pab BRDF test data set, aluminium saw-tooth profile

description - file formats - download original data - download data of regular grid

description

author

Peter Apian-Bennewitz, pab advanced technologies Ltd

revisions:
8th Dec 2012: updated datafiles to more Windows-friendly format
8th Aug 2012: initial version

license

You can do whatever you want with this data, however acknowledgement as pab advanced technologies Ltd being the source of the measured BRDF data must remain visible to the end-user (e.g. in papers, presentations, etc). No warranties whatsoever are implied.

motivation

Documented, public, full spherical BRDF datasets are apparently not very frequent, which limits development of software algorithms. To support this, the dataset serves as a test for BRDF import into simulation programs, especially to test advanced interpolation between incident angles. The measurement set-up used was the out-of-plane, mechanical scanning pgII gonio-photometer at our lab.
For other BRDF and BSDF data and fitting of mathematical models, see www.pab.eu/bme.

material

real/real-tannenbaum1.mini.jpg|
-+
real/real-tannenbaum2.mini.jpg|
-+

Material for this data is a commercially available Aluminium sheet material, with a saw-tooth shaped surface (see images). This makes the material very suitable as a test, since the maximum of this asymmetric BRDF of a non-rotational symmetric material is located at a non-standard coordinate and not around the incident_angle = outgoing_angle location.
The BRDF is broad enough to avoid possible artifacts on the software side due to peaks in the BRDF, but non-Lambertian enough to be of interest. This dataset is regarded as a first test for software algorithms. If this works, more complex and more ''peakish'' BRDFs can be tested. Measurements of finely-resolved peaks with the pgII is easy.
The thin 0.5mm Aluminium sheet is glued on a 5mm float glass pane, which provides a defined and flat mounting. Any curvature in the sample would add caustic effects during measurements, which are thus avoided.

datasets and file formats

coordinates

coordinates-labelled.mini.jpg|
-+

Coordinate system is a spherical theta-phi (Greek letters θ-φ) coordinate system, with theta measured to the surface normal. Angular units are degrees. A short introduction to coordinates of the pgII gonio-photometer is given at our sister side pab-opto.
In this particular case, phi=0 points normal to the saw-tooth profile, so the BRDF should be symmetric in ± phi, conclusively phi is measured from zero to 180deg in this example. An incident direction of phi=0 means that light shines onto the longer, less inclined part of the profile.

wavelength range

BSDF or DSF values are spectrally integrated over 400nm to 700nm. Colour appearance is not the primary target for this test data, we consider the interpolation of outgoing and incoming angles the most interesting feature of the importing software. Colours in visualisations are "false-colour" to show signal levels.

visualisations

mountain-description2.mini.jpg|
-+

A dataset for one incident angle comes with a screendump of mountain, a Linux/Open-GL program written to visualise BRDF data, especially data measured by the pgII gonio-photometer. The BRDF is plotted as a height-field over a disc, with theta extending outward from the disc centre. Measurement points are displayed as pink or yellow dots, along the scan lines of the detector head. The scatter-plane (defined by surface normal and incident direction) is shown as translucent plane.

Note on JPEG images: Visualisation images are JPEG encoded to save space and increase access speed. However, JPEG isn't that good at compression line art, so data points in images appear a bit washed out (especially pink datapoints on green background). Contact us if you like images in higher quality PNG format.

averaging and interpolation

mountain.4259.org.closeup.mini.jpg|
-+
mountain.4259.sa02.closeup.mini.jpg|
-+
mountain.4259.reg.closeup.mini.jpg|
-+
original data averaged 0.2° regular (θ,φ)-grid

Outgoing angles are pre-averaged to 0.2° resolution in the original data along the scan paths, in order to save file size. The minimum angular resolution of the pgII is actually finer. Angular areas where the detector itself would cast a shadow on the sample are automatically left out. Angular areas where the sample mount shadows the outgoing directions are left in for completeness. The latter areas are easily identified by a sharp drop off in value, to the noise-equivalent-bsdf.
Images show the original dataset, the published dataset with a smallest angle along the scan path of 0.2° and the auxiliary interpolated dataset.

DSF versus BSDF

The DSF is defined in ASTM2387 as the product of BSDF×cos(θout). Thus the DSF corresponds to the signal measured by an irradiance detector at position (θ,φ). Since it is mostly used in software, and avoids artifacts (see below), it is given here.
At θout=90°, the DSF value is theoretically zero, and in fact a small, random value depending on detector noise of the measurement device. This noise is amplified in the BSDF by the cosine devision, limiting useful BSDF values up to around θout=89°, depending on gonio-photometer and sample. Since the BSDF is multiplied by cos(θout) inside the simulation software anyway, the DSF seems the data format to prefer.
If your software needs the BSDF, you have to divide each datapoint by cos(θout).

original versus interpolated DSF

The original data-sets with a non-regular grid for (θoutout) are preferred and recommended, since angular resolution is adaptive. This is the way to go. Interpolated data on a regular (θoutout) grid is given here as an alternative for programs which don't yet read (θoutout)-data with an adaptive resolution.


