The player was alone (no targets) and could move freely. We used our camera motion and DoF blur effect implemented in the video game Quake III as described in section 3 and section 4. An experiment has been conducted to study the subjective pref- erence of users regarding the DoF blur effect and camera motion when eye-tracking is enabled or not. hardware The configuration video game with aforementioned our techniques were imple- used in the experiment described in the next section. On a PC with an Intel Pen- tiumD CPU at 3.4Ghz, 2.0Gb of RAM and a nVidia Quadro FX 3450/4000 SDI, performance was of 70 frames per second with the both effects activated, instead of 260 without, at a resolution of 1280 mented × 1024 and the pixels. They have been successfully implemented in the real time open-source engine of Quake III video game 1. We have enhanced two existing techniques suitable for first-person navigation in VE to make them work with an eye-tracking system: (1) a compensated camera motion and (2) a DoF blur effect. Thanks to the focus zone centered on fp scr, the focus is done on the semantically important character instead of the background, even if the character covers few pixels in the auto-focus zone. Figure 2 illustrates our algorithm implemented in real time in the Quake III video game engine. The implementation parameters are the same as in. Then, by ap- plying a gathering blur technique, we obtain the final blurred image without the problem of color leaking, thanks to depth comparisons. To compute the amount of blur per pixel we use a lens model that takes into account the focal distance. We simulate a part of the hu- man visual system: focal distance accommodation using a low-pass filter and DoF blur. The computation of the DoF blur effect is achieved using the classical techniques described in. Thereafter, we describe the implementation of a DoF blur effect adapted in real time to user’s focal distance in VE thanks to an eye-tracking system and to the algorithm described in section 2. It was suggested that an eye-tracking system could improve the results. In a previous study, it was shown that, in absence of an eye-tracking system, only a half of the participants enjoyed a DoF blur effect computed with a focus zone constantly positioned at the center of the screen. They concluded that it could be due to the very slow frame rate of their application and they suggested to implement and further evaluate real time DoF blur effect in VE. How- ever, their results did not show evidences of performance improve- ment when the DoF effect was computed using an eye-tracker. conducted an evaluation of DoF blur effect using a stereoscopic display. The second effect we propose to improve visual feedback using users’ focus point is a DoF blur effect. The resulting camera motion is thus compensated in real time based on the point the user is looking at in the VE, thanks to the eye-tracking system. Our other parameters were set using preliminary testing to: ( T x c, T y c, T z ve ) = ( 0 ms, 780 ms, 390 ms ), ( K x c, K y c, K z ve ) = ( 0, 0. In our final implemen- tation, α and α angles were set to 0. Then, ( e s, a s ) are elevation and azimuth angles of fp ve in the camera reference frame at position C in Figure 3B. ( e l, a l ) are elevation and azimuth angles of fp ve in camera reference frame when following the initial linear motion as shown in Figure 3A, i.e., before offsets are applied to camera position C.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |