Pape Youm - http://436chi-lounge.blogspot.com/2011/02/paper-reading-10-screen-space.html
Evin Schuchardt - http://csce436spring2011.blogspot.com/2011/02/paper-reading-10-screen-space.html
Reference Information
Title: A screen-space formulation for 2D and 3D direct manipulation
Authors: Jason L. Reisman, Philip L. Davidson, Jefferson Y. Han
Presentation Venue: UIST 2010: 22nd annual ACM symposium on User interface software and technology; October 4-7, 2009; Victoria, British Columbia, Canada
Summary
This paper discusses how the authors have extended Rotate-Scale-Translate interactions on touch screens used in a 2D context to three dimensions. The researchers explain some of their techniques with equations they used to map object-space points to screen-space points and explore the different difficulties that arose during their research and how they attempted to fix them.
In their initial experiences with their new system, users had difficulty controlling the two new DOF (degrees-of-freedom), but after several trials they learned the best way to rotate objects was to use both hands. However, this technique was not found to work when the three places at which the user touched the screen were all close to the same depth. Other difficulties they encountered early on was that large rotations sometimes took multiple gestures and a smooth motion of the fingertips did not always generate a smooth move of the object.
The researchers also discussed two rotational problems. Ambiguous rotations occur when the object rotates in the opposite direction from what is expected. To resolve this issue they adjusted the algorithm and experiment with different pressures applied by the user. The other rotational problem was rotational exhaustion which occurs when the distance between two contact points change as the object is moved by the user. To remedy this problem they tried to “correct” the error by limiting the DOF.
The paper also explains how the researchers dealt with error distribution and how they adjusted the different constraints they use in their algorithm to adjust the motion of the object. They tended to call these constraints penalties. One example of a penalty is how the researchers constrained the center of a globe to always remain in the same location, allowing for a user to rotate the globe.
The researchers conclude by providing the readers with several examples of their system when doing one-point, two point and three or more-point interactions.
Discussion
While this paper was interesting, it did get a little too technical for me. There were a lot of in depth explanations that kind of bored me, but to someone who knows more about this area of research, I’m sure it was a very interesting and thorough paper. Most of the papers I have read have not gone into such detail about their entire process, so the extra information will likely be very beneficial to other researchers in this area.
In the paper they mention that in future works they would like to do further general testing to better understand unexpected events and add more penalties to allow for improved manipulation. They would also like to optimize their algorithm.
 |
| Taken from paper: two-point and one-point interactions |
The addition of math formulas also made the reading for me hard, however it definitely solidifies the authors' paper. It conveys that the authors have really grasped to a certain extent the material at hand.
ReplyDeleteYes! In response to your comment on my blog concerning this same paper I thought it was awesome how they responded in each section to the problems that arose.
ReplyDeleteThe future goals in penalties I thought was really interesting. This allows simplicity when needed and could be great to allow greater dynamism in the system.