Abstract
For precise stimulus control, psychophysical experiments have been usually carried out with a high-performance graphics system set up in a darkroom. The critical limitation of this conventional setup is the difficulty in increasing the data size. To collect psychophysical data with a reasonable quality from a much larger sample of observers, we are developing new software for vision experiments over the internet. This software is based on HTML5 and WebGL implemented in the latest browsers. Editing, distribution, execution and data collection are all operated over web pages browsed by personal computers or tablets. We assessed precision and accuracy of this software with various web browsers running on Windows, Mac OS X, iOS and Android. Temporal precision was measured by sampling luminance of 30 Hz flicker with photodiode. Our software shows good temporal performances under Google Chrome and Internet Explorer regardless of OS, while the timing error increased under the other browsers. Since gamma correction table on the video cards are not operable from browsers, we implemented spatiotemporal dither to linearize the image intensity. Pixel-to-pixel drawing is achieved by complying OpenGL specification. We added a function to keep pixel-to-pixel drawing under pixel-magnifying devices such as Apple retina display. Finally we tested whether basic psychophysical experiments could be replicated with this software. We found, for example, the contrast sensitivity function measured with our new software agreed well with the results reported by past studies and those we collected with a conventional experimental setup. These results certificate that this software is platform-independent tool having a capacity to run many traditional darkroom experiments anywhere and anytime over the internet and/or with tablet devices.
Meeting abstract presented at VSS 2016