25 results about "Texture space" patented technology

This disclosure provides systems, devices, apparatus, and methods, including computer programs encoded on storage media, for a dicing oracle for texture space shading. A processor obtains an indication of a UV parameterization for each of a set of geometry units and a target number of pixels per region of the UV parameterization, where the UV parameterization includes a first set of UV coordinates for an object space. The processor renders the set of geometry units in order to obtain a second set of UV coordinates for a screen space and a set of derivatives for the second set of UV coordinates. The processor calculates a resolution for a mip region map based on the set of derivatives and the target number of pixels per region. The processor outputs an indication of the calculated resolution.

The invention relates to a graphics processor, an image processing method and electronic equipment, and relates to the technical field of image processing. The graphics processor comprises a gradient LOD calculation circuit and a shader. The LOD calculation circuit is used for obtaining the gradient LOD of a plurality of effective pixels in the image frame in the texture space, and the effective pixels are used for forming primitives in the image; and the shader is used for coloring the plurality of effective pixels according to the LOD of the plurality of effective pixels. Therefore, computing resources can be saved.

This disclosure provides systems, devices, apparatus, and methods, including computer programs encoded on storage media, for sampler feedback for texture space shading. A processor may assign each shading element in a set of shading elements associated with a geometry unit to a corresponding shading element group in a set of shading element groups. The processor may identify whether each shading element is a visible shading element. The processor may calculate a group offset value for each shading element group in the set of shading element groups based on a number of visible shading elements in the set of shading elements. The processor may calculate an element offset value for each shading element based on a number of visible shading elements. The processor may allocate, based on the element offset value and the group offset value, texture memory to the visible shading elements.

A high-precision mapping method for an underground poor texture space based on mobile-fixed collaborative deviation correction includes: acquiring sensor data of sensor nodes determined based on the structure of the underground poor texture space; calculating the initial position data of an UAV; receiving sensor data through the UAV, and generating position data of the UAV by combining the initial position data of the UAV and lidar detection data of the UAV; correcting the position data of the UAV through a segmented mobile-fixed collaborative odometer; acquiring point cloud data for describing the structure of underground poor texture space; performing data fusion on the sensor data, the initial position data of the UAV, the corrected position data of the UAV and the point cloud data to obtain fused data, and establishing a high-precision map of the underground poor texture space, with each sensor node used as a feature point in the mapping.

This invention discloses a visual localization method based on pose space texture continuity mapping, belonging to the fields of robotics and computer vision. In the mapping stage, image intrinsic parameters, pose, and sparse point cloud are first obtained through sparse reconstruction. Image features are extracted using a pre-trained model, and pose encoding and decoding models are trained to ensure consistency between pose encoding and image features in spatial metrics. Simultaneously, the decoding model outputs a multi-peak distribution to model pose ambiguity. Subsequently, a mapping model from image features to pose encoding is trained and optimized end-to-end to construct a scene map. In the localization stage, features of the image to be localized are extracted, pose encoding is obtained through the mapping model, and the pose distribution is output by the decoding model. A unique location or multiple candidate poses are determined based on the number of peaks in the distribution. This invention effectively improves the robustness and accuracy of robot localization in repetitive texture scenes by maintaining the continuity alignment between the pose space and texture space.

Example embodiments relate to techniques for volumetric performance capture with neural rendering. A technique may involve initially obtaining images that depict a subject from multiple viewpoints and under various lighting conditions using a light stage and depth data corresponding to the subject using infrared cameras. A neural network may extract features of the subject from the images based on the depth data and map the features into a texture space (e.g., the UV texture space). A neural renderer can be used to generate an output image depicting the subject from a target view such that illumination of the subject in the output image aligns with the target view. The neural render may resample the features of the subject from the texture space to an image space to generate the output image.

The invention relates to a real-time texture drawing method and system based on a lookup table, and the method comprises the steps: matching a texture space sampling point coordinate corresponding to each element obtained from a pre-constructed initial lookup table with a texture space vertex coordinate of each triangular primitive in an imported three-dimensional model; obtaining a triangular primitive matched with each element, obtaining a model space sampling point coordinate corresponding to each element according to the matched triangular primitive, and storing the model space sampling point coordinates corresponding to all the elements into an initial lookup table to obtain a final lookup table; according to drawing parameters input by a user, a drawing area is screened out from the final lookup table, texture color writing is conducted on the three-dimensional model according to texture space sampling point coordinates corresponding to elements existing in the drawing area, and real-time texture drawing is achieved. Therefore, accurate texture drawing is realized.

