SoC 2023

From VideoLAN Wiki
Jump to navigation Jump to search

gsoc2016-sun-373x373.png This page gives the current list of ideas for VideoLAN project, for the Google Summer Of Code 2023 program.

The list is long, but it is not exhaustive, and not limitative. Feel free to apply with your own idea!, we love original ideas and value them.

Introduction & Information

Asterix VLC.svg

This page covers the VideoLAN program as a mentoring organization for Google Summer of Code, in order to improve VLC, VLMC and dav1d (or the libVLC engine), but also the VideoLAN infrastructure and some other related projects.

We have projects in C, C++, ASM, JS, Wasm, Go, Obj-C, GPU Shaders, C#, Java/Kotlin and Swift.

Summer of Code

Google Summer of Code is a way for anyone to work on open source projects and become top developers, while being paid by Google.

VideoLAN was a Google Summer of Code mentoring organization in 2007, 2008, 2009, 2010, 2011, 2013, 2016, 2017, 2018, 2019, 2020, 2021 and 2022.

This page lists some ideas for Summer of Code projects on dav1d, libVLC and VLC media player, but also on VLC infrastructure projects.
We accept ideas also on other multimedia projects related to VLC media player.

VLC & libVLC

VLC media player is a cross-platform multimedia player, encoder and streamer application. It is one of the most successful open-source projects worldwide.

VLC media player is downloaded at an approximate monthly rate of 25 millions from the main website and that's not including third-party distributions (Linux)!

You can find more information on VLC on Wikipedia or on this wiki.

The engine of VLC is libVLC, and is used by VLMC and mobile versions of VLC.

dav1d

dav1d is the reference AV1 decoder that is shipped inside VLC, Chrome, Edge, Firefox and Windows.

It's a very low-level codec, mostly written in C and assembly.

Summer of Code rules

If selected and developed, SoC projects for VLMC and VLC will be included in later releases.

All projects are covered by the GPL (v2+) or LGPL (v2.1+) licenses depending on the module.

The VideoLAN Code of Conducts applies to all Summer projects.

How to Start

Find an idea

First, you need to find an idea.

This current page gives you a list of ideas. Those ideas are NOT exhaustive: you can bring your own idea! Some of the best ideas we've ever had were custom ideas!

The duration of the projects is indicative, and can be adapted, if needed.

Submit your idea

You need to submit your idea on the Google Summer of Code platform.

You should do so, very quickly, even before having finished compilation and the next steps, so that we can give you early feedback.

Compile VLC or libVLC

This may sound trivial, but it's harder than many expect. You must compile the project you want to work on.

See https://wiki.videolan.org/Category:Building/ for more informations.

You should come on IRC to get help to compile.

Provide a small patch

To demonstrate your skills, share a small patch with us. This will let you become familiar with Git, in case you don't know it already and our process on our gitlab.

Let's get in touch

If you have a doubt, are not sure about anything or want clarification, please get in touch with us.

We have 3 major communication channels:

  • Our mailing-lists to discuss patches and further development related topics;
  • Furthermore, we have our web forums for VLC-related end-user support - a VLMC section will be created once the product is published.
  • Finally, there is our IRC channel #videolan (for libVLC) and #vlmc on the Freenode network. It's open to any kind of discussion. Usage issues, questions how to compile VLC/VLMC, getting to know the fellow developers, etc.

Ideas for VLC & libVLC

Advanced Audio Filters

Project Description: We are looking for a skilled audiophile that knows a lot about audio theory and practice to work on new audio filters for VLC.

Duration: 350h

Tasks to do:

  • SRS WoW like or other 3D effects;
  • channels mixing, notably upmixing, like Prologic-II;
  • tracks mixing, and transitions;
  • scriptable new audio filters in lua and enable users to create whatever audio filtering function they want in a Lua script;
  • LADSPA or other libraries integration.

Requirements: This project needs some good audio knowledge and good C experience.

Qualification task: Port any audio filter from MPlayer

Proposed mentor: unidan

Unreal Engine libVLC module

The libVLC plugin for Unreal is a bit outdated and needs to be ported to the newer versions of Unreal Engine.

Also, we need to improve performances by using the new libVLC 4 API to share textures.

Requirements: This project needs C experience and understanding a bit Unreal development.

Duration: 350h

Qualification task: Compile the old Unreal module

Proposed mentor: Pierre

Re-implememt the netsync module and use the W3C TimingObject

This module synchronizes more than one VLC instances to play the same media at the same time. Cf. modules/control/netsync.c

This module was disabled when we switched to the new vlc_clock.

