digital transmission of subjective material appearance...

9
Digital Transmission of Subjective Material Appearance (Supplementary Material) Rodrigo Martín Michael Weinmann Matthias B. Hullin University of Bonn Institute of Computer Science II {rodrigo, mw, hullin}@cs.uni-bonn.de This document provides additional material to the paper ‘Digital Transmission of Subjective Material Appearance’, submitted to the 25 th International Conference on Computer Graphics, Visualization and Computer Vision (May 4, 2017). The contents of this supplement are the following: • Further information regarding the device and methodology employed for the acquisition and digitization of material samples. • The stimuli images used in Experiment 1 and Experiment 2, which comprises the original material photographs (Figures 1 and 2), the digitized spatially-varying BRDF materials (Figures 3 and 4) and photographs where the border regions have been digitally removed (Figure 5). • A set of screenshots showing the user interface from the custom Android application with which the psy- chophysical experiments were carried out (see Figure 6). • Means and confidence intervals for the participants’ response ratings in each experimental condition in the two experiments, separated by material (Figures 7, 8 and 9). On the digitization of materials In order to digitize the selected material samples, we made use of a commercial scanning device [XR16]. The apparatus includes a rotatable sampleholder on which the specimen is placed below a hemispherical gantry with 20 attached LED light sources, four illuminations with 10-band spectral filter wheels, four monochromatic cameras at zenith angles 5.0 , 22.5 , 45.0 and 67.5 and an additional motorized linear light scanner. After taking images of the material sample from different viewpoints and under different illumination conditions, a surface normal map is obtained and the reflectance behavior is stored in terms of an SVBRDF. The latter is based on a diffuse Lambert model and a specular anisotropic Ward model [War92] with bounded albedo [GM10] and modulated with a simple Fresnel term [Sch94]. The output Appearance Exchange Format (AxF) is natively supported across several CAD and rendering applications. 1

Upload: phamtruc

Post on 25-May-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

Digital Transmission of Subjective Material Appearance(Supplementary Material)

Rodrigo Martín Michael Weinmann Matthias B. HullinUniversity of Bonn

Institute of Computer Science II{rodrigo, mw, hullin}@cs.uni-bonn.de

This document provides additional material to the paper ‘Digital Transmission of Subjective Material Appearance’,submitted to the 25th International Conference on Computer Graphics, Visualization and Computer Vision (May4, 2017). The contents of this supplement are the following:

• Further information regarding the device and methodology employed for the acquisition and digitization ofmaterial samples.

• The stimuli images used in Experiment 1 and Experiment 2, which comprises the original material photographs(Figures 1 and 2), the digitized spatially-varying BRDF materials (Figures 3 and 4) and photographs where theborder regions have been digitally removed (Figure 5).

• A set of screenshots showing the user interface from the custom Android application with which the psy-chophysical experiments were carried out (see Figure 6).

• Means and confidence intervals for the participants’ response ratings in each experimental condition in the twoexperiments, separated by material (Figures 7, 8 and 9).

On the digitization of materialsIn order to digitize the selected material samples, we made use of a commercial scanning device [XR16]. Theapparatus includes a rotatable sampleholder on which the specimen is placed below a hemispherical gantry with20 attached LED light sources, four illuminations with 10-band spectral filter wheels, four monochromatic camerasat zenith angles 5.0◦, 22.5◦, 45.0◦ and 67.5◦ and an additional motorized linear light scanner. After taking imagesof the material sample from different viewpoints and under different illumination conditions, a surface normal mapis obtained and the reflectance behavior is stored in terms of an SVBRDF. The latter is based on a diffuse Lambertmodel and a specular anisotropic Ward model [War92] with bounded albedo [GM10] and modulated with a simpleFresnel term [Sch94]. The output Appearance Exchange Format (AxF) is natively supported across several CADand rendering applications.

1

Experimental stimuliThe additional video stimuli presented to the participants can be found at http://cg.cs.uni-bonn.de/en/publications/paper-details/Martin2017/.

leather01 - L1 leather02 - L2 leather03 - L3

leather04 - L4 leather05 - L5

Figure 1: Photographs from the real materials (leathers only) utilized as stimuli in Experiment 1.

fabric01 - F1 fabric02 - F2 fabric03 - F3

fabric04 - F4 fabric05 - F5

Figure 2: Photographs from the real materials (fabrics only) utilized as stimuli in Experiment 1.

2

digitized leather01 - L1d digitized leather02 - L2d digitized leather03 - L3d

digitized leather04 - L4d digitized leather05 - L5d

Figure 3: Digitized material representations (leathers only) utilized as stimuli in Experiment 1 and Experiment 2.

digitized fabric01 - F1d digitized fabric02 - F2d digitized fabric03 - F3d

digitized fabric04 - F4d digitized fabric05 - F5d

Figure 4: Digitized material representations (fabrics only) utilized as stimuli in Experiment 1 and Experiment 2.

3

cropped leather01 - L1c cropped leather05 - L5c cropped fabric01 - F1c

cropped fabric05 - F5c

Figure 5: Photographs from the subset of materials utilized as stimuli in Experiment 2, where the borders havebeen digitally removed.

4

User interface

photographs - PH digitized renderings - DR

digitized videos - PH full-modal - PH

Figure 6: Screenshots from the Android application with which the studies were carried out. Each picture corre-sponds to one of the four conditions composing Experiment 1.

5

Response ratings from Experiment 1

L1 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

L2 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

L3 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

L4 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

L5 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

Figure 7: Response ratings from Experiment 1 for the investigated material samples (leathers only). The centralcircles represent the participants’ mean rating, while the outer circles represent the bootstrapped 95% confidenceinterval for the mean.

6

F1 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

F2 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

F3 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

F4 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

F5 rating means and CI

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPH

FMDV

-3 -2 -1 0 1 2 3

Figure 8: Response ratings from Experiment 1 for the investigated material samples (fabrics only). The centralcircles represent the participants’ mean rating, while the outer circles represent the bootstrapped 95% confidenceinterval for the mean.

7

Response ratings from Experiment 2

L1 rating means and CI (Exp. 2)

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPHc

FM

-3 -2 -1 0 1 2 3

L5 rating means and CI (Exp. 2)

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPHc

FM

-3 -2 -1 0 1 2 3

F1 rating means and CI (Exp. 2)

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPHc

FM

-3 -2 -1 0 1 2 3

F5 rating means and CI (Exp. 2)

rough-smooth

hard-soft

thick-thin

stiff-flexible

shiny-matte

bright-dark

transparent-opaque

homogen.-heterogen.

expensive-cheap

natural-synthetic

beautiful-ugly

unrealistic-believable

DRPHc

FM

-3 -2 -1 0 1 2 3

Figure 9: Response ratings from Experiment 2 for the investigated material samples (leathers only). The centralcircles represent the participants’ mean rating, while the outer circles represent the bootstrapped 95% confidenceinterval for the mean.

8

1 REFERENCES[GM10] Geisler-Moroder, D. and Dür, A. Bounding the albedo of the ward reflectance model. In High Performance Graphics,

2010.

[Sch94] Schlick, C. An inexpensive brdf model for physically-based rendering. In Computer Graphics Forum, vol. 13, pp.233–246. Wiley, 1994.

[War92] Ward, G. Measuring and modeling anisotropic reflection. In ACM SIGGRAPH Computer Graphics, 26(2):pp. 265–272, 1992.

[XR16] X-Rite. Tac7 scanner. http://www.xrite.com/categories/Appearance/tac7, 2016. Accessed at 25th April 2017.

9