Skip to content

GitLab

Last edited by Lubosz Sarnecki
Page history

Google Summer of Code 2020 Ideas

Here is an list of ideas for GSoC 2020.

Implement Virtual Keyboard

A VR keyboard is part of every desktop VR experience. We currently only provide a keyboard via the OpenVR API, which is implemented by SteamVR. Optimally we will be able to run an internal VR keyboard on both APIs, implemented with our frameworks.

Recommended skills: C11, GLib, UX design, Vulkan, OpenXR, X11, Wayland

Difficulty: medium

Loading virtual environments via GLTF

Being able to choose a more interesting background in form of a GLTF scene. This will require work in our Gulkan rendering library, as well as experience in working with 3D scenes and assets. Blender should be used as the primary asset pipeline.

Recommended skills: C11, GLib, Blender, Vulkan, GLTF

Difficulty: medium

Implement wlroots client for xrdesktop

As a Wayland reference implementation we could implement a Weston fork using xrdesktop or a stand alone wlroots compositor, so xrdesktop can be run without Xorg, GNOME or KDE. This will be interesting for stand alone FOSS XR devices.

Recommended skills: C11, GLib, Wayland, Vulkan, OpenGL

Difficulty: hard

Make xrdesktop work with GNOME / KDE on Wayland

xrdesktop should work with Wayland on KDE and GNOME, but needs more work in terms of testing and ironing out the last missing parts. This includes patches to libinputsynth to implement generic synthesis of input in Wayland compositors, as well as the enhancement of the user experience. This might include upstream work. The final result should be tested and usable in KDE and GNOME.

Recommended skills: C11, GLib, Wayland, OpenGL, Vulkan, KDE, GNOME, C++

Difficulty: easy

Improve UX for mouse / keyboard input

A requested feature is to make xrdesktop work with traditional keyboard / mouse input only. This will most likely require more Xorg / Wayland work and will be kind of a hack on GNOME and KDE. This is also required for a stand alone implementation of xrdesktop for usage with the PC.

Recommended skills: C11, GLib, Wayland, X11

Difficulty: medium

Implement scripting by enabling GObject introspection

Enabling JavaScript and Python bindings is basically free in GObject. We already have gobject-introspection in the build system, but it is never compiled and tested. Writing minimal examples in JavaScript an Python for the whole xrdesktop stack will make development easy and attract new contributors. This will require API and runtime.

More information on this can be found in the issue: #3.

Recommended skills: C11, GLib, Python, JavaScript, Vulkan

Difficulty: hard

Make scene renderer multi threaded

To improve performance when running the scene renderer in a window manager, it will need to run on it's own thread. Initial work for this was already done in a Gulkan example, implementing basic Vulkan threading. The project will need to make the scene renderer threaded when used from GNOME Shell or KWin.

Recommended skills: C11, GLib, OpenGL, Vulkan, Threading, Graphics stack, OpenXR

Difficulty: hard

Implement xr_shell_unstable_v1 Wayland protocol for 3D clients

To make xrdesktop support future 3D UI protocol standards, we should implement the xr_shell_unstable_v1 protocol he uses in wxrc to enable stand alone 3D apps on xrdesktop. This would require also rendering improvments in the scene renderer.

https://git.sr.ht/~sircmpwn/wxrc/tree/master/protocol/zxr-shell-unstable-v1.xml

Recommended skills: C11, GLib, Wayland, Vulkan, OpenGL

Difficulty: hard

Extend support for more controllers

With our OpenXR backend we now support a wide range of controllers with very heterogeneous abilities. Implementing profiles and UX improvements for them is part of this project. You will also look deeper into the stack with using custom branches of Monado and OpenHMD, as well as writing patches. It helps if you have diverse controller hardware here.

Recommended skills: C11, GLib, OpenXR, UX

Difficulty: medium