SoC 2021

This page gives the current list of ideas for VideoLAN project, for the Google Summer Of Code 2021 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.

Introduction & Information

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 university students to have a paid internship by Google to work on open source projects and become top developers!

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

This page lists some ideas for Summer of Code projects on dav1d, VLMC, libVLC and VLC media player, but also on VLC infrastructure projects.

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 and Firefox.

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

VideoLAN Movie Creator

VLMC is a cross-platform non-linear video editing software based on libVLC technology. It was started as a final year student project at the French IT school EPITECH.

VLMC currently awaits a transition to the current libvlc API and it's actual 1.0 release!

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!

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 libVLC or VLMC

This may sound trivial, but it's harder than many expect. 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 merging patches.

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.

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

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.

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

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.

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.

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/tvTV 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 (aka "Marzipan") 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

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.

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. Like for example the network navigation section.

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

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: Soomin Lee, 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

Those will be done in C and C++

Proposed mentor: Thomas Guillem, Steve Lhomme

Develop a MPD server inside VLC

The idea is to add 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

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

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

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

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

Improve VLCKit Objective-C - Swift interoperability

Project description:

Currently VLCKit is not very Swift friendly. Indeed, it will be beneficial for VLCKit to have a better Swift interoperability interface than one generated by default.

Tasks to do:

• Study exposed VLCKit methods
• Improve Objective-C - Swift interop
• Make a test Swift project to test the interoperability

Requirements: This project requires Objective-C and Swift knowledge.

Proposed mentor: Felix Paul Kühne, Soomin Lee

Other 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.

• 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 ?)

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

Proposed mentor: David and j-b

Ideas for VLMC

Port to Android

Project Description: VLMC is now working on the desktop. It would be nice to port it also to the mobile world, starting by Android.

This project would require to adapt to smaller screens, and simplify the UI to fit the Android workflow.

Tasks to do:

1. Fix the VLMC build for Android. As libVLC works fine on Android, this is more focused on compiling the Qt part for Android.
2. Split more parts of the UI in components that are reusable on mobile.
3. Change the components to fit the small resolutions of the Android devices.
4. Rewrite some UI components in QML.
5. Write a simpler timeline widget for Mobile workflows.

Requirements: This project requires C++/Qt/qml knowledge, and access to Android devices.

Proposed mentors: Geoffrey/Hugo

Port to iOS

Project Description: VLMC is now working on the desktop. It would be nice to port it also to the mobile world, also on iOS.

This project would require to adapt to smaller screens, and simplify the UI to fit the iOS workflow.

Tasks to do:

1. Fix the VLMC build for iOS. As libVLC works fine on iOS, this is more focused on compiling the Qt part for iOS.
2. Split more parts of the UI in components that are reusable on mobile.
3. Change the components to fit the small resolutions of the iOS devices.
4. Rewrite some UI components in QML.
5. Write a simpler timeline widget for Mobile workflows.

Requirements: This project requires C++/Qt/qml knowledge, and access to iOS devices. Previous iOS development experience is preferred.

Proposed mentors: Felix/Hugo

Media Workflow rework

Project Description: VLMC is now working on the desktop, but the media workflow is quite slow and is missing some common use cases.

This should be fixed in MLT framework backend for libVLC and is the continuation from the work from the previous Summer of Code.

Tasks to do:

1. Understand the MLT libVLC backend codebase,
2. Clean the backend and port it to a cleaner code state in C++,
3. Benchmark the code and make it more efficient,
4. Work to merge this backend upstream,
5. Implement all the missing features in libVLC.

Requirements: This project requires C/C++ knowledge, and understanding of multimedia concepts is a good positive point.

Proposed mentors: Hugo

Ideas for dav1d

dav1d SSE2/AVX optimizations

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

It requires to:

• Understanding the SSSE3 code of dav1d
• Port it to SSE2
• Write AVX-512 optimizations

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

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

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

Contact 'j-b'

dav1d Threading improvements

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

This project aims to: - improve the threading of AV1, notably to separate the threads for the filters - provide an external threadpool for multiple players - provide a good auto-setup of threads counts, after benchmarking.

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

Requirements: This project requires C and threading 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.

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

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.

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.

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

Proposed mentor: chouquette, tguillem

rav1e, the AV1 encoder in Rust

The rav1e project aims to create a new AV1 encoder implementation in Rust.

To support this we need better AV1 analysis tools that give detailed information about AV1 streams including:

• bit rate across the frame
• frame allocation within a scene
• display of frame data (block partitioning, mode decisions, motion vectors, etc)
• objective quality over time
• average rate (for streaming)

Requirements: This project requires Rust knowledge.

Contact 'unlord'