Below you will find results for the subset single-material spheres of our real test set (Table 1, Column: Real Spheres in the paper). On the left part of this webpage you can briefly browse the results showing from left to right the initial images, our predicted illumination and the ground truth illumination (similar format as Figure 8 in the paper). You can scroll down to see all the different examples. If you click one of the examples then on the right part of the webpage the detailed results for this particular example appear, showing images, inputs (reflectance maps and background), baselines, the different variants of our method, and the reference illumination (similar format as Figure 5 in the paper). Note that, in this particular case (i.e. multiple single-material objects under the same illumination) where multiple background images exist, one for each singlet, only the Background 1, presented on the right part, is fed to the Triplet + Background CNN.
| Our predicted illumination | Ground-thuth illumination | ||||
|---|---|---|---|---|---|
![]() |
![]() |
||||
| Image 1 | Image 2 | Image 3 | |||
![]() |
![]() |
![]() |
|||
| Reflectance Map 1 |
Background 1 | Reflectance Map 2 |
Background 2 | Reflectance Map 3 |
Background 3 |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
| Singlet 1 | Singlet 1 + Background |
Singlet 2 | Singlet 2 + Background |
Singlet 3 | Singlet 3 + Background |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
| Best-of- Singlets |
Nearest Neighbor |
Mask-aware Mean |
Triplet | Triplet + Background |
Ground Truth |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |