A Study in Gloss

Secondment review by Sina Mehraeen

One’s ideas must be as broad as Nature if they are to interpret Nature
                                                                                     ― Arthur Conan Doyle

Here are the chronicles of my adventures during my secondment at the Department of Psychology, University of Southampton, under the supervision of Prof. Wendy Adams and Prof. Marc Ernst.

The Objectives of the trip:

  1. Gain an understanding of the interaction between illumination and material perception, including analysis of spherical high dynamic range natural illumination environments.
  2. Gain experience of working in a Psychology department in another EU country.

To begin at the beginning, I was excited to work with Wendy and to be exposed to the cool research she does on material perception, illumination environments and rendering virtual objects.  I had also never been to the UK, and was eager to get a glimpse of that part of the world.

The plan was twofold; first, I would conduct a pilot study on the interaction between illumination and gloss and how gloss might affect the way we manipulate a physical object. At the same time, I would try to learn as much as possible on rendering virtual objects in natural illumination environments and learn the statistics involved.

Sample HDR illumination environment sourced from the Southampton-York Natural Scenes (SYNS) Dataset. The SYNS dataset is a collection of high definition images (and their relevant metadata) from around 100 rural and urban locations.

After a short trip to a metal supermarket, we got some nice aluminum box pipes which we would use as our physical stimuli. The main question we were seeking to address with these physical objects is whether their glossiness affects the way people handle them. This is in line with previous studies looking at the relationship between gloss and touch (Kerrigan, Adams, & Graf, 2010 ; Adams, Kerrigan & Graf, 2016).

Aluminum box pipe

Once the objects were cleaned and nicely polished, we started testing out a suitable experimental setup. This proved more challenging than expected due to the unpredictable nature of natural lighting. In order to get the desired outcome, we really had to control the angle and intensity of illumination so using natural lighting was out of the question. We also tested various coatings to get the objects to be as glossy as possible. Currently, we have made some tweaks to the set up and we are in the process of data collection; so more info coming soon.

On the side and as a secondary objective, I began learning the basics of Blender to generate virtual objects and to import them as .obj files into Octane Render. Octane is fast rendering engine which allows for easy manipulation of minute elements which make up a scene, e.g. material qualities an object might possess and how those certain qualities (e.g. surface reflectance) interact with light. Once we had set those parameters, we placed them in a natural illumination environment and got something that looked like this:

A virtual object placed in a natural illumination environment

To conclude, I’d like to thank Wendy, Marc and Erich for their continuous support during my stay and also Davide and Paul for helping me with the setup. I’d also like to say a big thanks to everyone I had the good fortune of meeting at the Department of Psychology in Soton; for showing me around town and volunteering to participate in the experiment. Looking forward to seeing you all again soon.


ECVP 2019 ESR Abstracts

ECVP 2019 ESR Abstracts

The ECVP have now released the full abstracts of the attendees to the conference in Leuven. Below you will find the abstracts submitted by DyViTo ESRs. To see the full abstract book, including the abstracts that involve DyViTo Supervisors and friends of the project, please use this link.

Scale ambiguities in material recognition
Jacob R. Cheeseman*, Filipp Schmidt, Roland W. Fleming
Justus Liebig University Giessen

As a rule, observers can reliably identify the material properties of surfaces. Here, we investigated exceptions to this rule using a set of 87 photographs of materials (e.g., water, sand, stone, metal, wood) that appear to belong to different material classes depending on their apparent distance from the viewer. In three experiments, participants viewed each image and provided a categorical judgement of the depicted material, and a quantitative estimate of the distance between the camera and surface. Experiment 1 manipulated interpretations of these images by instructing two groups of participants to imagine a small or large distance between the camera and surface, while a third control group received no such instruction. In Experiments 2 and 3 interpretations were manipulated by providing visual cues for scale (e.g., objects of familiar size), which were presented alongside the target image or digitally inserted into the image. Results indicate that these manipulations can cause identical images to appear to belong to different material classes (e.g., water vs. marble), and that susceptibility to context information (i.e., material ambiguity) correlates with higher variability in distance estimates. Under challenging conditions, therefore, the recognition of some materials is vulnerable to simple manipulations of apparent scale.

