Samsung electronics co., ltd. (20240354997). HIGH-FIDELITY NEURAL RENDERING OF IMAGES simplified abstract

HIGH-FIDELITY NEURAL RENDERING OF IMAGES

Organization Name

samsung electronics co., ltd.

Inventor(s)

Dimitar Petkov Dinev of Sunnyvale CA (US)

Siddarth Ravichandran of Santa Clara CA (US)

Hyun Jae Kang of Mountain View CA (US)

Ondrej Texler of San Jose CA (US)

Anthony Sylvain Jean-Yves Liot of San Jose CA (US)

Sajid Sadi of San Jose CA (US)

HIGH-FIDELITY NEURAL RENDERING OF IMAGES - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240354997 titled 'HIGH-FIDELITY NEURAL RENDERING OF IMAGES

The abstract of the patent application describes a method for generating images by encoding input data into a latent space, decoding the encoded data through two separate decoders, and combining the outputs to generate an image.

• The encoded data is processed through layers of a first decoder and a second decoder to generate an updated feature map.
• The updated feature map is created by replacing portions of feature maps from different decoder layers.
• The final image is generated by further decoding the updated feature map through additional layers of the first decoder.

Potential Applications: - Image generation for various industries such as entertainment, design, and advertising. - Data compression and reconstruction in image processing applications.

Problems Solved: - Efficient image generation by utilizing multiple decoders and combining their outputs. - Enhanced flexibility in image generation by merging feature maps from different decoder layers.

Benefits: - Improved image quality and diversity through the combination of multiple decoder outputs. - Increased efficiency in image generation processes.

Commercial Applications: - This technology can be used in industries such as graphic design, virtual reality, and medical imaging for enhanced image generation capabilities.

Questions about the Technology: 1. How does this method compare to traditional image generation techniques? 2. What are the potential limitations of combining feature maps from different decoder layers in image generation processes?

Original Abstract Submitted

generating images includes generating encoded data by encoding input data into a latent space. the encoded data is decoded through a first decoder having first decoder layers by processing the encoded data through one or more of the first decoder layers. the encoded data is decoded through a second decoder having second decoder layers by processing the encoded data through one or more of the second decoder layers. an updated feature map is generated by replacing at least a portion of a feature map output from a selected layer of the first decoder layers with at least a portion of a feature map output from a selected layer of the second decoder layers. an image is generated by further decoding the updated feature map through one or more additional layers of the first decoder layers.