The invention discloses a special effect rendering method, device and equipment, a medium and a product, and the method comprises the steps: determining an initial directed acyclic graph corresponding to an initial rendering link according to the initial rendering link of a special effect; according to the initial directed acyclic graph, determining at least one reference rendering link and a texture number corresponding to the at least one reference rendering link; wherein the at least one reference rendering link comprises rendering links, corresponding to the initial directed acyclic graph, of various different texture numbers, and the texture number corresponding to the reference rendering link represents the number of minimum texture spaces needing to be occupied in the execution process of the reference rendering link; the target rendering link is determined from the at least one reference rendering link according to the texture number corresponding to the at least one reference rendering link, and the special effect is rendered according to the target rendering link, so that the number of texture spaces allocated to special effect rendering in the memory resources is reduced, and the rendering efficiency is improved. The problem that the occupation amount of special effect rendering on memory resources is too high is solved.

A rendering controller configured to render lights in a three-dimensional (3D) scene is disclosed. The 3D scene includes one or more objects. Each object is associated with a plurality of on-surface reservoirs. The rendering controller is further configured to provide light transport computation in texture space utilizing resampled importance sampling (RIS) or weighted reservoir sampling (WRS) based on the on-surface caches. The rendering controller is configured to compute shading efficiently (i.e., direct, and indirect illumination) of the 3D scene in multi-viewer applications present in the cloud, which involves complex lighting scenarios.

The invention relates to the technical field of image processing, in particular to a plankton image enhancement method based on a deep learning model. The method comprises the following steps: acquiring an original image of plankton to be enhanced; analyzing the original image through a deep learning model, and extracting plankton shell grains and flagellum to form periodic texture space arrangement characteristics and spectral absorption characteristics of a chlorophyll-dominated pigment region; and positioning a texture dense region based on the texture space arrangement features, performing pixel-by-pixel phase alignment calibration on the texture dense region, and optimizing the texture features of the texture dense region through feature superposition to form a texture optimization region. Through the deep learning model and the image processing enhancement technology, accurate optimization of the plankton image texture features and natural restoration of the color attributes are realized, so that the detail identification degree, the color authenticity and the overall integrity of the plankton image are improved.

In order to provide efficient triangle index mapping for object spatial shading, geometric processing comprises: dividing UV coordinate data of an object to be shaded into UV layers such that each UV layer comprises non-overlapping triangles with coordinates in the range of 0 to 1; triangular index (index, ID) information is generated for the UV layer to be used for spatial shading of the object, and ray tracing of triangular IDs is performed for sample points in the UV layer. Based thereon, a unique texture space having non-overlapping triangles in the range of 0 to 1 is provided to store texture space shading. A rendering device for such geometric processing is beneficial to providing an efficient solution to construct an object space shading framework.

The embodiment of the invention provides a scene rendering method and device, equipment and a storage medium. The method comprises the following steps: determining a corresponding visible texel set according to a mapping texel point of each pixel point in a current to-be-rendered frame in a target texture image of a specified rendering scene; the visible texel sets are reported to a server side, so that the server side combines the visible texel sets reported by multiple terminals into a visible texel union set, and the visible texel union set is issued to each terminal through a reference texture image subjected to texture rendering; and according to the reference texture image, rendering the current to-be-rendered frame and then displaying the current to-be-rendered frame. According to the embodiment of the invention, one-time efficient texture rendering and texture rendering multiplexing of multiple terminals in the texture space for the specified rendering scene can be realized, the repeated texture rendering overhead of the multiple terminals for each rendering of the specified rendering scene is greatly reduced, and the rendering frame rate and smooth display of each terminal for the specified rendering scene are improved.