Your goal will be to rewrite this module from 0, using the new vlc_clock API.

Tasks to do:

  • Expose the vlc_clock API to control modules
  • Implement the netsync module using the new vlc_clock API
  • Re-use or rewrite the network protocol, depending on its current state (legacy, old code, can it be improved ?)
  • Look at use W3C TimingObject

Duration: 350h

Requirements: Those will be done in C and it requires familiarity with POSIX network APIs (socket, UDP, TCP, bind, accept, listen...)

Proposed mentor: Thomas Guillem

VLC Qt interface redesign

Project Description: The VLC interface is quite outdated on Linux and Windows. It has a lot of features, but some are not properly exposed.

We are currently reworking the interface, but we need help.

Duration: 350h

Scope of the tasks to do:

Use the new designs shared on the mailing list, and help developping part of those, using Qml.

This new interface is simpler, more user friendly, and has a better "media center" feel into it.

It requires integration with the media library and with the current interface.

Qml is the technology needed to improve the current UI.

Requirements: This project requires Qt/C++ knowledge, and qml would be a nice plus.

Proposed mentor: Pierre

Improve chromecast support (subtitles)

Your goal will be to help an engineer who is already working on chromecast improvements.

Tasks to do:

  • Add support for images subtitles (DVD, Bluray, MVK): configure the chromecast module so that it blend the subtitle image video with a video frame
  • Help with text subtitles support
  • Various improvements and bugfixes that are chromecast related (cf. issues from our gitlab)

Duration: 350h

Requirements: Those will be done in C and C++.

Proposed mentor: Thomas Guillem and Alaric Senat

VLC.js

Project Description: We're currently porting VLC to the web, using WebAssembly. This is a current work in progress, so we need a lot of help to improve it

This requires working close to the toolchain, compilation and the web browsers.

Duration: 350h or 175h

Scope of the tasks to do:

There are lots of things to do (not all of them need to be done by the same person:

  • Improving compilation of the VLC core to Wasm,
  • Improving compilation of VLC 3rd party libraries,
  • Fixing the VLC core to support the Web platform,
  • Implement WebCodec inside VLC,
  • Improve audio pipeline inside VLC,
  • Improve emscripten :)

Requirements: This project requires C knowledge, and compilation.

Proposed mentor: JB

VLC macOS interface redesign

Project Description: The VLC interface is quite outdated on macOS and we are currently in the process of re-writing it to give it a modern feel, but also to integrate recent additions to libvlc regarding playback control and library management.

This project for the summer is to rework heavily this interface to make it beautiful and useful again.

Duration: 350h or 175h

Scope of the tasks to do:

There is a full design already done and tested. The major hurdle is to actually implement it the way we want it to be. The iOS/tvOS interface is simpler, more user friendly, and has a better "media center" feel into it, which influenced what we want to achieve on the Mac. Note that the objective is to use AppKit. UIKit will not be part of this project.

Iterating from the current UI and closely collaborating with the team currently working on it is a requirement.

Requirements: This project requires Obj-C knowledge, a thorough understanding of OOP and proven previous Mac development experience. You cannot use swift for this project.

Proposed mentor: David Fuhrmann, Felix Paul Kühne

Improve Android MediaCodec support

Project Description: The goal is to fix few bugs (black screen with some device/video, HDR issues), and to improve the download of MediaCodec surfaces to the CPU

Duration: 175h

Scope of the tasks to do:

Proposed mentor: Thomas Guillem

Rewrite some C modules in Rust

Project Description: The idea is to rewrite a few modules (principally demuxers) in Rust.

Your work will be based on Loic work: https://code.videolan.org/videolan/vlc/-/merge_requests/2738 (Not yet merged)

Scope of the tasks to do:

  • Learn how a C module work and rewrite it in Rust
  • This doesn't concern avi, mp4, mkv, and adaptive demuxers, that are too complicated for a student.

Requirements: Good C knowledge and very good Rust knowledge

Proposed mentor: Thomas Guillem

Add multiple audio tracks playback support

Project Description: Currently, only one audio track can be played at a time, we need to be able to play all tracks from a media in parallel.

Scope of the tasks to do:

  • Modify the audio_output_t API to support multiple streams (Hint, the work already started on the mmdevice/wasapi module)
  • Implement multiple streams support with one API of your choice
  • Adapt the vlc_player_t and LibVLC API for multiple audio streams support
  • Implement it on other APIs

Requirements: Very good C knowledge, familiar with some audio output APIs (PulseAudio, Pipewire, macOS AudioUnit/Hal, Windows Wasapi, Android Audiotrack/aaudio)

Proposed mentor: Thomas Guillem

Scripting VLC with wasm

Project Description: We're would like to try to replace the lua scripting support of VLC with Wasm, like wasmer, directly inside VLC.

