Skip to content

Pixel Brite Media Server

A complete library, organized automatically.

Media Server organizes movies, TV and music, keeps Pixel Brite’s apps in sync, and prepares playback for the screen and connection. Choose media folders once; matching and artwork are automatic.

Mac · Windows · Linux · Compatible NAS

Automatic matching & metadata

Artwork and details for movies, TV and music.

Choose folders for movies, TV or music. Media Server identifies the titles through Pixel Brite’s catalog and fills in their artwork and details. Matches are selected automatically, with the option to change a selection or edit its details.

Home on Mac, with a Married… with Children episode in Continue Watching, movies to resume, Up Next, and Sunburst by Bob Seger & the Silver Bullet Band in Now Playing
Media Server identified these files and added the titles, descriptions and artwork. Nothing here was entered by hand.

Movies

Browse posters, read a plot summary and see the release year, genres and cast before choosing a movie.

TV shows

Episodes are matched to their series and grouped into seasons, with episode titles and descriptions.

Music

Artist and album matching brings cover art, track names and release details together for browsing and listening.

New files added to a folder are picked up automatically. Library changes sync to the connected apps.

Adaptive video · ABR & HLS

Video quality that follows the connection.

In Automatic quality, Media Server offers video at several resolutions and bitrates. Adaptive bitrate streaming (ABR) lets the player move between them as bandwidth changes, using a smaller stream when needed and raising quality when the connection improves.

  1. More bandwidthHigher picture quality
  2. Bandwidth dropsA smaller stream
  3. Connection improvesQuality rises again

Preserve the original

Compatible video and audio play without re-encoding. When only the audio needs conversion, the original video can stay intact.

Stream as it’s prepared

HTTP Live Streaming (HLS) delivers a movie in small sections, prepared as the player requests them. You can start watching before the whole movie is converted.

Play on more screens

When playback needs a different format, Media Server converts the video or audio for the device. Resolution and bitrate can adapt to the available connection.

Adaptive+ music

Lossless music that adapts.

Adaptive+ starts at original quality, including lossless. For lossless recordings, Media Server also prepares lower-bitrate streams so the player can use less data when bandwidth falls and return to lossless when it recovers.

Explore music streaming

Hardware-accelerated conversion

Watch HDR movies on SDR screens.

When a screen needs SDR video, Media Server adapts HDR brightness and color while preserving highlight detail. Pixel Prism handles the video processing on the server.

Pixel Brite includes hardware acceleration on Mac, Windows and Linux.

Explore Pixel Prism picture processing See format and HDR support

Audio quality & efficiency

Better audio with less processing.

Pixel Brite uses Apple AAC on Apple devices and NMR AAC on Windows and Linux. NMR uses a model of human hearing to preserve more detail in music and dialogue at the same bitrate.

Less audio CPU time than Jellyfin
Up to40%
Less measured signal error than Plex
Up to57%

Linux audio tests · Ryzen 9 7950X3D

Compare audio quality and processing

DTS-HD MA 5.1 → stereo AAC

Audio processing

Total CPU time · Lower is better

  • Plex54.3 s
  • Jellyfin56.2 s

Audio accuracy

Closer to the original · Higher is better

  • Plex37.4 dB
  • Jellyfin39.3 dB

Eight three-minute audio conversions · 48 kHz stereo AAC at 160 kb/s

About these audio comparisons

Across four audio conversion tests, Pixel Brite used 32–38% less CPU time than the Plex audio encoder and 33–40% less than Jellyfin. The converted audio also had less measured signal error in every test. CPU time measures processing work, rather than elapsed conversion time. Plex’s audio encoder is shown once for both plans; video acceleration is compared separately.

NMR stands for noise-to-mask ratio. Its perceptual analysis considers how audible compression noise would be, then chooses how to encode each part of the signal within the available bitrate.

Audio accuracy is the signal-to-noise ratio of the converted audio against the original, after aligning the signals. Higher values mean less error. The 57% figure is the largest reduction in error power relative to Plex, calculated from those measurements. These are objective signal measurements, not listening-test scores.

