Speakers bio and abstracts - GStreamer Conference 2012

Talk Abstract

The GStreamer 1.0 release will be out before this years GStreamer Conference and this talk will provide a final summary of the changes and improvements GStreamer 1.0. The talk will also outline the plans for the 1.0 series going forward.

Talk Abstract

Wayland is a new display server architecture that reuses the Linux low-level graphics stack but makes the compositor the server and relies exclusively on client side rendering. We're approaching the 1.0 release of Wayland, which means that we're ready to freeze the protocol and client side API. Along with the protocol, we're developing a reference server implementation called Weston. In the talk we will take a look at what will be in Wayland and Weston 1.0, with a bit of extra focus on how video and gstreamer is integrated.

Talk Abstract

Video4Linux is a fast-changing subsystem where a lot of work is done to support the complex video hardware of embedded systems. This presentation will give an overview of recent developments and the work that is planned for the near future, in particular the ongoing effort to support zero-copy video pipelines.

Talk Abstract

At NVESD we work with imaging sensors and often have a need to capture, process, analyze, and transmit information from them. We have found that GStreamer is well suited for solving many of our requirements. These include capturing uncompressed video from sources such as Camera Link frame grabbers, tagging frames with highly accurate timestamps, and associating frames with other metadata such as sensor readings and GPS/IMU data. We are also working on the use of the Material Exchange Format (MXF) for storing uncompressed video along with metadata with the goal of standardization by the Motion Imagery Standards Board (MISB). As part of this effort we are implementing support for SMPTE 336M KLV metadata in pipelines. Stephen Burks will discuss our motivations and approach.

Talk Abstract

GStreamer 1.0 brings new features for memory handling, particularly the management of buffers. Now it is possible to implement elements that make use of memory areas that are not accessed directly by CPU, such as the video memory, continuous memory areas and so on.

The purpose of this talk is to show how we can use these new interfaces for developing GStreamer elements, that are tightly integrated with the hardware. In particular, we will show how we implemented a simple video sink for the pandaboard, using the Direct Rendering Management (DRM) interface.

This element, called kmssink, uses many of the new concepts for memory handling, added in GStreamer v1.0, such as allocators, buffer pools, and so on. We will review these concepts and how they were used in the element.

Talk Abstract

On February 2012, Fluendo and Collabora announced a partnership to create a Software Development Kit for GStreamer, with the objective of providing a tested and stable version of GStreamer, with a full development environment to build GStreamer applications in the 3 most important operating systems: Linux, Mac OS X and Windows.

The main goal of this SDK is to break the walls of GStreamer's development in platforms like OS X and Windows, where building and deploying GStreamer wasn't straightforward, providing native installer packages with precompiled binaries, development files and integration with IDE's like Visual Studio and XCode as well as a complete documentation with tutorials and instructions to use the SDK in all the platforms.

In this talk I will present the decisions made to create the GStreamer SDK for all the target platforms, such as the GStreamer components selected for the first version of the SDK, the GUI toolkits included to develop graphical applications, the way we achieved modularization and versioning of the SDK and the packaging tools used to distribute it.

I will also present cerbero, the multi-platform build system created to build and package the SDK, showing how it can be used to compile GStreamer and other projects for different platforms, architectures and distributions. I will finish talking about the several problems encountered while developing the SDK in all the platforms and architectures, trying to show the hard process of getting a cross-platform distribution of GStreamer and why this SDK was needed to make our loved GStreamer the best option for cross-platform multimedia development!

Talk Abstract

While GStreamer already achieves very good performance on various platforms, architectures and use-cases, there is still room for improvement. What satisfies one use-case might not be sufficient for another use-case. Bottlenecks that were not noticeable on a desktop computer becomes a show-stopper on an embedded platform. Latency that was not impacting local file playback can make real-time communication impossible even on a powerful machine.

During this talk Edward Hervey will first go through the various metrics one wants to track: cpu usage, memory usage and latency being the main ones. He will expose how they impact the various use-cases, and how they are tied together.

