With new technology, computer generated imagery (CGI) is closer than ever to becoming indistinguishable from photography. Architectural artists are keen on recreating the chaos which surrounds us; small details like finger prints, scratches, dirt etc. which is what lifts the visualisation from clinical and pure and into hyper realistic visualisations. However, with these realistic CGI it may mislead any potential client as the end result may be very different from what the artists had in mind.

The lack of validation of CGI surface materials may be one of the problems that ultimately leads to different visual appearance. In this project a novel approach was designed to validate real surface materials in two different environment with different levels of complexity (level of semantics), whilst keeping human perception in mind. The purpose of having two environments was to see, if there were any differences in visualising surfaces in a simple environment compared to a more complex environment and if the threshold of accepting what was similar were consistent in the two environments.

Three materials were processed in the project; a highly specular material (whiteboard), a diffuse material (post-it note) and a glossy material (table), which were all calibrated as close as possible to a real reference. Small deviations were made to the calibrated sample in order to get a range of samples that differed from the calibrated sample.

The results from the experiments shows that whilst the assessors were able to discern between the changes in the two environments, the low semantic environment were rated more in concordance between assessors and in general had more defined groupings for the specular and diffuse materials compared to the same materials implemented in the high semantic environment. The glossy material had very little groupings in both of the environment and indicates a larger range of acceptance for this type of materials.