This requires working close to the toolchain, VLC and understanding C.

Duration: 350h

Scope of the tasks to do:

There are lots of things to do:

  • Study Wasm
  • Embed wasm inside VLC
  • Map VLC Scripts
  • Improve emscripten and wasm:)

Requirements: This project requires C knowledge, and compilation.

Proposed mentor: tguillem

VLC iOS UI update

Project Description: We're currently in the process of rewriting and updating the entire UI for VLC iOS

There is a lot of components that need refactoring and need to get an updated UI.

The Android port of VLC has done most of that and was successful. We need the same level of features.

Duration: 350h or 175h

Tasks to do:

  • Get an overview of the current App and components that need an update
  • Refactor and give the appropriate components a new look
  • See what is missing compared to the Android version
  • Code it :

Requirements: This project requires Obj-C and Swift knowledge and ideally knowledge of writing tests for iOS but this can be learned.

Proposed mentor: Felix Paul Kühne


Interactive movie support

The idea is to add support for interactive movies inside VLC using the matroska.org mkv format. An interactive movie is like a gamebook but for movies.

Tasks to do:

  • Extend the MKV spec to add support for titles, menus, that will be used by interactive movies
  • Improve a MKV muxing tool (ffmpeg, mkvtoolnix) to support the extended spec
  • Add interactive movies support inside VLC's MKV module

Requirements: Those will be done in C and C++

Duration: 350h

Proposed mentor: Thomas Guillem, Steve Lhomme

Develop a MPD server inside VLC

The idea is to continue adding support for VLC to act as a MPD server.

Tasks to do:

  • Study the MPD protocol
  • Code a MPD server as a control module that will connect to the medialibrary

Duration: 350h

Requirements: Those will be done in C and C++, and it requires familiarity with an MPD client.

The base was done in a previous year, one just needs to finish it.

Proposed mentor: Thomas Guillem

Improve the web interface of VLC in Vue.js

The idea is to improve the new web interface of VLC, that is currently in Vue.JS. It requires significant improvements

Duration: 350h

Tasks to do:

  • Review the current code
  • Clean and improve the modularity
  • Adapt to the new playlist code
  • Connect to the new medialibrary
  • Improve the look

Requirements: Those will be done in JS and C++, and it requires familiarity with the Vue.JS framework Proposed mentor: David

Improve libNDI and integrate in VLC

Project description: Improve the libNDI project supporting the NDI protocol to support more formats.

Duration: 350h

Tasks to do:

  • Study the NDI protocol, implement and test and integrate inside VLC.

Requirements:

  • NDI understanding
  • C knowledge.

Proposed mentor: j-b

Implement DVD-Audio deciphering

Project description: Support DVD-Audio deciphering using dvdcpxm

Duration: 350h

Tasks to do:

  • Understand DVD-Audio
  • Implement VLC module based on those modules

Requirements:

  • Audio likeness
  • C knowledge.

Proposed mentor: j-b

Other short ideas for VLC & libVLC

Those ideas are not detailed, but they are ideas that we could help to spring new ideas. We can help work with you to make those more detailed.

Those ideas should be 175h long

  • Integrate Rust inside VLC, as a demuxer or a parser
  • Improve Vulkan output for VLC, including HDR support
  • Improve id3 tag and metadata handling in VLC
  • Bridge module for GMI'C or other video filters
  • Automated Testing Environment like ffmpeg Fate (port ?) for demuxing, non-hw decoding
  • Integrate libavfilter in VLC
  • Improve the libVLCSharp bindings for VLC in C#
  • Provide setups for popular streaming services / sout templates (ui ?)
  • Improve cue support in VLC

Ideas for VideoLAN infrastructure

Improve the VideoLAN crash reporter in Go and Vue.js

The idea is to improve the current crash reporter of VLC, called CrashDragon.

The tasks are the following:

  • Review the current code
  • Improve the API in Go
  • Write a new Vue.js frontend

Those will be done in Go and JS

Duration: 350h

Proposed mentor: David and j-b

Ideas for dav1d

dav1d RISC-V optimizations

Improving the performance of the AV1 decoder is very important for VLC and the whole ecosystem.

It requires to:

  • Understand of RISC-V assembly
  • Understand a bit what a video decoder is
  • Write RISC-V functions

Requirements: This project requires C and ASM knowledge, as well as system programming skills

Duration: 175h

Contact 'j-b'

dav1d GPU Compute Shaders

Improving the performance of the AV1 decoder is very important for VLC and the whole ecosystem.

