-
公开(公告)号：US20240257439A1
公开(公告)日：2024-08-01
申请号：US18612293
申请日：2024-03-21
Applicant: NVIDIA Corporation
Inventor： Shiqiu Liu , Christopher Ryan Wyman , Jon Hasselgren , Jacob Munkberg , Ignacio Llamas
CPC classification number: G06T15/06 , G06T5/20 , G06T5/70 , G06T15/506 , G06T15/60 , G06T2210/21
Abstract: Disclosed approaches may leverage the actual spatial and reflective properties of a virtual environment—such as the size, shape, and orientation of a bidirectional reflectance distribution function (BRDF) lobe of a light path and its position relative to a reflection surface, a virtual screen, and a virtual camera—to produce, for a pixel, an anisotropic kernel filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface. In order to accomplish this, geometry may be computed that corresponds to a projection of a reflection of the BRDF lobe below the surface along a view vector to the pixel. Using this approach, the dimensions of the anisotropic filter kernel may correspond to the BRDF lobe to accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface.
-
公开(公告)号：US11977388B2
公开(公告)日：2024-05-07
申请号：US16282210
申请日：2019-02-21
Applicant: NVIDIA Corporation
Inventor： Jon Hasselgren , Jacob Munkberg
CPC classification number: G05D1/0231 , G05B13/027 , G05D1/0088 , G06N3/02 , G06N3/04 , G06N3/043 , G06N3/045 , G06N3/088
Abstract: The performance of a neural network is improved by applying quantization to data at various points in the network. In an embodiment, a neural network includes two paths. A quantization is applied to each path, such that when an output from each path is combined, further quantization is not required. In an embodiment, the neural network is an autoencoder that includes at least one skip connection. In an embodiment, the system determines a set of quantization parameters based on the characteristics of the data in the primary path and in the skip connection, such that both network paths produce output data in the same fixed point format. As a result, the data from both network paths can be combined without requiring an additional quantization.
-
公开(公告)号：US10776985B2
公开(公告)日：2020-09-15
申请号：US16355214
申请日：2019-03-15
Applicant: NVIDIA Corporation
Inventor： Shiqiu Liu , Christopher Ryan Wyman , Jon Hasselgren , Jacob Munkberg , Ignacio Llamas
Abstract: Disclosed approaches may leverage the actual spatial and reflective properties of a virtual environment—such as the size, shape, and orientation of a bidirectional reflectance distribution function (BRDF) lobe of a light path and its position relative to a reflection surface, a virtual screen, and a virtual camera—to produce, for a pixel, an anisotropic kernel filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface. In order to accomplish this, geometry may be computed that corresponds to a projection of a reflection of the BRDF lobe below the surface along a view vector to the pixel. Using this approach, the dimensions of the anisotropic filter kernel may correspond to the BRDF lobe to accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface.
-
公开(公告)号：US20190287294A1
公开(公告)日：2019-09-19
申请号：US16355214
申请日：2019-03-15
Applicant: NVIDIA Corporation
Inventor： Shiqiu Liu , Christopher Ryan Wyman , Jon Hasselgren , Jacob Munkberg , Ignacio Llamas
Abstract: Disclosed approaches may leverage the actual spatial and reflective properties of a virtual environment—such as the size, shape, and orientation of a bidirectional reflectance distribution function (BRDF) lobe of a light path and its position relative to a reflection surface, a virtual screen, and a virtual camera—to produce, for a pixel, an anisotropic kernel filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface. In order to accomplish this, geometry may be computed that corresponds to a projection of a reflection of the BRDF lobe below the surface along a view vector to the pixel. Using this approach, the dimensions of the anisotropic filter kernel may correspond to the BRDF lobe to accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface.
-
公开(公告)号：US20220327765A1
公开(公告)日：2022-10-13
申请号：US17852132
申请日：2022-06-28
Applicant: NVIDIA Corporation
Inventor： Shiqiu Liu , Christopher Ryan Wyman , Jon Hasselgren , Jacob Munkberg , Ignacio Llamas
Abstract: Disclosed approaches may leverage the actual spatial and reflective properties of a virtual environment—such as the size, shape, and orientation of a bidirectional reflectance distribution function (BRDF) lobe of a light path and its position relative to a reflection surface, a virtual screen, and a virtual camera—to produce, for a pixel, an anisotropic kernel filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface. In order to accomplish this, geometry may be computed that corresponds to a projection of a reflection of the BRDF lobe below the surface along a view vector to the pixel. Using this approach, the dimensions of the anisotropic filter kernel may correspond to the BRDF lobe to accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface.