This disclosure provides systems, apparatus, devices, and methods for a tessellation oracle for texture space shading, including computer programs encoded on a storage medium. A processor obtains an indication of a UV parameterization for each geometry unit in a set of geometry units and a target number of pixels for each region of the UV parameterization, wherein the UV parameterization includes a first set of UV coordinates of an object space. The processor renders the set of geometry units to obtain a second set of UV coordinates of the screen space and a set of derivatives of the second set of UV coordinates. The processor calculates a resolution of the mip region map based on the set of derivatives and the number of target pixels for each region. The processor outputs an indication of the calculated resolution.

This disclosure provides systems, devices, apparatuses, and methods, including computer programs encoded on a storage medium, for sampler feedback for texture space shading. The processor may assign each shading element of the set of shading elements associated with the geometry unit to a corresponding group of shading elements of the set of groups of shading elements. The processor may identify whether each coloring element is a visible coloring element. The processor may calculate a group offset value for each of the set of pigmented elements based on a number of visible pigmented elements of the set of pigmented elements. The processor may calculate an element offset value for each colored element based on the number of visible colored elements. The processor may assign a texture memory to the visible colored element based on the element offset value and the set of offset values.

Embodiments are disclosed for neural texture mapping. In some embodiments, a method of neural texture mapping includes obtaining a plurality of images of an object, determining volumetric representation of a scene of the object using a first neural network, mapping 3D points of the scene to a 2D texture space using a second neural network, and determining radiance values for each 2D point in the 2D texture space from a plurality of viewpoints using a second neural network to generate a 3D appearance representation of the object.

The invention provides a sampler texture feedback method, a graphics processor, electronic equipment and a storage medium, and relates to the technical field of graphics processing. The method comprises the following steps: receiving and analyzing a sampler feedback instruction through a shader processing module to obtain texture sampling data for feedback processing; and a texture sampling module reuses a texture sampling path based on texture sampling data to execute texture space mapping, generates a texture hierarchy data storage address, and writes a request to update a target cache unit by using a texture address with a feedback flag bit, so that the target cache unit records the use condition of a corresponding texture region and hierarchy. According to the technical scheme, under the condition that independent hardware and high instruction overhead are not increased, the texture access statistics and the texture sampling process can be cooperatively completed on a hardware assembly line, so that the texture feedback data generation efficiency can be improved, the access pressure of a feedback path can be reduced, and the accuracy of the texture level use condition can be enhanced.

Techniques and systems are provided for three-dimensional model segmentation. For instances, a process can include: rendering a three-dimensional (3D) model of an object based on a 3D mesh model and a texture of the object; generating a view of the 3D model based on the rendered 3D model; generating a texture space mask based on the texture of the object using a first machine learning (ML) model; generating a labeled mask based on a second ML model; assigning labels to the texture space mask based on the labeled mask to obtain a labeled texture space mask; and generating a 3D mask based on the labeled texture space mask.

Approaches presented herein can utilize a network that learns to embed three-dimensional (3D) coordinates on a surface of one or more 3D shapes into an aligned two-dimensional (2D) texture space, where corresponding parts of different 3D shapes can be mapped to the same location in a texture image. Alignment can be performed using a texture alignment module that generates a set of basis images for synthesizing textures. A trained network can generate a basis shared by all shape textures, and can predict input-specific coefficients to construct the output texture for each shape as a linear combination of the basis images, then deform the texture to match the pose of the input. Such an approach can ensure alignment of textures, even in situations with at least somewhat limited network capacity. To unwrap shapes of complex structure or topology, a masking network can be utilized that cuts the shape into multiple pieces to reduce the distortion in the 2D mapping.

The invention discloses a gift 3D model lightweight processing method for terminal loading, and relates to the technical field of data processing, and the method comprises the steps: taking a curvature concentration region at a boundary line intersection as a topology constraint reference, generating a topology connection constraint path, and limiting a local deformation conduction range according to the topology connection constraint path; constructing a texture mapping control field, and identifying a high distortion area of texture mapping in the texture mapping control field; deducing a self-adaptive adjustment matrix of geometric topology according to the topological adjacency relation of the high-distortion area, and controlling relocation of a topological patch according to the self-adaptive adjustment matrix; generating a texture space resampling mask according to the texture coordinate change after the topological surface patch is repositioned, limiting a local mapping reconstruction area, executing texture compression processing, and outputting a gift three-dimensional model; according to the method, collaborative optimization of local geometry and texture when the gift three-dimensional model is loaded at the terminal is realized, and the lightweight precision and the visual reduction effect of the model are effectively improved.