This project requires to port one of the filter, like SGR or Wiener to one of the Shader languages. iPhones or Xbox One would be a good target.

This is a tricky project, but is doable during the summer

Duration: 350h

Requirements: This project requires C and GPU Shaders knowledge, as well as system programming skills

Contact 'j-b'

dav1d statistics extractions

The dav1d AV1 decoder is a new high performance AV1 decoder by VideoLAN.

Current open source tools for AV1 analysis use instrumentation in the reference decoder libaom to extract decode-time metadata for display and reporting, but support for sophisticated analysis is lacking.

To speed development of AV1 tools like the rav1e, it would be helpful to add similar decoder metadata extraction APIs to the dav1d decoder so that rapid testing of encoder algorithms is easier. This includes the ability to quickly produce statistics, visualizations and other reporting that can be used for tuning encoder parameters or guiding development. Advanced ideas include adding similar encoder metadata API to rav1e that add encode-time visualizations.

Requirements: This project requires C knowledge.

Duration: 175h

Contact 'unlord'

Ideas for libplacebo

Direct3D 11 backend

Project Description:

libplacebo uses a GPU abstraction with a number of backends. The goal would be to add a new backend based on Direct3D 11, since Vulkan and OpenGL support on Windows are often of limited quality, especially for older hardware.

Lots of example code for how this implementation would look can be found as part of the mpv project .

Large parts can be copy/pasted and adapted to the libplacebo API.

Tasks to do:

  1. Add a new `pl_gpu` backend based on Direct3D 11
  2. Integration into the build system, test framework and CI infrastructure

Requirements:

  • Knowledge of C as well as, ideally, graphics API fundamentals. (But the latter can be learned as part of the project)
  • Ability to develop and test on Windows

Duration: 175h

Contact 'haasn'

Dolby Vision Profile 5 (IPT-PQ)

Project Description:

Dolby IPT-PQ is a HDR color space similar to ITU-R ICtCp, but with proprietary Dolby modifications (reshaping algorithm). Your goal is to implement this reshaper in the form of a GLSL shader, using knowledge from known Dolby patents and dumped headers.

Tasks to do:

  1. Figure out, and (if necessary) reverse engineer the stream format for the Dolby reshaping algorithm described in several of their patents.
  2. Implement this algorithm in GLSL
  3. Integration into libplacebo (optional)
  4. Test against reference implementations of Dolby Vision profile 5

Requirements:

  • Knowledge of GLSL and C. Knowledge of colorspaces in general is an obvious plus, but the theory here is not important - only the implementation.
  • (Possibly) Ability to reverse engineer any still-unknown or differing-from-patents parts of the stream headers.

Knowledge of libplacebo internals is not required, since the skeleton code for this already exists - what's missing is the reshaping algorithm.

Duration: 350h

Contact 'haasn'

GPU motion interpolation (mvtools)

Project Description

Your goal is to develop GPU shaders for motion-adaptive frame interpolation in the style of [mvtools](https://github.com/dubhater/vapoursynth-mvtools).

This is an open-ended project. If not completed, any progress towards this goal is good enough.

Sub-goals:

  1. Recreate the motion vector search algorithms from MAnalyze
  2. Implement the pixel masking and pixel flow algorithms from MFlowFps

These can be tackled and complete out-of-order.

Requirements:

  • Good knowledge of both C and GLSL, especially compute shaders and other GPGPU techniques. (CUDA or OpenCL skills also transfer, though the shader will have to be GLSL)
  • Ideally, general knowledge of video processing techniques (e.g. motion vector search) - at least enough to be able to understand what mvtools code is doing.

Duration: 350h

Contact 'haasn'

Ideas for VLC dependencies

libmicrodns refactoring

Our current mDNS discoverer is working, but is not so respectful of the RFC. Possible improvements include:

  • Device TTL support
  • Device removal detection
  • Better request pacing
  • Delegate socket interactions to the caller
  • Unit testing
  • Fuzzing

Requirements: This project require C knowledge, as well as system programming skills

Duration: 175h

Proposed mentor: tguillem


VideoLAN Google Summer of Code (GSoC/SoC) mentoring projects
2007 • 2008 • 2009 • 2010 • 2011 (GCi 2011 • SOCIS x264 2011) • 2012 • 2013 • 2016 • 2017 • 2018 • 2019 • 2020 • 2021 • 2022 • 2023 • 2024
Retrieved from "https://wiki.videolan.org/index.php?title=SoC_2023&oldid=62190"