Illumination discrimination in the absence of a fixed surface-reflectance layout

Supplemental Material

Ana Radonjić, Xiaomao Ding, Avery Krieger, Stacey Aston, Anya C. Hurlbert and David H. Brainard

Instructions
Experiment 1: pdf
Experiment 2: pdf

Comparison illumination tables.
Table specifying the difference between the target and comparison illuminations in ΔE L*u*v* units are provided as tab delimited text (txt). In producing the tables, target illuminant CIEXYZ coordinates are used as the reference white point (for illuminant spectra to XYZ to L*u*v* conversion).

We also provide image-based lookup tables for each of our experiments. In these tables we provide the differences (ΔE L*u*v* units) between each comparison illumination test image and the mean target image (averaged across all 30 target images). In producing the tables, we used image-based white pointsfor mean image XYZ to L*u*v* conversion; these differ slighty across experiments (see manuscript for more details on these computations). Each table provides values for all test images in the fixed-surfaces condition (columns 2-5) and in the shuffled-surfaces condition (columns 6-9). Column headers indicate illumination-change direction and condition. Tables are provied as tab delimited text for Experiment 1, Experiment 2 and Experiment 1A.

Stimulus scene

 Stimulus geometry. Blender file specifying the position of each element in the stimulus is available here (Blender is required to view this file).
Render toolbox mapping files which are used for rendering are provided as tab delimited text (txt). Mapping files provide scene specification for rendering that is common across all test scenes.
Render toolbox conditions files which specify surface-to-reflectance mapping for each rendered scenes are provided as tab delimited text (txt) for each condition (files for target and comparison scenes are provided separately). The files provided here are for rendering of the left image only. The files for rendering of the right image are identical except the eye position (which is -3.2 for the right and 3.2 for the left image).
Each row of the file corresponds to one image. Header labels in each collumn denote the image feature that is specified in each column. The first column corresponds to image name (includes: a letter indicating the condition ('C'-fixed-surfaces, 'S'-shuffled-surfaces), scene type specification ('T' for target, illumination direction for comparison illuminations) and the image number (either the target number of the illumination step).
Experiment 1/Experiment A1:
    Fixed-surfaces condition: target scenes (txt)     comparison scenes (txt)
    Shuffled-surfaces condition: target scenes (txt)     comparison scenes (txt)
Experiment 2:
    Fixed-surfaces condition: target scenes (txt)     comparison scenes (txt)
    Shuffled-surfaces condition: target scenes (txt)     comparison scenes (txt)
Note: In Experiment 2 and Experiment A1, the intended surface reflectance of four floor tiles was replaced by Blender-assigned pinkish surface in all rendered images. The label for these tiles in the Blender file and the mappings file is ''FloorMat9" (reflectance value is included in the surface reflectance txt file below).
Illuminants. Spectral power distribution for the target and comparison illuminants as tab delimited text (grouped by illuminant). Spectra are provided in arbitraty units. Files are grouped by illuminant. Target illuminant is denoted as 'blue0'. 1st column shows wavelength in nm (380 - 780 nm, 4 nm steps); different spectral sampling was used for rendering  (400 - 700 nm, 10 nm steps). For the remaining columns, header labels correspond to comparison illuminant spectra with numbers denoting the nominal comparison illuminant step (difference relative to the target).
Blue Yellow Green Red
Surfaces. Simulated surface reflectance functions as tab delimited text. (txt)
Table format: 1st column: wavelength in nm (400 - 780 nm, 4 nm steps); different spectral sampling was used for rendering  (400 - 700 nm, 10 nm steps).
For the remaining columns, header labels correspond to surface reflectance labels specified in the mapping files. Recovered reflectance of the Blender-assigned surface used due to a bug in experimental code in Experiment 2 and Experiment A1 is provided in the last column.

Individual subject data

Follow-up questionnaires: Experiment 1 (pdf), Experiment 2 (pdf), Experiment A1 (pdf).

Thresholds. The tables below provide individual subject data for each experiment and condition as tab delimited text.
Each row corresponds to one subject (subject numbers are assigned arbitrarily) and shows measured thresholds for each illuminant direction for each block of trials (as indicated by the header labels). In the shuffled-surfaces condition, asterisks denote estimated thresholds when the measured threshold in a given block falls out of the stimulus range. In the fixed-surfaces condition, NaN values denote cases where the thresholds fell outside of stimulus range. 

Experiment 1: Fixed-surfaces, Shuffled-surfaces.
(note: subject #11 was excluded after the second fixed-surfaces session; second block in the shuffled-scene condition was not run in this case and corresponding threshold values are denoted with NaNs.)

Experiment A1: Fixed-surfaces, Shuffled-surfaces.
(note: subject #11 was excluded after the second fixed-surfaces session; second block in the shuffled-scene condition was not run in this case and corresponding threshold values are denoted with NaNs; subject 11 in this experiment is a different individual than subject #11 in Experiment 1 above.)

Experiment 2: Fixed-surfaces, Shuffled-surfaces.

Eye-movement data.

Proportion of dropped-trials per each observer, condition and trial block of trials is provided as txt table.

Observer color mapping code for Figures 6, 7, 8 and 9 as well as figures B1, B2, B3 and B4 in the supplement is shown below.

ObserverMappingCode

Raw eye position data. For each observer and condition we provide three pairs of table, one for each trial interval (target, first comparison, second comparison). Each pair consists of a x coordinates table and a y coordinate table.
The first row in a table indicates a block number (1 or 2), the second row indicates the trial number within that block. For each trial, we then provide x / y eye coordinate for the duration of the trial (in pixels relative to the upper left corner of the screen). Position at first millisecond of the trial is shown in row 3, position in the next millisecond in row 4 and so forth. Each table is padded with Inf values. These indicate that a given trial has ended. NaN values indicate the missing data on the eye position in the course of the trial.

Surface reflectance distribution. For each observer we plot the relative distribution of fixations across reflectances. Fixations are aggregated across trials and plotted for each condition and interval. Fixed-surfaces condition is shown in black, shuffled-surfaces condition is shown in white. For each condition, target interval data is indicated with solid line, first comparison interval data with dashed and second comparison interval data with dotted line. Each surface reflectance is indicated with a square symbol (x-axis) whose RGB values correspond to colors used in Figure 5. Black leftmost reflectance corresponds to the stimulus scene background. Pink rightmost reflectance corresponds to the reflectance assigned by Blender (consequence of the typo in the mapping file).

Eye fixation distribution across conditions. The figures follow the same formating style as Figure 5 in the paper. For each condition and interval fixations are aggregated across all trials and overlaid onto the corresponding locations in the stimulus image (white circles). For each condition, each interval is presented in a separate figure. Figures show fixations over the entire display. Tile colors shown in this figure are for illustration purposes only.

Experiment 2:

(1) Observer AZM
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(2) Observer BMJ
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(3) Observer VLE 
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(4) Observer VVU
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(5) Observer IDH
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(6) Observer HUL
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(7) Observer IJJ
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(8) Observer EOM
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(9) Observer DTM
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)

(10) Observer KTV
Raw data:
Fixed-surfaces:     TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY
Shuffled-surfaces: TargetIntX    TargetIntY    1stComparisonIntX    1stComparisonIntY    2ndComparisonIntX    2ndComparisonIntY   
Eye fixations distributions:
Fixed-surfaces:     TargetInterval   1stComparisonInterval   2ndComparisonInterval
Shuffled-surfaces: TargetInterval  1stComparisonInterval   2ndComparisonInterval
Surface reflectance distribution (pdf)