This specification discloses a method, apparatus, storage medium, and electronic device for following visual offset. First, an input vector corresponding to the original pose data of a target object is obtained, wherein the target object corresponds to at least one sub-logical object. Then, the vertical component of the texture coordinates of the sub-logical object in the texture space of the target object is obtained at the CPU. Based on the vertical component and preset manual correction parameters, a representative depth value of the sub-logical object is obtained. Based on the input vector, the representative depth value, the focal plane depth value, and the global parallax scaling factor, the texture space offset corresponding to the sub-logical object is obtained. The texture space offset is inversely projected from the texture space to world space to obtain the corresponding world coordinate displacement. Based on the original reference position of the sub-logical object and the world coordinate displacement, the final position of the sub-logical object is obtained, and the final position is assigned to the transformation component of the sub-logical object so that the sub-logical object follows the visual offset of the target object.

The embodiment of the invention relates to a three-dimensional head portrait generation method and device, equipment, a storage medium and a program product. The method comprises the following steps: acquiring a target two-dimensional image containing a head portrait, inputting the target two-dimensional image into a three-dimensional head portrait generation model, and generating a multi-channel texture image and a three-channel position offset image in a two-dimensional texture space; generating an initial head portrait grid model based on the target two-dimensional image, and generating an initial two-dimensional position image based on the initial head portrait grid model; superposing the three-channel position offset image based on the initial two-dimensional position image to generate a target two-dimensional position image, and generating a target head portrait grid model based on the target two-dimensional position image; and performing head portrait rendering based on the multi-channel texture image and the target head portrait grid model, and generating a full-view three-dimensional head portrait asset corresponding to the target two-dimensional image. Therefore, the fine degree of the three-dimensional head portrait asset and the integrity of the three-dimensional head portrait asset in each view angle are improved, and the consistency between the three-dimensional head portrait asset and the input two-dimensional head portrait image is improved.

The present application belongs to the technical field of image processing, and provides a double-branch color point cloud 3D texture grid reconstruction method, device and electronic equipment, which comprises: performing multi-view projection and grid reconstruction on the input color point cloud respectively to obtain multi-view sparse images and non-texture 3D grid; performing multi-view consistency repair on the multi-view sparse images to generate an initial UV texture map; performing texture repair on the initial UV texture map based on back projection and weighted mixing algorithm to generate a complete UV texture map; performing texture resolution enhancement processing on the complete UV texture map to obtain a final UV texture map; and pasting the final UV texture map back to the surface of the non-texture 3D grid to generate a 3D texture grid. The scheme cooperatively optimizes the image space and the texture space through a double-branch framework, uses multi-view consistency repair to improve the texture definition and global consistency, and improves the definition of the texture map through texture repair and texture resolution enhancement.

This application discloses a method and related products for optimizing the texture space of 3D model data. The method includes acquiring 3D model data to be optimized; performing preprocessing operations on the 3D model data to obtain preprocessed 3D model data; globally parameterizing the preprocessed 3D model data based on harmonic mapping to obtain its corresponding new texture coordinates; and filling the new texture space formed by the new texture coordinates using scanline filling based on the texture information of the 3D model data, thereby optimizing the texture space of the 3D model data. Using the solution of this application, texture space optimization can be achieved efficiently, and a new, compact texture space can be generated.

To provide efficient triangle index mapping for an object-space shading, a geometry processing includes splitting UV coordinate data of an object to be shaded into UV layers so that each UV layer comprises non-overlapping triangles with coordinates within a 0-1 range, and generating triangle index, ID, information for the UV layers to be used for the object-space shading with a ray tracing of triangle IDs for sample points in the UV layers. As a result, a unique texture space with non-overlapping triangles within the 0-1 range is provided to store texture-space shading. A rendering apparatus configured for such geometry processing is beneficial to provide an efficient solution for building an object-space shading framework.