Research Topics

Human perception of colors and rendering

Many virtual reality requires a good immersion feeling. The user position in the virtual world gives camera point of view. And the rendered image should give a good feed back of what the user should see. In order to improve the image quality we use state of the art global illumination algorithms. But these algorithms only take physics into account. They do not take care of the human percertion of colors. In color science, the color appearance models (CAM) describe the human perception of colors. Our main objective is to take CAM key features into account in the rendering engine. The chromatic adaptation also called white balance is one of these CAM key features. Even state of the art chromatic adaptation algorithms do not give good results in the case of global illumination because they makes strong assumptions about the ’real’ image to white balance. Unfortunately theses assumptions are no more true in ’virtual’ images. So we purpose a new chromatic adaption algorithm suitable with global illumination. This algorithm takes direct and indirect illuminations (i.e. diffuse inter-reflection) into acount. Our algorithm does not depend of the global illumination engine, it works with radiosity, stochastic ray tracing and photon mapping engines.

Interactive visual rendering

Most virtual reality applications need real-time visual rendering. However, the virtual images may sometimes look “too synthetic” or “too perfect” and, therefore, might provide a weak immersion feeling. Thus, we have studied innovative techniques to adapt the visual rendering interactively to the user’s gaze, in order to enhance the final graphic rendering and the immersion feeling. Hereafter, we describe our recent results obtained in the field of ”interactive visual rendering”:

1. a novel Depth-of-Field visual blur effect,
2. the use of eye-tracking to improve visual rendering.

Depth-of-Field visual blur effect Depth-of-field (DoF) of the human’s eyes is the range of distances near the point of focus where the eyes perceive the image as sharp. Objects behind and in front of the point of focus are blurred. Depth-of-field and its associated blur effects are well-known and classical depth cues in human vision, and they have been already introduced in computer graphics years ago. We have studied several techniques to improve real-time Depth-of-Field blur rendering: a novel blur computation based on the GPU, an auto-focus zone to automatically compute the user’s focal distance without an eye-tracking system, and a temporal filtering that simulates the accommodation phenomenon. Then, using an eye-tracking system, we have analyzed users’ focus point during first-person navigation in order to better set the parameters of our algorithm. Lastly, we have conducted an experiment to study the influence of visual blur effects on performance and subjective preference of first-person shooter gamers. Our results suggest that our blur effects could improve fun or realism of rendering, making them suitable for video gamers, depending however on their level of expertise.

We have studied the use of users focus point to improve visual rendering in virtual environments. First, we have studied how to retrieve users focus point in the 3D virtual environments using an eye-tracking system. Then, we have proposed the adaptation of two rendering techniques:

1. A camera motion which simulates eyes movement when walking
2. Depth-of-Field (DoF) blur effect. Second, we have conducted an experiment to study users’ subjective preferences concerning these visual effects during first-person navigation in VE. It showed that participants globally preferred the use of these effects when they are dynamically adapted to the focus point in the VE. Taken together, our results suggest that the use of visual effects exploiting users’ focus point could be used in several VR applications involving first-person navigation such as the visit of architectural site, training simulations, video games, etc.

Real-time Rendering for Multiview Autostereoscopic Displays

Multiview autostereoscopic displays are now available at affordable cost and are set to become widely used in virtual reality applications and 3D games. With their wide viewing zone, this type of display easily accommodates multiple viewers and no head tracking is required. However, real time rendering on these displays poses a number of difficult problems, the first one being of course the simultaneous generation of several views of the same 3D scene. Besides, the particular sampling pattern of the displayed image requires specific anti-aliasing procedures and this results in limiting the usable depth range. The purpose of our work is thus to tackle these problems. In particular, we have tested various virtual cameras settings with a view to keep the region of interest within the usable depth range of the display. We have also developed rendering methods allowing the generation of the interlaced multiview image with a commodity graphic hardware. Our method accounts for the depth of field effect in case of multiview autostereoscopic display.

Links

CV (in french)

Publications

Figures

Figure 1: Global illumination without chromatic adaptation

Figure 2 : Global illumùination with chromatic adaptation

Figure 3 : Depth of field blur in First player game (Sébastien Huillaire, PhD, work)