-
公开(公告)号：US20190251397A1
公开(公告)日：2019-08-15
申请号：US16256820
申请日：2019-01-24
申请人： NVIDIA Corporation
发明人： Jonathan Tremblay , Aayush Prakash , Mark A. Brophy , Varun Jampani , Cem Anil , Stanley Thomas Birchfield , Thang Hong To , David Jesus Acuna Marrero
CPC分类号： G06K9/6257 , G06K9/6228 , G06K2209/21 , G06N3/04 , G06N3/084 , G06T15/04 , G06T15/20 , G06T15/50 , G06T2210/12
摘要： Training deep neural networks requires a large amount of labeled training data. Conventionally, labeled training data is generated by gathering real images that are manually labelled which is very time-consuming. Instead of manually labelling a training dataset, domain randomization technique is used generate training data that is automatically labeled. The generated training data may be used to train neural networks for object detection and segmentation (labelling) tasks. In an embodiment, the generated training data includes synthetic input images generated by rendering three-dimensional (3D) objects of interest in a 3D scene. In an embodiment, the generated training data includes synthetic input images generated by rendering 3D objects of interest on a 2D background image. The 3D objects of interest are objects that a neural network is trained to detect and/or label.
-
公开(公告)号：US20190244430A1
公开(公告)日：2019-08-08
申请号：US16251575
申请日：2019-01-18
发明人： Jonathan Steed , Aaron Krauss , Mike Scavezze , Wei Zhang , Arthur Tomlin , Tony Ambrus , Brian Mount , Stephen Latta , Ryan Hastings
CPC分类号： G06T19/006 , G06T15/50 , G06T2215/16
摘要： Embodiments related to efficiently constructing an augmented reality environment with global illumination effects are disclosed. For example, one disclosed embodiment provides a method of displaying an augmented reality image via a display device. The method includes receiving image data, the image data capturing an image of a local environment of the display device, and identifying a physical feature of the local environment via the image data. The method further includes constructing an augmented reality image of a virtual structure for display over the physical feature in spatial registration with the physical feature from a viewpoint of a user, the augmented reality image comprising a plurality of modular virtual structure segments arranged in adjacent locations to form the virtual structure feature, each modular virtual structure segment comprising a pre-computed global illumination effect, and outputting the augmented reality image to the display device.
-
公开(公告)号：US20190236838A1
公开(公告)日：2019-08-01
申请号：US16377474
申请日：2019-04-08
发明人： Seungin PARK , Minsu Ahn , Hyong Euk Lee , Inwoo Ha
CPC分类号： G06T15/50 , G06K9/00624 , G06K9/6202 , G06T15/506 , G06T17/00 , G06T2215/12
摘要： A three-dimensional (3D) rendering method for rendering a 3D model includes rendering a 3D model from each viewpoint of direct light sources, obtaining scene information on scenes rendered from each viewpoint of the direct light sources, generating a matched image in which the scenes are matched based on the scene information, and sampling at least one indirect light source from the matched image.
-
公开(公告)号：US20190208158A1
公开(公告)日：2019-07-04
申请号：US15861526
申请日：2018-01-03
申请人： Plantronics, Inc.
摘要： A display apparatus uses a LED array having a LED pixel pitch, having LED pixels emitting light with viewing angles θ. A transmissive diffuser panel is mounted with the LED array, having a directly lit, non-planar diffuser panel surface spaced away from the LED array by at least distance D. The distance D and the viewing angles θ being are effective to merge illumination from multiple LED pixels in the LED array on the directly lit diffuser panel surface. A controller is connected to the LED array having circuitry to control the LED array in response to image data to induce display via the diffuser panel of a time varying image with spatially varying colors and intensities.
-
公开(公告)号：US20190122419A1
公开(公告)日：2019-04-25
申请号：US16225273
申请日：2018-12-19
CPC分类号： G06T15/06 , G06T15/005 , G06T15/50 , G06T15/80
摘要： Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.
