows guide.
• macOS / Linux: cd docker && ./setup.sh, then ./encode.sh — or drive muxm directly with docker compose run. See docker/README.md.

Encoding inside Docker is CPU-only (no VideoToolbox), so on macOS prefer the Homebrew install above and keep Docker as a fallback.

Dependencies

Required:

• ffmpeg and ffprobe – core encoding and media inspection
• jq – JSON parsing for stream metadata and reporting
• bc – arithmetic evaluation for frame-rate and bitrate calculations

Optional (auto-disabled if missing):

• ffmpeg with libass (ffmpeg-full on Homebrew) – required for subtitle burn-in (--sub-burn-forced); the standard ffmpeg package works for all other features
• ffmpeg with AV1 encoder (libsvtav1 or libaom-av1) – required for av1-hq and streaming-av1 profiles; a missing AV1 encoder is a fatal error (unlike missing DV tools, it cannot be silently disabled). Not included in the standard Homebrew ffmpeg formula.
• dovi_tool – Dolby Vision RPU extraction/injection; DV handling is automatically disabled if missing
• MP4Box (gpac) – DV container-level signaling verification
• tesseract – OCR engine required by pgsrip for PGS bitmap subtitle conversion
• sub2srt or pgsrip – PGS bitmap subtitle OCR to SRT (configurable via --ocr-tool)

Setup Helpers

# First-time setup (manual installs only): dependencies, man page, and tab completion
muxm --setup

# Or run individually:
muxm --install-dependencies
muxm --install-man
muxm --install-completions

# Generate a config file pre-filled with a profile's defaults
muxm --create-config user atv-directplay-hq   # ~/.muxmrc
muxm --create-config project streaming-hevc    # ./.muxmrc

# Remove installed components
muxm --uninstall-man
muxm --uninstall-completions

Upgrading

Homebrew:

brew upgrade muxm

The man page and tab completions are refreshed automatically on every upgrade — no extra steps needed.

Manual installs:

Pull the latest version and re-run setup to update the man page, completions, and dependencies:

cd MuxMaster && git pull
sudo cp muxm /usr/local/bin/muxm
muxm --setup

---

🚀 Usage

muxm [options] <source> [target.mp4]

Arguments

• <source> – Input media file (e.g., movie.mkv)
• [target] – Output file (optional; defaults to <source>.<output-ext>). If the derived output path would overwrite the source file, muxm auto-appends a version number: movie(1).mp4, movie(2).mp4, etc.

Key Flags

Flag	Description
--profile NAME	Apply a format profile (archive, hdr10-hq, atv-directplay-hq, atv-directplay-animation, av1-hq, streaming-hevc, streaming-av1, animation, universal, youtube-upload)
--dry-run	Simulate without writing output
--crf N	Set video CRF value
-p, --preset NAME	x265 encoder preset (e.g., slow, medium)
--video-codec libx265|libx264	Video encoder
--tonemap	Tone-map HDR to SDR
--dv / --no-dv	Enable or disable Dolby Vision handling (overrides profile)
--dv-convert-p81 / --no-dv-convert-p81	Convert DV to Profile 8.1 when native injection fails (default: on)
--audio-lang-pref LANGS	Audio language preference (comma-separated, e.g., eng,jpn)
--audio-force-codec CODEC	Force all audio to a specific codec
--audio-force-bitrate RATE	Override bitrate for all non-lossless audio output (e.g., 256k); overrides codec-specific bitrate variables
--audio-lossless-passthrough	Allow lossless codecs to pass through
--sub-lang-pref LANGS	Subtitle language inclusion filter (comma-separated set; empty keeps all). Not a ranking — in single-track mode the kept track per type is the first match in source order, not the first language listed. (Contrast --audio-lang-pref, which feeds audio scoring.)
--sub-burn-forced	Burn forced subtitles into video
--sub-preserve-format	Keep ASS/SSA subtitles in native format (MKV only; use --no-sub-preserve-format to force SRT conversion)
--sub-preserve-bitmap / --no-sub-preserve-bitmap	Stream-copy PGS bitmap subtitles to MKV without OCR (default: on; no effect for MP4/MOV)
--output-ext mp4|mkv|m4v|mov	Output container
--report-json	Generate a JSON report alongside the output file
--checksum	Write a checksum sidecar for the output (SHA-256 by default)
--checksum-algo sha256|blake2b|auto	Checksum algorithm (auto prefers b2sum, falls back to sha256)
--strip-metadata	Strip non-essential metadata
--skip-if-ideal	Skip processing if source matches target
--replace-source	Replace the original source file (interactive confirmation)
--force-replace-source	Replace the original source file (no prompt; scripting-friendly)
--print-effective-config	Show resolved config after config file imports

Multi-profile encoding: Pass comma-separated profiles to encode the same source once per profile. Output filenames are auto-suffixed with the profile name, and each profile runs sequentially:

muxm --profile youtube-upload,streaming-hevc source.mkv
# → source.youtube-upload.mp4  source.streaming-hevc.mp4

By default each profile uses its own natural container (e.g. archive → MKV). An explicit --output-ext applies to every profile, so the per-profile filename and the actual container always agree:

muxm --output-ext mp4 --profile archive,streaming-hevc source.mkv
# → source.archive.mp4  source.streaming-hevc.mp4   (both real MP4)

Output container inference: When you provide an explicit output filename, muxm infers the container from its extension — output.mkv forces MKV regardless of the profile default.

