Skip to content

Open source ambient lighting implementation for television sets based on the video and audio streams analysis for Windows, macOS and Linux (x86 and Raspberry Pi). Includes real-time HDR tone mapping and multi-threading for better performance.

License

Notifications You must be signed in to change notification settings

corckie/HyperHDR

This branch is 4 commits ahead of, 233 commits behind awawa-dev/HyperHDR:master.

Folders and files

NameName
Last commit message
Last commit date

Latest commit

48e8926 Β· Feb 5, 2022
Oct 31, 2021
Nov 23, 2021
Nov 25, 2021
Aug 18, 2021
Nov 18, 2021
Dec 30, 2021
Sep 25, 2021
Nov 28, 2021
Feb 5, 2022
Sep 9, 2021
Nov 25, 2021
Oct 25, 2020
Mar 1, 2021
Nov 18, 2021
Aug 18, 2021
Nov 28, 2021
Nov 23, 2021
Aug 18, 2021
Dec 18, 2021
Sep 19, 2021

Repository files navigation

About

Open source ambient lighting implementation for television sets based on the video and audio streams analysis, using performance improvements especially for USB grabbers. Single and multi-threaded video processing optimization for Windows, macOS and Linux x64 & ARM (Raspberry Pi and others) for SDR/HDR streams captured by USB grabbers using pre-calculated LUT tables. Direct support for USB grabbers under Windows 10 (Microsoft Media Foundation), Linux (v4l2) and macOS (AVFoundation). Also software screen grabbers are available now πŸ†•

v17

Download packages & sources

Official releases:
https://github.com/awawa-dev/HyperHDR/releases

Installers of the latest testing version (artifacts available in the latest build):
https://github.com/awawa-dev/HyperHDR/actions

Default LUT table is already included (in the installers) but you can generate your own or import 3dl Autodesk lut table.
For Raspberry Pi you may use prepared SD card images: manual
Default hostname for SD images is HyperHDR so connect to http://hyperhdr:8090/
SSH and SPI are enabled on default.

How to compile

Building HyperHDR from sources πŸ†•

Support and contact

HyperHDR's support section (or https://hyperhdr.blogspot.com/ )

Manuals and guides for building and configuring your own ambient lighting system with HyperHDR:

Official Wiki πŸ†•

Build-log from my SK6812 RGBW system and one thing about calibration

Main features of HyperHDR:

  • Really low CPU usage on SoCs like Raspberry Pi using v4l2 grabbers
  • Support for multithreading that makes Raspberry Pi capable of processing HQ video stream (Rpi 1 & Zero should also benefit from the optimization alone)
  • Built-in LUT table generator
  • HDR/BT2020 color, treshold & gamma correction (LUT)
  • SDR treshold & gamma correction for selected codecs (LUT)
  • Automatic signal detection with smart learning capability for USB grabbers πŸ†•
  • Support for USB grabbers under Windows 10
  • Support for USB grabbers under macOS (x64/M1)
  • Support for software screen grabbers: DirectX11 (Windows), CoreGraphics (macOS), X11 (Linux) πŸ†•
  • Built-in audio visualization effects
  • Optimized multi-instances. You can use for example your TV LED strip and multiple WLED or Philips Hue light sources.
  • Support for WS821x, APA102 (HD107, SK9822 etc) and SK6812 RGBW LED strips using fastest possible cable solution for generic ESP8266/ESP32 external LED drivers: HyperSPI πŸ†•
  • Support for WS8201, WS821x, APA102 (HD107, SK9822 etc) and SK6812 RGBW LED strips ultrafast USB serial port AWA protocol for ESP8266/ESP32 at 2Mb baud with data integrity check and white channel calibration: HyperSerialEsp8266 and HyperSerialESP32 πŸ†•
  • WLED fork for ESP8266 & ESP32 at 2Mb baud (also @921600 baud πŸ†•) and almost all popular types of LED strips is available: HyperSerialWLED

