Summaries for week 9

DART: A Toolkit for Rapid Design Exploration of Augmented Reality Experiences

While AR is being more and more studied and developed, a recent approach is to think of it as a medium and no more as a pure technology.Developed with designers, DART (Designer’s Augmented Reality Toolkit) is a tool built upon Macromedia Director aimed at being usable efficiently by non-experienced people.

First, AR tools are often meant to be used with programming languages instead of graphical tools, which is a non-sense since common users have opposite needs. Furthermore, building a 3D model is much longer than a 2D storyboard. Managing tracking technology is also very difficult, and the last main issue is matching both real and virtual worlds. DART manages these issues, and to do it it seemed natural to work with designers, since they are the ones who build the final product upon technology. Instead of building a new tool, DART is integrated within Macromedia Director, which many people are used to.

The first feature of DART is the ability to manage virtual actors, which can be animatic characters. Secondly, DART supports synchronized capture and playback of video to capture data. Another feature is the handling of events, which are split in cues and actions. Cues are triggered by certain actions. Finally, tracking data is handled by world-based coordinates and transformations do not depend on the application.

Thus, DART is an efficient early prototype underlining how its integration in a existing tool can enlarge creativity and ease of use. Future work will be to manage several cameras and to have the same authoring and debugging tools as Alice.

 

Alice: Lessons Learned From Building a 3D System for Novices

Since computer graphics are likely to be more and more used in the coming years, Alice aims to be a tool to allow non-experienced undergraduate users to make 3D interactive graphical programs. Indeed, more and more people are interested in this field without having programming knowledge, however some logic will be needed but will have to be kept easy to understand and without strictly mathematical notations. For this purpose, one hundred users are observed during the elaboration.

When using Alice, one first has to create an opening scene by choosing a model, which can then be modified, interact with other objects or be a reference point for the camera. These commands can be performed with the mouse or via a command box. Also, all is animated so that the user can see its model move along the path, and thus spot a mistake, which can naturally undone. For what concerns the implementation, Alice is rendered with D3DRM and implemented in Python.

Tests were led by making couples of subjects learn Alice with the tutorial with no exterior help, except if in a deadlock. Suggestions were also gathered from experienced users. The most important feature is the use of directions instead of mathematical referencials to move objects. Also, common terms are used instead of mathematical terms for ease of understanding. For measurements (rotation for instance), subjects have been asked which one is the favorite one. A last point is that the objects expected the program to require no typing and handle physical laws to give cues or prevent impossible situations.

Although Alice was built for undergraduate students, younger users seem to use it well, making it successful. Future work will focus on avoiding typing in the most of case by making users graphically choose the number and order of parameters. Nowadays, it would be interesting to have such an API being usable with speech.

 

The Effect of Latency and Network Limitations on MMORPGs (A Field Study of Everquest2)

For 10 years, MMORPGs have grown to now reach millions of people playing in these huge virtual environments. To manage these worlds, two solutions exist, the zone based solution which separate the environments in several zones inducing delays between them, and the seamless environment with different zones sharing border places inducing a high level of complexity Thus, this paper addresses the problem of latency in MMORPGs, by studying the case of Everquest2, a 2nd generation MMORPG where latency can be a big deal during battles, which are at the center of the game.

Related work on MMOGs include three approaches.  The first one is the ingame social content and evolution, focusing or social interactions. The second one is the impact of MMORPGs on the real world, for instance because players have to play to buy virtual items, and the third one is game type analysis and user experience to understand the peculiarities of a certain field of games. The last one is mobile games, trying to improve mobile conditions.

Here, tests were led with two good players playing with both the same computer game and configurations (except where in team), with the PCs connected via a switch. Three tests were led, the first one concerned movement and combat, with a latency from 0ms to 5000ms. The hypothesis is that the game time will increase with latency, as well as damages as deaths, for a maximum playable latency of 500ms. Results prove the hypothesis right, except that the breaking point is as 1250ms, with good conditions up to 1000ms, because of the queuing system of the game. The second test was identical but with a group combat, with similar results than the first experience, except that players had to change their tactic starting from 1000ms. The last test concerned movements within the environment with latencies for one or both subjects, expecting a dumb-client model and a server-based environments. This test has been proven correct by the experience since characters were not at the same place on both PCs.

Therefore, Everquest2 is built so that it can be played accurately even with high latencies, but tests hold also be ran with more players. Furthermore, local actions are always possible without waiting the authorization from the server, but this could lead to illegal movements. Finally, on a network level, the size of the packets is always the same, only their rate increase, which would fit for a mobile MMORPG, although it would need the understanding of huge peer-to-peer networks and should be started with first generation MMORPGs. Future work will deal with this problem as well building a platform to manage different networks. But an interesting feature may be to load an environment when entering a zone, so that local interactions can be done correctly (fighting monsters for instance) and above all so that movements are locally checked.

Comments are closed.