The hardest part of optimization is actually identifying and understanding where the bottleneck or issue lies. Therefore in a second part, Hervey will go over the available methodologies and tools one can use to profile the various metrics. This will range from those applicable to any project, to those specific to GStreamer. The goal of this second part is to pinpoint what our bottleneck or issues are, how to avoid false-positives and the limitations and advantages of each profiling technique.

Various examples of past and current optimizations will be shown, how they were identified and solved. This will include tools and scripts that were created for that purpose.

While this talk is mostly aimed at developers using GStreamer, it should also prove interesting to other developers, as well as general public wanting to understand profiling and optimization.

Talk Abstract

GSS is a new streaming server application that grew out of work done by Entropy Wave. It is a custom standalone HTTP server written in GLib-style C and based on GStreamer, Libsoup for HTTP processing, Bootstrap for HTML user interface, and JSON-GLib for request processing. Streaming protocols include HTTP, RTSP, various flavors of RTMP, Icecast, HTTP Live Streaming, and WebRTC. Primary streaming formats include WebM, Ogg/Theora/Vorbis, Ogg/Theora/Opus, FLV, and MPEG-TS. As a library, GSS is useful for creating any standalone HTTP server in the GLib ecosystem. This talk will discuss GSS architecture, how to deploy GSS, and how to modify and extend.

Talk Abstract

Our GStreamer on Android port has made quite a lot of progress since last year. So has Android itself. How have we worked around latest platform changes? What's the status of the NDK port. What about replacing stagefright on Ice Cream Sandwich? What are we working on right now? How can you help?

Idea of the talk is to present a detailed sum up of the work made on enabling GStreamer on Android during last year. To both sync with people working with GStreamer on this platform and hopefully gather some community interest and contributions.

Talk Abstract

Pro video requires impeccable reliability, release after release. To prevent regressions, we need a practical way to do continuous integration testing against a rather exhaustive set of real-world video files. We also need to test extensive NLE scenarios with video files from professional cameras. And to make life easy for developers, we need the feedback loop to be as fast as possible.

Novacut is building a GStreamer-based testing tool to do just this, and we're harnessing the cloud to get the compute resources needed run such tests in a sane amount of time (and also not break the bank).

Talk Abstract

While GStreamer 1.0 looms close on the horizon a lot of development is still happening in the 0.10 branch, most of which will migrate accross to 10. Tim-Philipp Müller is the maintainer and release manager for GStreamer 0.10.x. This talk will give you a whirlwind tour of what's new and improved in and around GStreamer over the course of the last year.

The talk includes a high-level overview of all kinds of cool things we've improved and done lately, especially. from an application developer perspective. Topics will range from things like AC-3 passthrough now working, va-api support and transcoding improvements. The talk will also talk a bit about the support plans for GStreamer 0.10.x once GStreamer 1.0 has been released.

Talk Abstract

Gravitational waves are waves of spacetime curvature generated by the movement of masses. They are a prediction of the general theory of relativity, and although their existence has been inferred they have never been directly detected. In the last decade, a world-wide network of gravitational-wave detectors has been developed with the goal of establishing the first direct observation of gravitational waves from the most extreme events in the universe such as the collisions of black holes and neutron stars. Rapid detection and localization of gravitational waves will allow gravitational-wave detectors to be part of a larger transient astronomy community and will provide exciting, new information about our Universe.

Gravitational-wave detection presents significant signal processing challenges, and we have used GStreamer to address many of them. We regularly run GStreamer applications having hundreds of processes with thousands of threads per process spanning tens of thousands of CPU cores. Our experience scaling GStreamer up to this level has been remarkably painless, thanks in large part to the GStreamer community. In this talk, we'll discuss our experiences, and present some of our own wish-list items for future GStreamer development.

Talk Abstract

An update of GStreamer on TI OMAP4+ platforms. In the last year, a new mechanism called dmabuf has been developed at the kernel/driver level for zero copy buffer sharing between different devices (video codecs, gpu and display, camera, etc). Compared to how things worked 2 years ago when the OMAP+GStreamer architecture was last presented, this dmabuf based architecture allows for more commonality between x86 and various other ARM platforms, while at the same time providing greater flexibility. And it is based on infrastructure that is part of the upstream linux kernel rather than custom TI mechanisms. We will cover how we've added support for this in GStreamer 0.10 and how it fits in to GStreamer 1.0, as well as what is still missing.

Talk Abstract

GStreamer is a big and successful project that implements complex formats. We use it to read media streams that sometimes come from untrusted sources.

The size and complexity of GStreamer and all the format implementations it brings means that the presence of security bugs in a media playback pipeline is not totally unlikely. An attacker could forge a stream that would exploit such a bug so that he can execute code of his making in the context where the exploited code was running.

A common way to alleviate this kind of issue is to run the software handling untrusted data in a context where its privileges are very limited, so that a successful exploit of a bug in the software would only grant the attacker execution rights in that limited context. We call such a context a sandbox.

I have run some experiments[1] with setuid-sandbox[2], a stand-alone version of the sandboxing system used by chrome on GNU/Linux, that shows the feasibility of running at least part of a pipeline in a sandbox. In this talk, I will explain this work[3], its limitations and discuss what can be done to improve on it.

Talk Abstract

This talk will look at an implementation of a multi-room entertainment system - allowing synchronised playback of music or movies in multiple rooms simultaneously. Commercial implementations of such systems (squeezebox, Sonos) can be costly - but the basic mechanism is quite simple.

Talk Abstract

GES is a library that allows rapid development of advanced multi-platform audio/video editors using GStreamer. From your smartphone to your high-end server cluster, GES is designed to be the solution to all your needs. This presentation will give you an overview of GES, its history and goals. It will also feature a case stufy of the Pitivi video editor, a major GES user, a general technical overview and how GES concepts match well-established non-linear editing paradigms.

The talk will also cover the current features (why you want to use GES) and the roadmap ahead for the library.

Talk Abstract

ALSA has been deployed as the basic lowlevel sound infrastructure, and is still one of the most actively developed area in Linux. This talk will cover a brief introduction of ALSA, the recent development in both the kernel and user-space sides, and the planned new features in near future.

Talk Abstract

GStreamer provides base class for many common element types, including just recently added base classes for audio/video decoders and encoders. People often ask, if it is a good idea to use a base class for this or that problem and which base class they should use. This talk will introduce them, explains what they're doing and when they should be used, describes the concepts and how they are used and what advantages of their use is. In the end some future ideas for base classes and improvements to the existing base classes will be discussed.

Talk Abstract

Introduction to the powerful logging system provided by GStreamer and how to use it when writing plugins. How to read log output to understand what happens in your pipelines and learn about GStreamer's internals. I will extensively demonstrate my gst-debug-viewer UI, which can easily handle million-line log files and boost your productivity.

Talk Abstract

Opus is a new lossy audio codec, standardized at the IETF by Microsoft, Mozilla, Broadcom, and others. It can replace a large collection of existing codecs for both high-quality archival music storage and low-bitrate, interactive speech, as well as everything inbetween. This talk will cover some of the interesting technical aspects of the codec, its features, and how those features interact with various aspects of gstreamer, hopefully without starting a flamewar about container formats.

Talk Abstract

The Mesa project has been speeding up its development over the last few years catching up to modern OpenGL. This talk will cover some of our recent developments and where our plans take us over the next year. A particular focus will be made on what impact newer OpenGL may have on gstreamer, for GPGPU-style media decode and for hardware-assisted post-processing.

Talk Abstract

Luis de Bethencourt will be hosting a round table with 3 GStreamer Experts: Jan Schmidt, Sebastian Dröge, and Tim P. Müller. This personal, instead of technical, Q&A will be driven by the audience questions to the table about their experience learning, progressing, and contributing in GStreamer. Leading and inspiring by example the experts will talk about the help, tricks, mistakes and bumps each one of them experienced in their road to GStreamer expertise.