Run muxm --help for the full flag reference.

---

⚙️ Configuration

muxm has a layered configuration system designed so you set your preferences once and override only when you need to.

Settings are resolved through multiple levels, applied in this order (lowest → highest precedence):

Hardcoded defaults
  → /etc/.muxmrc           (system-wide)
    → ~/.muxmrc            (user defaults)
      → ./.muxmrc          (project-specific)
        → --profile        (format profile)
          → CLI flags      (highest — always wins)

Setting a Default Profile

Add to any .muxmrc file:

# ~/.muxmrc — always use Apple TV Direct Play unless overridden
PROFILE_NAME="atv-directplay-hq"

CLI --profile always overrides a config-file PROFILE_NAME.

Creating a Config File

Use --create-config to generate a .muxmrc file pre-configured for a profile:

muxm --create-config <scope> [profile]

Scope	Path
system	/etc/.muxmrc
user	~/.muxmrc
project	./.muxmrc (current directory)

Pass CLI flags to pre-set specific values:

muxm --create-config user atv-directplay-hq --crf 20 --preset medium

See man muxm for the full list of 40+ supported override flags. Use --force-create-config to overwrite an existing file.

See docs/config_profile.md for a per-profile reference of the exact variables each profile sets — a useful starting point when deciding what to override.

Verifying Effective Configuration

When configs cascade through multiple layers, it’s easy to lose track of what’s actually active. --print-effective-config resolves every layer and shows you the final result before anything is encoded:

# See what the resolved config looks like after all sources merge
muxm --profile hdr10-hq --crf 20 --print-effective-config

Every variable is displayed grouped by section, with the active profile name and the source of each override (CLI, config file, or profile default). No more guessing what a run will do.

---

📋 Additional Features

Beyond profiles and the core encoding pipeline, muxm ships with a set of operational features that make it safer and easier to use in practice:

• Skip-if-Ideal – Before encoding, muxm inspects the source to determine if it already matches the target profile. If it does, the file is linked or copied without re-encoding, saving time and avoiding generation loss. When multi-track audio or subtitle modes are active, the ideal check verifies that every source track would survive the filter — if any would be dropped, the source is not ideal and remuxing proceeds with explicit per-stream maps. Enabled per-profile or via --skip-if-ideal.
• Hardware-Accelerated Encoding (macOS) – On macOS, --hw-accel videotoolbox (or auto) offloads encoding to Apple’s VideoToolbox for dramatically faster encodes, with --hw-accel-quality to tune the quality/speed trade-off. Software encoding stays the cross-platform default. See docs/HW_ACCEL.md for the architecture and docs/VIDEOTOOLBOX_CALIBRATION.md for the per-profile quality calibration.
• Collision Handling – When the derived output filename matches the source (e.g., encoding movie.mp4 with the default .mp4 extension), muxm auto-versions the output to movie(1).mp4, movie(2).mp4, etc. instead of failing. Use --replace-source for interactive in-place replacement or --force-replace-source for scripted workflows.
• Conflict Warnings – Running --profile archive --no-dv doesn’t error out — it warns you that the combination is contradictory and proceeds with your explicit flags taking precedence. Flags that are incompatible with multi-track modes (e.g., --audio-track, --audio-force-codec, --sub-burn-forced) trigger a graceful demotion: multi-track mode drops to single-track mode with an informational note, and the explicit CLI flag wins. The tool trusts you but lets you know when something looks wrong.
• Dry-Run Mode – --dry-run executes the entire decision pipeline (profile resolution, codec detection, DV identification, audio selection) and prints what it would do, without writing any output files.
• JSON Reporting – --report-json generates a machine-readable JSON report alongside the output file, documenting every decision, warning, codec mapping, and stream disposition from the run.
• Checksum Verification – Writes a checksum sidecar for the output to verify integrity after transfer or archival. Defaults to SHA-256 (.sha256); use --checksum-algo blake2b for BLAKE2b (.b2, requires b2sum), or --checksum-algo auto to prefer BLAKE2b when available. Enabled by default for archive; available for other profiles via --checksum.
• External Subtitle Discovery – At startup, muxm automatically scans for sidecar subtitle files alongside the source: .srt, .ass, .ssa, .vtt, .sup, and .idx. Common filename conventions are recognized — movie.en.srt, movie.forced.en.srt, and similar patterns — and discovered tracks are fed into the subtitle pipeline alongside embedded streams, subject to the same language and type filters already in effect. Enabled by default; use --no-ext-subs to opt out, or --ext-subs-dir to point at a different search directory. See man muxm for the full filename convention reference.
• Man Page – A full muxm(1) manual page with complete flag reference, profile documentation, and examples, accessible via man muxm. See Installation → Setup Helpers for setup commands.
• Tab Completion – bash/zsh tab completion for all flags, profiles, presets, and config scopes. Completes media file extensions when providing input files. See Installation → Setup Helpers for setup commands.
• Live Progress Bar – A clean single-line indicator shows percentage complete and elapsed time during encoding, replacing raw ffmpeg output and clearing itself when the encode finishes.
• Smart Disk Space Preflight – Estimates the encoded output size before starting and aborts early if the output or temp volume lacks room, rather than failing mid-encode and leaving a partial file. Use --no-disk-check to disable (or set DISK_CHECK=0 in .muxmrc); see man muxm for how the estimate is calculated.
• Signal Handling – Interrupting a run (e.g. Ctrl-C) triggers a clean shutdown: partial output and temp files are removed and a brief summary is printed before exit. No stray temp files left behind.
• DEBUG=1 Trace Mode – Set DEBUG=1 in your environment (DEBUG=1 muxm --profile streaming-hevc movie.mkv) to log every major decision point — profile resolution, codec detection, DV identification, stream selection — to stderr, making unexpected behavior straightforward to diagnose without modifying the script.