Colour Variations within Light Fields: Interreflections and Colour Effects
Cehao Yu* (1), Elmar Eisemann (2), Sylvia Pont (1)
1: Perceptual Intelligence lab, TUDelft; 2: Computer Graphics and Visualization Group, TUDelft

The human visual system incorporates knowledge about local chromatic and lightness effects of interreflections (Bloj et al., Nature, 1999). Here we study basic principles behind chromatic effects of interreflections using computational modelling and photometric measurements. The colour of interreflections varies as a function of the number of bounces they went through. Using a computational model we found that those colour variations can show brightness, saturation and even hue shifts. Using a chromatic Mach Card, a concave folded card with both sides made of the same colour, we demonstrated those three types of colour effects empirically. Finally, we tested the effects of such coloured interreflections on light fields in 3D spaces. Via cubic spectral illuminance measurements in both computer simulations and full mock up room settings under different furnishing scenarios we measure the chromatic variations of first order properties of light fields. The types of chromatic variations were found to depend systematically on furnishing colour, lighting and geometry, as predicted, and also vary systematically within the light field, and thus throughout the space. We will next compare the physical light fields with visual light fields (including chromatic properties) and test perceived material colours, for (combinations of) the three types of effects.

Visual and haptic softness dimensions
Müge Cavdan* (1), Knut Drewing (1), Katja Doerschner (1,2,3)
1: Justus Liebig University Giessen, Germany; 2: Department of Psychology, Bilkent University, Ankara, Turkey; 3: National Magnetic Resonance Research Center, Bilkent University, Ankara, Turkey

When investigating visually or haptically perceived softness of materials researchers have typically equated softness with compliance. However, softness entails more aspects than this single dimension: a rabbit’s fur is soft in a different way than sand on Siesta beach and both’s softness is not necessarily related to the materials’ compliance. Here we investigated the dimensionality of perceived softness in visual and haptic domains. We asked participants to rate various materials on different adjectives. In the haptic experiment, participants were blindfolded and rated materials after haptically exploring them, whereas in the visual experiment they made the same ratings while looking at close up images of the same materials used in the haptic experiment. Principal component analyses revealed that both haptic and visual perception of softness are similarly organized in perceptual space, both containing dimensions of granularity, visco-elasticity, and deformability. However, furriness existed only in the haptic experiment. Moreover, the explained variance was higher in the haptic experiment, which suggests that the perceived dimensions of softness might be more accessible through haptic exploration than by looking at images of materials. Overall, these results contribute to our understanding of how visual and haptic information about material properties are processed and integrated.

Recognising materials over time
Ellen E M De Korte* (1), Andrew J Logan (2), Marina Bloj (1)
1: School of Optometry and Vision Science, University of Bradford, United Kingdom; 2:Department of Vision Sciences, Glasgow Caledonian University, United Kingdom

Materials change over time; colours fade and surfaces are scratched. These changes alter the retinal input and yet we still recognise them as the same material. When textiles are washed and laid out to dry we still identify them as the same fabric even though their colour visibly changes. The present study evaluated the appropriateness of an existing calibrated photograph set as a stimulus for studying the perception of appearance changes of materials over time. Participants (N = 4) reported which of the 2 pairs of images shown displayed the largest perceptual difference. Images were blocked (210 trials per block and participant) by material (Banana, Copper, Granite, Quilted Paper). Individual observers’ perceptual scales, estimated with Maximum Likelihood Difference Scaling via the General Linear Model estimation method, for each material were similar and showed that some, but not all, photographs were perceptually distinct. Thus, the calibrated photographs seem suitable for our purposes. Next steps will include image-based manipulations to establish which parameters drive the development of perceptual scales. Specifically, this will involve converting images to grayscale and manipulate image marks, such as brown staining in Banana images, in order to test the effects of colour and characteristic marks, respectively.