Title : Gaze-contingent multi-resolutional graphic library



Project Lead : Santiago Santiago Martn Gonzlez From : University of Oviedo (None)

Dates : from -- to --

Description :

Motivation and objectives :
Motivation: Human vision has many interesting characteristics. One of them is that Human eyes subtend a wide field-of-view and provide high resolution detail at the center of gaze (fovea) with decreasing resolution in the remaining area (periphery). The fovea comprises less than 1% of retinal size but takes up over 50% of the visual cortex in the brain. From a biological point of view, the photoreceptor cells in the fovea region are mainly cones, and outside the fovea they are rods. Rods function mainly in dim light and provide black-and-white vision, while cones support daytime vision and the perception of colour. Rods are also more sensitive to motion detection. This characteristic of human vision suggests that most of our computing power can be concentrated to render the small foveal area. That is, limited computing resources are sufficient to draw low resolution images on the periphery. This would allow us to decrease (or re-orient) the computational load, especially in a CAVE system. Gaze-contingent multi-resolutional displays is the term used in the bibliography for this approach to computer graphics. This result has also direct application in the human visual system research field. There are many related examples, for example the study of saccadic eye movements. Human observers make saccadic eye movements in order to bring new information into the high-acuity fovea centralis of the retina. It is commonly assumed that higher level processing of visual information is confined to the foveal object and that processing of peripheral information only plays a role for selecting the next saccade target. However some researchers disagree with this theory. Related to this problem, activities and computer games have been developed to measure and eventually improve stereopsis in patients with amblyopia (lazy eye). Some optometrists define the amblyopia as a failure in the dominance of the foveal vision compared with the peripherical vision. But it is not easy to confirm that assertion unless we were able to design tests using both foveal and extra-foveal stimuli. Utilizing gaze-contingent displays to simulate depth of field, is one of the suggested methods to help solve problems such as accommodation-convergence conflict and diplopia. Human eyes typically converge and accommodate at the same point. In 3D displays, however, while accommodation occurs at a point on the flat surface of the projection screen, convergence occurs in the front or behind the screen where the image forms. This conflict creates eye strain and can be disturbing. Another interesting field of research refers to the vestibular system, which contributes to balance in most mammals. It sends signals primarily to the neural structures that control our eye movements, and to the muscles that keep us upright. Diseases of the vestibular system usually induce vertigo and instability. Due to the complex interactions among sensory, motor, and central processes involved in posture and balance, a specialized clinical assessment technique called Computerized dynamic posturography (CDP) has been developed. Static posturography is carried out by placing the patient in a standing posture on a fixed instrumented platform connected to sensitive detectors, which are able to detect the tiny oscillations of the body. The platform is contained within an enclosure which can also be used to generate apparent visual surround motions. As peripheral vision is more sensitive to motion detection, gaze-contingent displays could be used to analyze the patient response in new experiments. Objectives: The general objective is to develop the methods to test a gaze-contingent multi-resolutional CAVE. Although we are interested in using this system to study the human visual system, we consider that the objectives of this first project should be focus on the analysis and tuning of the graphic system. The UCL facilities provide both the CAVE and the eye-tracking device, but it is necessary to develop the graphic libraries to create gaze-contingent multi-resolutional 3D contents. Different approaches to generate those contents should be studied. The first option is to dynamically change the resolution of the objects inside the scene: if the objects lie inside the foveal cone of vision the resolution will be high; if not the resolution will be low (number of vertices in the mesh, density of the textures, etc). The second approach renders the scene twice, using different render settings (mainly, pixel resolution target) and it blends them, taking advantage of the graphic library settings. Hardware accelerated fragment programming techniques offer a new flexible option. The effects of those techniques on computational load will be analyzed, as well as their flexibility in different scenarios of research. Those results will allow us to evaluate the most suitable technique.

Teams :
IdeasCAD research group is a young research team, conducted by Dr. Santiago Martin and Dr. Ramon Rubio. Both are lectures at the Engineering Graphic area of the University of Oviedo. They share academic interests and have complementary research fields. IdeasCAD is focused in the design, simulation, and manufacturing cicle. Ecodesign, generative design, virtual reality, reverse engineering, human factors and additive manufacturing are key words for IdeasCAD. Those technics are applied not only in the industrial field but also in other areas like education or medical research.

Dates :
starting date : 07 July, 2014
ending date : 18 July, 2014

Facilities descriptions :
http://visionair-browser.g-scop.grenoble-inp.fr/visionair/Browser/Catalogs/REACTOR.UK.html

Recordings & Results :
A rendering system, specifically designed to allow gaze-moderated manipulations of spatial frequency and other effects, and initially developed at Oviedo, was successfully implemented on the UCL CAVE cluster.

Conclusions :
- Developed a new system for mounting the Arrington eye-tracker on shutterglasses. This was not a planned activity, but necessitated due to an exisiting mounting frame being unsuitable for our experiments - Developed a system that combines eye-tracking and head tracking data to alter rendering around the fixation point - Developed and conducted proof-of-concept tests in the CAVE. - Currently writing a journal article based on applying this system in CAVE environments




Project Images :

fig4.png
fig1.jpg fig2.jpg fig3.png



Other project resources :

ACTIVITY REPORT.pdf


.



Visionair logo

VISIONAIR / Grenoble INP / 46 avenue Felix Viallet / F-38 031 Grenoble cedex 1 / FRANCE
Project funded by the European Commission under grant agreement 262044