-
Patent Agency Ranking
公开(公告)号：US20200349755A1
公开(公告)日：2020-11-05
申请号：US16935431
申请日：2020-07-22
Applicant: NVIDIA Corporation
Inventor： Shiqiu Liu , Christopher Ryan Wyman , Jon Hasselgren , Jacob Munkberg , Ignacio Llamas
Abstract: Disclosed approaches may leverage the actual spatial and reflective properties of a virtual environment—such as the size, shape, and orientation of a bidirectional reflectance distribution function (BRDF) lobe of a light path and its position relative to a reflection surface, a virtual screen, and a virtual camera—to produce, for a pixel, an anisotropic kernel filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface. In order to accomplish this, geometry may be computed that corresponds to a projection of a reflection of the BRDF lobe below the surface along a view vector to the pixel. Using this approach, the dimensions of the anisotropic filter kernel may correspond to the BRDF lobe to accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface.
-
公开(公告)号：US20250103049A1
公开(公告)日：2025-03-27
申请号：US18654351
申请日：2024-05-03
Applicant: NVIDIA Corporation
Inventor： Jon Hasselgren , Jacob Munkberg
IPC: G05D1/00 , G05B13/02 , G05D1/227 , G05D1/249 , G06N3/02 , G06N3/04 , G06N3/043 , G06N3/045 , G06N3/088
Abstract: The performance of a neural network is improved by applying quantization to data at various points in the network. In an embodiment, a neural network includes two paths. A quantization is applied to each path, such that when an output from each path is combined, further quantization is not required. In an embodiment, the neural network is an autoencoder that includes at least one skip connection. In an embodiment, the system determines a set of quantization parameters based on the characteristics of the data in the primary path and in the skip connection, such that both network paths produce output data in the same fixed point format. As a result, the data from both network paths can be combined without requiring an additional quantization.
-
公开(公告)号：US11941745B2
公开(公告)日：2024-03-26
申请号：US17852132
申请日：2022-06-28
Applicant: NVIDIA Corporation
Inventor： Shiqiu Liu , Christopher Ryan Wyman , Jon Hasselgren , Jacob Munkberg , Ignacio Llamas
CPC classification number: G06T15/06 , G06T5/002 , G06T5/20 , G06T15/506 , G06T15/60 , G06T2210/21
Abstract: Disclosed approaches may leverage the actual spatial and reflective properties of a virtual environment—such as the size, shape, and orientation of a bidirectional reflectance distribution function (BRDF) lobe of a light path and its position relative to a reflection surface, a virtual screen, and a virtual camera—to produce, for a pixel, an anisotropic kernel filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface. In order to accomplish this, geometry may be computed that corresponds to a projection of a reflection of the BRDF lobe below the surface along a view vector to the pixel. Using this approach, the dimensions of the anisotropic filter kernel may correspond to the BRDF lobe to accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface.
-
公开(公告)号：US11373359B2
公开(公告)日：2022-06-28
申请号：US16935431
申请日：2020-07-22
Applicant: NVIDIA Corporation
Inventor： Shiqiu Liu , Christopher Ryan Wyman , Jon Hasselgren , Jacob Munkberg , Ignacio Llamas
Abstract: Disclosed approaches may leverage the actual spatial and reflective properties of a virtual environment—such as the size, shape, and orientation of a bidirectional reflectance distribution function (BRDF) lobe of a light path and its position relative to a reflection surface, a virtual screen, and a virtual camera—to produce, for a pixel, an anisotropic kernel filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface. In order to accomplish this, geometry may be computed that corresponds to a projection of a reflection of the BRDF lobe below the surface along a view vector to the pixel. Using this approach, the dimensions of the anisotropic filter kernel may correspond to the BRDF lobe to accurately reflect the spatial characteristics of the virtual environment as well as the reflective properties of the surface.
-
公开(公告)号：US20200272162A1
公开(公告)日：2020-08-27
申请号：US16282210
申请日：2019-02-21
Applicant: NVIDIA Corporation
Inventor： Jon Hasselgren , Jacob Munkberg
Abstract: The performance of a neural network is improved by applying quantization to data at various points in the network. In an embodiment, a neural network includes two paths. A quantization is applied to each path, such that when an output from each path is combined, further quantization is not required. In an embodiment, the neural network is an autoencoder that includes at least one skip connection. In an embodiment, the system determines a set of quantization parameters based on the characteristics of the data in the primary path and in the skip connection, such that both network paths produce output data in the same fixed point format. As a result, the data from both network paths can be combined without requiring an additional quantization.
-
-
-
-
-
-
-
-
-
Search Results
10
records
(took 0.013 seconds)