Changelog (v17) πŸ†•

  • Overall performance without tone mapping for USB grabbers improved x10 (MJPEG) and x3 (YUV) over Hyperion NG 2.0.0.8A thanks to optimization & using of multi-threading
  • Direct support for USB grabbers under Windows 10, Linux and macOS (really fast & of course multi-threaded)
  • Support for software screen grabbers: DirectX11, CoreGraphics, X11 πŸ†•
  • User interface upgraded to modern standards (Bootstrap 5) πŸ†•
  • Improved LUT table for SDR(yuv) and HDR video streams πŸ†•
  • Support for CEC (turn ON/OFF grabbers, remote keys to command HDR tone mapping) πŸ†•
  • Support for my new HyperSPI project for Rpi. Fastest possible cable solution for almost every generic ESP8266/ESP32 LED driver πŸ†•
  • Fork of WLED with USB serial port AWA protocol at @2000000 speed (:sparkles: there are also special versions for @921600) for ESP32 & ESP8266 and almost all types of LED strips: HyperSerialWLED
  • Support for WS821x RGB, SK6812 RGBW, APA102 like LED strips using USB serial port AWA protocol for ESP8266 at @2000000 baud with data integrity check: HyperSerialEsp8266 πŸ†•
  • Support for WS821x RGB, SK6812 RGBW, APA102 like LED strips using USB serial port AWA protocol for ESP32 at @2000000 baud with data integrity check: HyperSerialESP32 πŸ†•
  • Automatic signal detection with smart learning capability for USB grabbers πŸ†•
  • Re-implemented backup import / export functions for ALL instances πŸ†•
  • New video stream crop method in JSON API and GET multi-command support πŸ†•
  • Auto-resume option for the USB grabber πŸ†•
  • JSON API documentation in a form of live playground in 'Advanced' tab πŸ†•
  • List of available COM ports for the adalight driver πŸ†•
  • Fix: in specific cases some devices could not react to the 'no video signal' event when it's triggered πŸ†•
  • New installer for Raspberry Pi 3 & 4 64bit OS (AARCH64), faster up to 30% over 32bit OS armv7l version πŸ†•
  • Fix for WLED new network protocol πŸ†•
  • LED grouping aka PC mode aka gradient mode, can help with eye fatigue when used with the monitor, each LED in the group has same average color πŸ†•
  • Add timeout for the anti-flickering filter πŸ†•
  • Panel for easy video resolution & refresh mode selection in the grabber section πŸ†•
  • Support for QT6 πŸ†•
  • Lower CPU usage when automatic signal detection triggers 'no-signal' πŸ†•
  • Fixed power saving issue in macOS version πŸ†•
  • Audio visualization effects (Windows, macOS and Linux)
  • Support for YUV, MJPEG, RGB24, I420, NV12 and XRGB encoding
  • Overall ('Quarter of frame' in the USB grabber section) and per an instance ('Sparse processing' in the Processing tab) options to control quality/performance balance.
  • Hardware brightness, contrast, saturation, hue control for USB grabbers (Windows and Linux)
  • Philips Hue driver (inc. Entertainment API) partially rewritten and working. Customized new options for powering on/off the lamps
  • New option to choose video encoding format (for multi format grabbers for ex. Ezcap 269, MS2109 clones)
  • special LUT table dedicated for Ezcap 320 grabber available in the download section
  • Add configurable Signal Threshold Counter option for signal detection
  • LUT table tone mapping, mainly for HDR correction and fast color space transformation (YUV).
  • New advanced & weighted advanced LED mean color algorithm in Imageβ†’LED mapping
  • Improved backlight algorithm to minimize leds flickering on the dark scenes (configurable in the Smoothing)
  • Add old style color calibration (HSL) using luminance, saturation et.
  • Build for newer Raspbian Buster. It's a complete migration from older Raspbian Stretch
  • Option for hyperhdr-remote, JSON API and web GUI remote to turn on/off HDR tone mapping
  • Option for luminescence & saturation for hyperhdr-remote
  • Ready to write SD images of HyperHDR
  • Fix for SK9822 leds on SPI (aka fake APA102)
  • Windows, macOS DMG and Linux DEB & RPM installers contain default LUT table

FAQ:

You don't need to use an HDR source to use HyperHDR. You can just benefit from: high performance & optimized video proccessing, capability of using multithreading to avoid bottleneck resources, support for modern USB grabbers under Windows 10/Linux/macOS, decicated fast USB LED drivers (HyperSerialEsp8266, HyperSerialESP32 and HyperSerialWLED) or SPI driver (HyperSPI), screen grabbers including DirectX11 with HDR tone mapping, easy to use JSON API (both POST and GET), easy to setup automatic video signal detection, music capabilities and anti-flickering filter for best movie experience.

Use linux 'top' command with per core view (press 1) or preferable 'htop'. On Rpi 2/3/4 max limit is 400% (4 cores per 100%). The problem will occure when one of the core's usage is close to the 100% limit, not when overall usage is for example between 200-300% where each core if far from the individual limit.

Check the performance statistics that are updated every minute in Systemβž”Log page.

Use preferable YUV(YUY2) or XRGB(RGB32) encoding formats if it's possible. They provide better quality and lower CPU usage.

Thanks to our colleague @mjoshd there is a HyperHDR integration plugin for Home Assistant. Check the details on the website of the project https://github.com/mjoshd/hyperhdr-ha πŸš€ πŸ†•

We do not support driving WS281x and especially SK6812 LED strips directly from the Raspberry Pi although it's theoretically possible: link. If you made it and it works, fine, but most of our users weren't so lucky. You should use external ESP8266/ESP32 (preferable with CH340G or CP2104 onboard) and the voltage level shifter. ⚠️

Usage of WS281x LED strip with Rpi directly (PWM mode) requires root privilages. Otherwise you may get 'Error message: mmap() failed' (read more) ⚠️

It's possible to switch off/on HDR tone mapping remotely with home automation system. You can build commands for HyperHDR using our JSON API playground πŸ†•

Before and after HyperHDR LUT correction on HDR/BT2020 video that was broken by the USB grabber.
Without it your ambient lighting colors will be washed-out:
alt text alt text alt text alt text

License

The source is released under MIT-License (see http://opensource.org/licenses/MIT).
GitHub license

About

Open source ambient lighting implementation for television sets based on the video and audio streams analysis for Windows, macOS and Linux (x86 and Raspberry Pi). Includes real-time HDR tone mapping and multi-threading for better performance.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • C++ 63.2%
  • JavaScript 12.1%
  • C 8.8%
  • HTML 7.9%
  • CMake 5.0%
  • Objective-C++ 1.4%
  • Other 1.6%