Datasets:
Tasks:
Image-to-3D
Modalities:
Image
Formats:
parquet
Languages:
English
Size:
10K - 100K
ArXiv:
License:
Update README.md
Browse files
README.md
CHANGED
|
@@ -601,14 +601,14 @@ size_categories:
|
|
| 601 |
|
| 602 |
|
| 603 |
<p align="center">
|
| 604 |
-
<img src="https://img.shields.io/badge/Project-Page-blue" style="height:
|
| 605 |
-
<img src="https://img.shields.io/badge/Paper-PDF-red" style="height:
|
| 606 |
-
<img src="https://img.shields.io/badge/Code-GitHub-yellow" style="height:
|
| 607 |
</p>
|
| 608 |
|
| 609 |
|
| 610 |
<div style="max-width: 800px; margin: 0 auto; padding: 0 20px; text-align: justify;">
|
| 611 |
-
<h2 style="text-align: center; margin-bottom:
|
| 612 |
<p style="font-size: 20px; text-align: justify; line-height: 1.6;">
|
| 613 |
Modern 3D reconstruction and novel view synthesis approaches have demonstrated strong performance on scenes with opaque Lambertian objects. However, most assume straight light paths and therefore cannot properly handle refractive and reflective materials. Moreover, datasets specialized for these effects are limited, stymieing efforts to evaluate performance and develop suitable techniques. In this work, we introduce a synthetic RefRef dataset and benchmark for reconstructing scenes with refractive and reflective objects from posed images.
|
| 614 |
Our dataset has 50 such objects of varying complexity, from single-material convex shapes to multi-material non-convex shapes, each placed in three different background types, resulting in 150 scenes.
|
|
|
|
| 601 |
|
| 602 |
|
| 603 |
<p align="center">
|
| 604 |
+
<img src="https://img.shields.io/badge/Project-Page-blue" style="height: 27px; margin: 5px;">
|
| 605 |
+
<img src="https://img.shields.io/badge/Paper-PDF-red" style="height: 27px; margin: 5px;">
|
| 606 |
+
<img src="https://img.shields.io/badge/Code-GitHub-yellow" style="height: 27px; margin: 5px;">
|
| 607 |
</p>
|
| 608 |
|
| 609 |
|
| 610 |
<div style="max-width: 800px; margin: 0 auto; padding: 0 20px; text-align: justify;">
|
| 611 |
+
<h2 style="text-align: center; margin-bottom: 25px;">Abstract</h2>
|
| 612 |
<p style="font-size: 20px; text-align: justify; line-height: 1.6;">
|
| 613 |
Modern 3D reconstruction and novel view synthesis approaches have demonstrated strong performance on scenes with opaque Lambertian objects. However, most assume straight light paths and therefore cannot properly handle refractive and reflective materials. Moreover, datasets specialized for these effects are limited, stymieing efforts to evaluate performance and develop suitable techniques. In this work, we introduce a synthetic RefRef dataset and benchmark for reconstructing scenes with refractive and reflective objects from posed images.
|
| 614 |
Our dataset has 50 such objects of varying complexity, from single-material convex shapes to multi-material non-convex shapes, each placed in three different background types, resulting in 150 scenes.
|