Today's entry is another round-up, but this time focused on a number of recent presentations by Microsoft at Gamefest 2011. These presentations are not directly related to Kinect for Windows SDK, instead focused on Kinect for the XBox 360, but there's still a number of great bits of information for all Kinect developers. The presentations include both the PowerPoint decks and video...
Gamefest 2011 Presentations
Kinect volume data can be exploited to extend and refine user interactions. This talk presents methods of constructing hand and finger representations, and building more direct physical UI interactions. In particular, it will cover the benefits, applications, and challenges of these solutions.
It’s been almost a year since the launch of Kinect and the learning process still continues, but experience has proven to be very important because testing Kinect titles is a challenging task. Adjusting to the experience of using your body as the input device, creating test-cases, adapting to procedures and managing resources in the given conditions are aspects that will be discussed in Ubisoft’s presentation.
Now that Kinect has launched to great success, it's time to look back and reflect on the things that went well and to learn from the things that didn't. This talk will start with a primer about the basics of Kinect Compliance and how we get the best out of our lab space. We then cover deep dives about the hot topics of Identity and hybrid submissions. The Identity section will explain the common pitfalls that tend to complicate compliance and how to avoid them. Finally, the hybrid section will describe our approach to submissions that use a mix of controller and Kinect gameplay.
This talk discusses best practices for dealing with NUI noisy joint position data in Kinect titles. Our goal is to help Kinect developers in this regard by showing how to choose an appropriate filtering technique and fine tune its parameters to match the specific needs of their titles. We discuss many topics, including the noise characteristics, noise propagation, latency versus smoothness tradeoff, and using forecasting to reduce latency. We also discuss the most useful filtering techniques for NUI joint data.
Controller-free gaming affects menu design and in-game controls more than any other part of the game. This is new ground for most developers, and an art—the Xbox team has seen examples of success from across the spectrum. In this talk, we will explore some example paradigms from released games and elsewhere, and describe what we need to see during the concept approval and UI review stages before your title can ship. This information is valuable to Producers and Designers wanting to understand—and to improve upon—current best practices in the crucial area of player feedback.
Do you want to make a premium Kinect title? We will explore experiences in shipped Kinect titles that have frustrated players, and present easy to implement solutions. With these tricks, your title will shine, improving the gamer's experience and feedback about your title.
Can you build a complete gesture detection system for your next Kinect title that can recognize a wide array of gestures, can work with side poses and self-occluded joints, and that doesn't require a large database of prerecorded gestures or training? This talk presents the outcomes from a series of experiments in heuristic-based gesture detection using Kinect, and uses the lessons learned to lay the fundamentals of a practical heuristic-based gesture detection system for your next Kinect game.
Thinking beyond speech itself, how can we create more natural-seeming voice interfaces and new gameplay mechanics? With a realistic approach to what speech recognition technologies allow us to do, and an emphasis on the Kinect hardware, we present several methods of interaction, including vocal and pre-vocal, speaker identification, and other acoustic events.
Update 11/14/2011 @12:00 PM PST:
A bunch more Gamefest 2011 downloads are now available;
Voice offers a powerful natural user input mechanism, offering the possibility of completely intuitive and seemingly omnipotent control to the player. How should speech be designed to best take advantage of this capability, to set up the player for maximum success, and to create magical experiences even when the player’s spoken word isn’t always perfectly understood? In this talk, we discuss not only the technical aspects of implementing speech recognition by using Kinect audio libraries, but also best practices for designing, testing, and tuning grammars for intuitive and successful speech recognition experiences across languages and locales.
The microphone array within the Kinect sensor opens a new axis of input for titles to take advantage of. Kinect also features sophisticated hardware and software processing to support spoken interactions as natural user input. This talk covers relevant sound features of the Kinect hardware and software pipelines, including echo cancellation, speech recognition, chat, sound position tracking, and other areas. Both technical and design implications for supporting verbal engagement are addressed.
We will explore an approach to navigating an open environment with Kinect. Using a walking-in-place paradigm, the system transforms the player's movement into natural looking locomotion on their avatar. Various techniques for detecting and analyzing rhythmic motion are covered, including power calculation, zero-crossing detection, auto-correlation, and Fourier analysis (FFT and DFT).
Understanding of how Kinect works and what it does well is key to making great games that use the hardware and software to the fullest. Join us for an all-inclusive discussion about how the Kinect sensor, skeletal tracking, Identity, and speech recognition work together to create the magic behind Kinect.
Kinect titles often rely on the end product of the Kinect system pipeline—the skeleton. Many Kinect titles, however, enhance that data or improve gameplay and visuals by going back to the source—the depth and color images. These images can be noisy and difficult to work with, and even simple processing can quickly grow expensive. Also, image calculations can be implemented either on the CPU or on the GPU, and the choice of processor has implications for performance and latency. We present a selection of tips for efficiently and effectively manipulating raw image data by using lessons learned from developing the AdvancedSegmentation, DepthVisualizer, and FastUntile samples in the Xbox 360 Development Kit.
Join the Certification Group's Functional Kinect presentation where a Functional Kinect lead will run through the latest developments and updates from the Functional labs. Join us as we delve into new features, test case updates, and common issues and hints/tips about how the Functional labs test your Kinect titles and find the issues that matter most to you. You will gain an insight into the most common Kinect failing issues, and pick up hints and tips to potentially improve your pass rate. You will see how we are evolving the Certification process in Functional to be simpler and easier for publishers and developers by giving you exclusive access to the all-new Kinect Certification training videos. If you expect to submit a title to Kinect Certification or Xbox 360 this is a must see!