-
公开(公告)号：US20190066391A1
公开(公告)日：2019-02-28
申请号：US16118200
申请日：2018-08-30
申请人： GO GHOST, LLC
发明人： Erik ANDERSON , Mathew FRIDE
CPC分类号： G06T19/20 , G06F16/51 , G06F2221/2107 , G06T15/06 , G06T15/50 , G06T2219/2004 , G06T2219/2012 , H04L9/0861
摘要： A method of directly modifying ray tracing samples generated by a ray tracing renderer. Modifications to samples may be made after rendering and before rasterizing, in contrast to typical compositing workflows that manipulate pixels of a rasterized image. Modifications may be based on user input. Rasterization may be performed afterwards at any desired resolutions, for example to adapt to different displays. Samples may be tagged with object identities, facilitating object selection without the need for object masks. Pseudo-random ray patterns typically used by renderers may be supported directly. Many operations may be performed directly on samples, including color changes, object repositioning, and merging of samples from different scenes. Secure samples with scrambled ray directions may be modified directly. Modified samples may be compressed for storage or transmission, and may be streamed to receiving devices that rasterize the samples for display, possibly after decompressing or decrypting the received streams.
-
公开(公告)号：US20190005710A1
公开(公告)日：2019-01-03
申请号：US16008859
申请日：2018-06-14
发明人： Steven Richard Irrgang , Thai Quan Huynh-Thu , Juno Kim , Vanessa Jeanie Honson , Matthew Raphael Arnison
CPC分类号： G06T15/50 , G06T15/04 , G06T15/506
摘要： A method of rendering a graphical object comprises accessing a mapping relating a mesoscale structure and a light scattering parameter of a material to a perceptual appearance characteristic; determining a perceptual appearance characteristic of the graphical object, the graphical object reproduced on an interface to represent an object formed from the material, the perceptual appearance characteristic determined in accordance with the mapping using an initial mesoscale structure and a light scattering parameter of the material; receiving a signal indicating a modification in structure relating to the initial mesoscale structure; and determining, using the mapping, an adjustment of the light scattering parameter preserving the determined perceptual appearance characteristic, based on the modification of the initial mesoscale structure. The method further comprises adjusting the light scattering parameter of the material according to the adjustment; and rendering the graphical object using the modified mesoscale structure and the adjusted light scattering parameter.
-
公开(公告)号：US20180310025A1
公开(公告)日：2018-10-25
申请号：US15958005
申请日：2018-04-20
发明人： Jaakko KERÄNEN , Kimmo ROIMELA , Johannes PYSTYNEN
IPC分类号： H04N19/597 , H04N13/161 , G06T15/08 , G06T19/00
CPC分类号： H04N19/597 , G06T7/55 , G06T15/08 , G06T15/50 , G06T17/005 , G06T19/006 , H04N13/161
摘要： A method and technical equipment are provided. The method comprises receiving media content comprising images and depth information; generating a sparse voxel octree from the received images and depth information; projecting the received image colors to each voxel of the generated sparse voxel octree based on visibility; converting the received images into a set of radiance samples for each voxel of the generated sparse voxel octree; and processing the radiance samples by performing one of the following: fitting the radiance samples to a parametric color model to generate estimated radiance samples for a current viewing direction; or analyzing the radiance samples with the content of the generated sparse voxel octree to separate an actual surface color and reflectance properties from a reflected lighting.
-
公开(公告)号：US20180293792A1
公开(公告)日：2018-10-11
申请号：US15946572
申请日：2018-04-05
摘要： This disclosure provides a method for providing spatial gradients of 3D CAD models of solid objects with respect to various parameters used in defining the solid object. A modeler is designed and implemented to provide spatial gradients of solid objects (when represented on a computer as 3D CAD models) with respect to the various parameters that define the objects. Although primarily based on finite difference operations, several additional steps are performed to ensure that the gradients computed are correct and valid for a general class of solid objects and the parameters that define the objects. No commercially available 3D CAD tool provides such ability for a general class of solid objects.
-
公开(公告)号：US10008030B2
公开(公告)日：2018-06-26
申请号：US15252863
申请日：2016-08-31
发明人： Hyun Sung Chang , Young Ju Jeong , Hyoseok Hwang
CPC分类号： G06T15/50 , G06T7/30 , G06T2207/10004 , H04N13/302 , H04N13/327 , H04N13/351 , H04N13/373
摘要： A panel image that may be displayed in a display apparatus to display a 3D image that includes reduced differences with a source image may be generated based on generating a predicted 3D image based on a panel image, comparing the predicted 3D image to the source image, and adjusting the panel image to reduce differences between the predicted 3D image and the source image. Such a process may be performed iteratively. The predicted 3D image may be generated based on applying a convolution kernel to the panel image in a convolution operation. The convolution kernel may be based on a first function and a second function. The first function may be associated with a brightness distribution of one or more visual fields of the panel image. The second function may be associated with a Gaussian distribution of brightness of a panel image pixel.
-
-
-
-
-
-
-
-
-
搜索结果
1261 条记录 (耗时 0.032 秒)
