Supplementary Material for SVBRDF-Invariant Shape and Reflectance Estimation from Light-Field Cameras

Results on Entire MERL Dataset

We provide shape and reflectance estimation results for all 100 materials in the MERL dataset, both on a sphere and a complicated shape. We compare the results to PLC [1], SDC [2], and PSSM [3]. It can be seen that our method achieves good results on most material types except fabric, which does not follow the half-angle assumption. However, for all the material types, we still outperform the other methods.

Note: if you cannot see some images, try refreshing the webpage, since there are many images to load.

Material types: Metal  Fabric  Phenolic  Plastic  Acrylic  Ceramics  Rubber  Steel  Wood

METAL

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
alum-bronze
aluminium
blue-metallic-paint
blue-metallic-paint2
gold-metallic-paint
gold-metallic-paint2
gold-metallic-paint3
green-metallic-paint
green-metallic-paint2
red-metallic-paint
silver-metallic-paint
silver-metallic-paint2
two-layer-gold
two-layer-silver
brass
chrome
nickel

FABRIC
Since the half-angle assumption fails for fabric, the shape estimation fails for most cases. The BRDFs also look weird, even for the ground truths we computed, since they can hardly be represented as a function of the half-angle.

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
beige-fabric
black-fabric
blue-fabric
dark-specular-fabric
green-fabric
light-brown-fabric
pink-fabric
pink-fabric2
pink-felt
red-fabric
red-fabric2
white-fabric
white-fabric2

PHENOLIC

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
black-phenolic
red-phenolic
specular-black-phenolic
specular-blue-phenolic
specular-green-phenolic
specular-maroon-phenolic
specular-orange-phenolic
specular-red-phenolic
specular-violet-phenolic
specular-white-phenolic
specular-yellow-phenolic
yellow-phenolic

PLASTIC

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
black-soft-plastic
gray-plastic
green-plastic
maroon-plastic
pink-plastic
red-plastic
red-specular-plastic
yellow-matte-plastic
yellow-plastic
pvc
delrin
polyethylene
polyurethane-foam
nylon
teflon

ACRYLIC

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
blue-acrylic
green-acrylic
violet-acrylic
white-acrylic

CERAMICS

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
color-changing-paint1
color-changing-paint2
color-changing-paint3
dark-blue-paint
dark-red-paint
gold-paint
light-red-paint
orange-paint
pearl-paint
purple-paint
silver-paint
white-paint
yellow-paint
aventurnine
black-obsidian
alumina-oxide
hematite
black-oxidized-steel
pink-jasper
silicon-nitrade
tungsten-carbide
white-diffuse-bball
white-marble

RUBBER

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
blue-rubber
neoprene-rubber
pure-rubber
violet-rubber
green-latex

STEEL

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
chrome-steel
grease-covered-steel
steel
ss440

WOOD

Input Image	Ground Truth Depth	Our Depth	Our Depth Error	Our BRDF	PLC [1]	PLC Depth Error	SDC [2]	SDC Depth Error	PSSM [3]	PSSM Depth Error
cherry-235
colonial-maple-223
fruitwood-241
ipswich-pine-221
natural-209
pickled-oak-260
special-walnut-224

[1] M. W. Tao, J.-C. Su, T.-C. Wang, J. Malik, and R. Ramamoorthi. Depth estimation and specular removal for glossy surfaces using point and line consistency with lightfield cameras. IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2015.

[2] M. W. Tao, P. P. Srinivasan, J. Malik, S. Rusinkiewicz, and R. Ramamoorthi. Depth from shading, defocus, and correspondence using light-field angular coherence. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.

[3] H.-G. Jeon, J. Park, G. Choe, J. Park, Y. Bok, Y.-W. Tai, and I. S. Kweon. Accurate depth map estimation from a lenslet light field camera. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.