Measured October 19, 2026 with Pixel Brite development builds, Plex 1.43.4.10903 and Jellyfin 12.2.20261005-225228 on a Linux virtual machine with 32 logical CPUs and 41.1 GiB RAM. Charts show medians of three runs, with one processing thread per conversion and matching surround downmix levels.

These tests compare audio conversion engines, excluding the server, video, network delivery and GPU. The charts use Pixel Brite’s native AAC encoder on Linux; Apple’s encoder is not measured here. The Apple apps share the audio optimizations, but these PC percentages do not quantify mobile CPU or battery savings.

Full catalog sync

Ready to browse on every Apple device.

The Apple apps sync the full catalog, so searching, sorting and filtering happen on the device, even offline. Media Server keeps library changes in sync and delivers the movie or music when you press play.

Explore the on-device library

Personal profiles

Keep each person’s watch history and listening separate, and choose which libraries they can access.

Explore profiles

Remote access included

Connection links and device approval connect the apps away from home, without port forwarding.

Playback beyond the Apple apps

The included web player brings browsing and playback to computers, phones and tablets, including Android browsers.

Library import

Import from Plex or Jellyfin.

Keep artwork, movie and TV matches, watched status, resume progress, ratings and manual playlists. Importing leaves the original database and media files untouched.

See what you can import
Performance comparisonsExplore library scans, streaming and video conversion, with charts and test conditions.

Library performance

100,000 files cataloged in 5m 5s.

50,000 movies and 50,000 episodes, from Add folders to a browsable catalog. Matching, metadata and cast details included.

Expected matches verified
100,000
Browse response · p95
5.4 ms
Peak server memory
735 MiB

Local APFS folders · Apple M4 Max · 128 GiB RAM · One run

Every file matched its expected title. Browsing and catalog sync continued without errors during the scan.

Development lab test with known-match placeholder files and rate limits disabled. Image downloads excluded.

Test conditions and limits

What was timed

From the Add folders click on a running server until all 100,000 files were matched, saved and ready to browse. File preparation and server startup were excluded.

Matching and metadata transfer alone took 36.45 seconds in a separate run of the same filenames and catalog-service build.

Known-match filenames

Non-playable, 37-byte files used real catalog filenames with prevalidated matches and no catalog IDs. This tests scan speed, not matching accuracy for arbitrary filenames.

What is included

Metadata, cast, credits and artwork URLs. Images load on demand and were excluded. Operating-system and catalog-service caches were not cleared.

Machine and build

Measured October 24, 2026 with a development build on macOS 27.0.1, a 16-core M4 Max, 128 GiB RAM and a local APFS database. The catalog service ran on an Apple M4 over a private network. Normal database durability settings; rate limits disabled.

Browsing and sync

A client requested 20 movies about once per second. Browse p95 is the response time at the 95th percentile. Memory is the server process’s peak resident memory, excluding the operating system, browser and other processes.

Catalog updates received during the scan were checked against the completed catalog, including people and credits. Native app catalog loading and playback were not measured.

Slow-connection check

A separate test matched 2,000 files in 11 seconds with the catalog connection limited to 512 KiB/s per connection and 100 ms added round-trip latency. All expected matches, browsing and catalog sync succeeded. Measured October 23 with an earlier development build.

HLS transmuxing

Eight streams under 512 MB of server RAM.

Eight players received the original 1080p H.264 video and AC-3 audio, repackaged for streaming without re-encoding.

Server RAM

Mac mini M4 · 16 GB installed · Dashed line: 512 MB

  • Before playback255 MB
  • Eight streams · median349 MB
  • Eight streams · peak398 MB

CPU during the same test

Before playback
0.17%
Eight streams
2.6%

Mean usage across the Mac’s 10 CPU cores.

The server also prepared an AAC audio version for compatibility; that work is included.

Three 60-second runs with eight independent clients. RAM includes the server, web interface, streaming connections, menu app and media workers; excludes macOS and playback devices. Peak includes startup. Music analysis was off.

Test conditions and playback results

Direct play, transmuxing and transcoding used different files and bitrates. These results measure each workload, so they cannot compare the speed or overhead of the three modes. Caching can make idle RAM higher than playback RAM.