 

original data with adaptive resolution, non-regular grid for (θoutout)

format of ASCII data files

Data for a set of outgoing angles and one incident angle is stored in one data file. These ASCII files consists of a header (comment lines starting with a # sign) , followed by lines of "theta_out phi_out DSF". Decimal sign is '.' (dot), column separator is tab (tabulator), end-of-line is CR-NL (carriagereturn-newline, DOS style). Files should be readable by most programs using Linux, Windows or Mac. Comments lines include details on measurement, for example the numerically integrated direct-hemispherical reflection. In case of specific questions, the "database_id" identifies each measurement.
Outgoing angles are not given on a regular theta-phi grid, and the reading software is expected to handle this. Angular regions of high DSF values have been automatically re-scanned with higher angular resolution: The BRDF/DSF angular resolution is adaptive, which optimises data volume and measurement time.
Incident angles are given on a regular theta_phi grid as shown in the following table. To allow for testing software interpolation between incident angles, some extra datafiles for in-between incident angles are given as well.

archives

Files for incident angles are packed into one archive, including images for visualisations. Available formats are 7-zip archive and compressed tar file:
Scheme for filenames:  data/XYZ_TTT.T_PPP.P_xxxx.dsf  where XYZ specifies the material, TTT.T specifies theta_in, PPP.P specifies phi_in and xxxx specifies the measurement id.
BRDF data: brdf_sawtooth_dataset1_pab.7z or brdf_sawtooth_dataset1_pab.tar.bz2

incident angles matrix

φ=000.0 φ=015.0 φ=030.0 φ=045.0 φ=060.0 φ=075.0 φ=090.0 φ=105.0 φ=120.0 φ=135.0 φ=150.0 φ=165.0 φ=180.0
θ=015.0 i i i i i i i i i i i i i
θ=030.0 i i i i i i i i i i i i i
θ=045.0 i i i i i i i i i i i i i
θ=060.0 i i i i i i i i i i i i i
θ=070.0 i i i i i i i i i i i i i

 

interpolated data on regular (θoutout) grid

format of ASCII data files

All data for a set of outgoing angles and one incident angle is stored in one data file. These ASCII files consists of a header (comment lines starting with a # sign) , followed by lines of "theta_out phi_out DSF". Decimal sign is '.' (dot), column separator is tab (tabulator), end-of-line is CR-NL (carriagereturn-newline, DOS style). Files should be readable by most programs using Linux, Windows or Mac. The DSF is defined by ASTM2387 as the product of the BSDF and cos(theta_out). Comments lines include details on measurement, for example the numerically integrated direct-hemispherical reflection. In case of specific questions, the "database_id" specified in the file identifies each measurement.
Outgoing angles are fixed with θ-step=1° and φ-step=3° . So, θ values are 1°,2°,3°...89° and φ values are 0°,3°,6°...357°. If in doubt, check contents of datafiles.
Incident angles are given on a regular theta_phi grid as shown in the following table. To allow for testing software interpolation between incident angles, some extra datafiles for in-between incident angles are given as well.

archives

Files for incident angles are packed into one archive, including images for visualisations. Available formats are 7-zip archive and compressed tar file:
Scheme for filenames:  grid/XYZ_TTT.T_PPP.P_xxxx.grid  where XYZ specifies the material, TTT.T specifies theta_in, PPP.P specifies phi_in and xxxx specifies the measurement id.
BRDF data: brdf_sawtooth_dataset1_pab.grid.7z or brdf_sawtooth_dataset1_pab.grid.tar.bz2

incident angles matrix

φ=000.0 φ=015.0 φ=030.0 φ=045.0 φ=060.0 φ=075.0 φ=090.0 φ=105.0 φ=120.0 φ=135.0 φ=150.0 φ=165.0 φ=180.0
θ=015.0 i i i i i i i i i i i i i
θ=030.0 i i i i i i i i i i i i i
θ=045.0 i i i i i i i i i i i i i
θ=060.0 i i i i i i i i i i i i i
θ=070.0 i i i i i i i i i i i i i

thank you for your interest. If this data is of use to you, I'd be delighted if you drop me a line.
Peter Apian-Bennewitz, pab advanced technologies Ltd, 2012