Imaginary Views with DGS
- 0 Collaborators
Generate a 3D asset with a very few images using DGS. ...learn more
Project status: Under Development
Virtual Reality, Mobile, HPC, Networking, Game Development, Artificial Intelligence, Graphics and Media
Intel Technologies
Intel Iris Xe,
Intel Iris Xe MAX,
Intel Deep Link,
Intel Integrated Graphics,
Intel Media SDK,
Intel GPA,
Intel Python,
Intel CPU,
Other
Overview / Usage
Using a reduced number of photos companies can make visual content faster, easier, and higher quality. View synthesis is great for generating synthetic training data for machine learning algorithms. The addition of machine learning to capture allows for the use of lower end cameras like mobile phones to capture better quality imagery.
Methodology / Approach
Advancement in machine learning algorithms has allowed for the computational understanding of the space between photos. Through the use of DGS GPUs these view dependent synthetic data sets can be generated on traditionally unscannable objects like those with shine, sheen, reflectivity, refractivity, and translucency.