Measured October 22, 2026 with Media Server 2026.10.22, macOS 15.7.7 and 16 GiB RAM. Eight independently enrolled clients retrieved streams without decoding or rendering them. These are server memory measurements, not minimum system requirements.

RAM is median server memory use; CPU is average usage across the Mac’s 10 cores. Each idle/playback pair comes from the run with the middle playback RAM result. Peak is the highest sample across all three runs, including startup. The server and its helper processes were sampled every half second.

Direct play

Server RAM

1080p AV1 + Opus

  • Before playback288 MB

Peak RAM: 384 MB
Average CPU: 3.9%

Repackage for streaming

Server RAM

1080p H.264 + AC-3

  • Before playback255 MB

Peak RAM: 398 MB
Average CPU: 2.6%

Video conversion

Server RAM

1080p HEVC + DTS

  • Before playback300 MB

Peak RAM: 989 MB
Average CPU: 4.9%

Video conversion compared

Less RAM for eight simultaneous conversions.

On the same Mac mini M4, Pixel Brite used less RAM than Plex @ $749 and Jellyfin while converting eight streams. All three kept up with playback.

1080p video

HEVC video + DTS audio → H.264 + stereo AAC

57% less RAM than Plex @ $749

  • Pixel Brite0.69 GB
  • Plex @ $7491.61 GB
  • Jellyfin1.71 GB

4K HDR video

4K HDR + TrueHD audio → 1080p SDR + stereo AAC

25% less RAM than Plex @ $749

  • Pixel Brite2.00 GB
  • Plex @ $7492.66 GB
  • Jellyfin2.75 GB

Eight streams · Three 3-minute runs per test · Hardware acceleration · Local HTTP delivery

RAM includes each server and its helper processes; Pixel Brite’s separate menu app is excluded. Each server used its own encoding and tone-mapping settings. Picture quality was not matched.

Test conditions and measurements

Measured October 22, 2026 on a Mac mini M4 with 16 GB RAM and macOS 15.7.7. Pixel Brite used an October 22 development build; Plex @ $749 used version 1.43.5.11029 and Jellyfin used version 12.2.0.

1080p: All three servers read the same 1920 × 1040 HEVC SDR source with DTS audio and produced H.264 video at the same resolution with stereo AAC.

4K HDR: All three servers read the same 3840 × 1600 HDR10 HEVC source with TrueHD audio and produced 1920 × 800 SDR H.264 with stereo AAC.

Hardware processing was verified. Both tests requested H.264 at a nominal 8 Mb/s with stereo AAC.

Figures are medians of three run medians. Streams started at different positions with a five-second playback buffer. Shared macOS video services are excluded. Maximum stream capacity was not measured.

PC server performance

Less RAM for eight video conversions.

Pixel Brite, Jellyfin and both Plex plans converted the same 1080p HEVC movie to H.264. All four configurations kept up with eight streams.

Ryzen 9 7950X3D + RTX 4090 · Linux

21%less RAM than Plex @ $749

Server RAM

Median during playback

  • Jellyfin1.83 GB
  • Plex FreeSoftware conversion7.52 GB
  • Plex @ $749Hardware acceleration2.19 GB

CPU use

Share of all 32 logical CPUs

  • Jellyfin1.7%
  • Plex FreeSoftware conversion42.7%
  • Plex @ $749Hardware acceleration2%

Eight simultaneous streams · 1080p HEVC → H.264 + stereo AAC · Lower is better. Encoding settings differed; picture quality was not matched.

About this comparison

Measured October 2026 on an Ubuntu virtual machine with 32 logical CPUs, 41.1 GiB RAM and an RTX 4090. Pixel Brite development build; Plex 1.43.4.10903; Jellyfin 12.2.20261005-225228.

RAM includes the server and its helper processes, with shared memory apportioned between processes. Graphics memory and operating-system services are excluded. CPU is average usage as a share of the virtual machine’s 32 logical CPUs.

Each server delivered three minutes per stream over local HTTP without rendering on a player. Hardware acceleration was enabled except for Plex Free, which used software conversion. Requested video and audio targets were 8 Mb/s and 160 kb/s; actual bitrates, encoder presets and buffering differed. Picture quality was not matched.