Cauwerts Coralie – UCLouvain/ENTPE
This study is part of a broader research project investigating preferences for color combinations in architectural spaces with psychophysical techniques. Working in real environments is challenging, especially with daylight, because of the need to control experimental conditions. We consider images as alternative visual stimuli to actual spaces. The use of pictures in lighting and color research instead of actual environments has already been investigated. Previous works concluded that both camera and computer generated-pictures have the potential to predict how the not yet built environment will be experienced. But some limitations have also been pointed out. Each step in the production of pictures – from scene creation to display observation – can indeed lead to inaccuracies.
The paper will focus on scene creation and will report photometric and colorimetric accuracy as well as sharpness of three types of images:
1) Images captured with professional digital cameras and HDR technique;
For more than ten years, HDR photography has increasingly been used by lighting researchers as a luminance data acquisition tool. They have also been produced for investigating the emotional response of the occupants to built environments. A luminance error less than 10% can be expected. Only few data are available regarding color accuracy.
2) Images captured with hyperspectral cameras;
Hyperspectral imaging technology is today mature enough to produce high definition images while capturing for each spatial pixel spectral information with an optical resolution equivalent to spot spectroradiometers. In our context of color research in architecture, it makes hyperspectral cameras perfect candidates to produce visual stimuli.
3) Images generated by a spectral rendering software (Ocean light simulator).
Since the first validation works of computer-generated images carried out in a context of color research and architecture in the 2000s, rendering packages have greatly improved. Nevertheless, most physically-based rendering software are complex and sometimes difficult to use. And the current lack of material properties database can still lead to discrepancies between real and rendered materials.
Photometric and colorimetric accuracy will be quantified in comparison to spot measurement with a spectro-radiometer. Visual acuity chart will be used to evaluate how well the camera or the software creates sharp images. Finally, advantages and disadvantages (including material cost and time consumption) of each type of image will be discussed.