---

❓ FAQ

muxm says it requires Bash 4.3+ but I’m on macOS. macOS ships Bash 3.2 (2007) due to licensing. If you installed via Homebrew (brew install TheBluWiz/taps/muxm), this is handled automatically — the formula rewrites the shebang to use Homebrew’s bash. For manual installs, run brew install bash and make sure /opt/homebrew/bin/bash (Apple Silicon) or /usr/local/bin/bash (Intel) appears before /bin/bash in your $PATH.

Dolby Vision handling seems to be disabled / I don’t see DV in my output. DV processing requires dovi_tool and, for MP4 container signaling, MP4Box (gpac). If either is missing, muxm silently disables DV features rather than failing. Run muxm --install-dependencies to install them, or check with muxm --dry-run — the output will show whether DV was detected and what the pipeline plans to do with it.

My file was copied instead of re-encoded. Is that a bug? Probably not. If the source already matches the target profile (correct codec, container, color space, and audio layout), muxm skips the encode and copies/links the file. This is the Skip-if-Ideal feature — it saves time and avoids generation loss. You’ll see a message indicating the skip. To force a re-encode, pass --no-skip-if-ideal or override a setting (e.g., --crf 18) so the source no longer matches.

Which profile should I use? If you’re playing through Plex on an Apple TV 4K: atv-directplay-hq. If you want broad device compatibility across Plex/Jellyfin/Emby clients: streaming-hevc. If your clients support AV1 decode (Fire TV Stick 4K Max, Chromecast with Google TV, modern browsers) and you want smaller files: streaming-av1. For a high-quality AV1 archive on AV1-capable hardware: av1-hq. If you’re archiving a disc rip and want all audio and subtitle tracks preserved losslessly: archive. If you need it to play on everything including old hardware and phones: universal. For anime or cartoons: animation. For anime or cartoons that need to direct-play on an Apple TV: atv-directplay-animation. For clean HDR10 without DV complexity: hdr10-hq. If you’re uploading to YouTube and want to give YouTube’s encoder the cleanest possible source to re-encode from: youtube-upload. When in doubt, start with --dry-run to preview what a profile will do to your file.

Can I process a batch of files? muxm is a per-file tool by design. For a batch, a simple shell loop works:

for f in *.mkv; do muxm --profile streaming-hevc "$f"; done

For library-scale automation, consider pairing muxm with find, xargs, or a job runner like GNU Parallel.

How do I know what settings are actually active? Run muxm --print-effective-config (with any profile and flags you plan to use). It resolves every config layer — defaults, system/user/project .muxmrc files, profile, and CLI flags — and prints the final result. Nothing is encoded; it just shows you exactly what would happen.

The output file is larger than I expected. CRF-based encoding targets quality, not file size. A visually complex source (grain, fast motion, high detail) will produce a larger file at the same CRF than a clean digital source. You can lower quality slightly with a higher --crf value (e.g., --crf 22 instead of the default), or use a slower --preset (slow or slower) which achieves better compression at the same quality — at the cost of longer encode time.

---

📄 License

MuxMaster is freeware for personal, non-commercial use. Any business, government, or organizational use requires a paid license.

Full license text available in LICENSE.md.

---

🐛 Bug Reports

Found a bug? Please open an issue on GitHub. Include the output of muxm --version, the profile and flags you used, and any relevant log output. A --dry-run dump or --report-json output is especially helpful.

This is a solo project — not taking outside code contributions right now, but bug reports and feedback are always welcome.

---

📬 Contact

If you’re using MuxMaster, I’d love to hear about it — what’s working, what’s not, what workflows you’re using it for. This is a solo project and real-world feedback shapes what gets built next.

• Bug reports → GitHub Issues
• Everything else (feedback, licensing, questions) → [email protected]

---

👤 Author

Jamey Wicklund (@TheBluWiz)

---

📝 Changelog

See CHANGELOG.md for the full release history.