A three-dimensional Gaussian scene color rendering method and computer device

Abstract

The present invention relates to the field of three-dimensional Gaussian scene color rendering, and in particular to a three-dimensional Gaussian scene color rendering method and computer device. The method comprises: converting the Gaussian ellipsoid of the RGB color space into the HSL color space, extracting the basic hue of each Gaussian ellipsoid; solving the color when the Gaussian ellipsoid is observed from different angles, using the basic hue, and the saturation and brightness of the observation point through the combination of spherical radial basis functions to represent the color; dividing the space of user activities according to fixed length, width and height, and obtaining the index of the Gaussian ellipsoid seen from different positions and different angles in the space block and the index of the SRBF seen in the space block through calculation preprocessing for the divided space blocks; when the user accesses the 3D Gaussian ellipsoid resource online, it is finally rendered to the user through real-time calculation, transmission and loading. The present invention is suitable for three-dimensional Gaussian scene color rendering.

Description

Technical Field

[0001] The invention relates to the field of three-dimensional Gaussian scene color rendering, and in particular to a three-dimensional Gaussian scene color rendering method and a computer device. Background Art

[0002] With the development of virtual reality technology, real-time rendering and transmission of three-dimensional scenes has become a research hotspot. Among the many three-dimensional scene representation methods, 3DGS (3D Gaussian Splatting) has attracted much attention due to its excellent rendering quality and efficiency. 3DGS uses Gaussian ellipsoids as basic units, and each Gaussian unit has unique properties such as position, rotation, scale, and color.

[0003] At present, 3DGS mainly uses SH (Spherical Harmonics) to express the color attributes of Gaussian units. Although SH expression has certain advantages in mathematics, it has the following problems in practical applications:

[0004] 1) SH needs to transmit 48 parameters for each Gaussian unit to represent the complete spherical color distribution, no matter how much is actually visible;

[0005] 2) The global nature of SH does not match the local perception of users in the interactive environment, resulting in inefficient data transmission;

[0006] 3) The color and lighting effects of SH are mixed together and difficult to separate and optimize.

[0007] Therefore, a new Gaussian unit color rendering method is needed that can accurately express the viewing angle-dependent color effects and achieve efficient data transmission. Summary of the invention

[0008] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a three-dimensional Gaussian scene color rendering method and a computer device to achieve efficient and separable color rendering.

[0009] The present invention adopts the following technical solutions to achieve the above-mentioned purpose. In a first aspect, the present invention provides a three-dimensional Gaussian scene color rendering method, comprising:

[0010] S1, converting the Gaussian ellipsoid of the RGB color space into the HSL color space, and extracting the basic hue of each Gaussian ellipsoid;

[0011] S2, solving the color when observing the Gaussian ellipsoid from different angles, using the basic hue, and the saturation and brightness of the observation point fitted by a combination of spherical radial basis functions to represent the color;

[0012] S3. Segment the user activity space according to fixed length, width and height. For the segmented space blocks, obtain the index of Gaussian ellipsoids at different positions and angles in the space block through calculation preprocessing, and record the index of SRBF (Spherical Radial Basis Function) in the space block.

[0013] S4. When users access 3D Gaussian ellipsoid resources online, the Gaussian ellipsoid and SRBF index are combined, calculated, transmitted and loaded in real time, and finally rendered to the user.

[0014] Furthermore, step S4 specifically includes:

[0015] S401, initialization stage;

[0016] Initialize the client, initialize the rendering engine, and initialize the data receiving buffer;

[0017] Initialize the server, load 3D scene data, load pre-calculated visibility tables, and initialize the priority transmission queue;

[0018] S402, first scene loading;

[0019] The client sends an initial request to obtain the current position coordinates and viewing direction, and packages the obtained data to send to the server;

[0020] The server receives and processes the initial request sent by the client, calculates the spatial block index based on the position coordinates, and obtains the pre-calculated visibility data corresponding to the spatial block;

[0021] Calculate the transfer priority:

[0022] The first priority is the Gaussian ellipsoid in the viewing cone. The Gaussian ellipsoid in the viewing cone and the SRBF list related to the user's viewing angle are obtained and sent to the transmission queue according to the label of the first priority.

[0023] The second priority is the Gaussian ellipsoid visible in the current space block. The visibility table is queried, and the visible Gaussian and its related SRBF are read according to the table, and sent to the transmission queue according to the second priority label;

[0024] The third priority is the Gaussian ellipsoid visible to the adjacent spatial blocks. The adjacent spatial block list is calculated, and the visibility table is queried to obtain the visibility data of the adjacent spatial blocks, which are sent to the transmission queue according to the third priority label;

[0025] Data Packaging:

[0026] According to the priority in the transmission queue, read the header information, and query whether the basic information of the Gaussian ellipsoid has been sent through the unique Gaussian ellipsoid index. For the first transmission, pack the basic data of the Gaussian ellipsoid, including: basic color tone, position, direction and scale;

[0027] According to the unique Gaussian ellipsoid index, query the SRBF parameters related to the viewing angle. If the relevant SRBF parameters have not been sent, send the relevant SRBF parameters, including the relevant weight and direction parameters;

[0028] Record the packaged data and identify the basic information and SRBF parameter information that have been sent according to the unique Gaussian ellipsoid index;

[0029] Send the packaged data to the client;

[0030] S403, client rendering loop;

[0031] The client receives the data packet sent by the server, parses the Gaussian ellipsoid data, parses the SRBF parameters, and updates the scene cache;

[0032] Get the current viewing angle parameters, calculate the SRBF color, render the Gaussian ellipsoid and display the image;

[0033] Periodically send changes in position and view angle to trigger update requests.

[0034] In a second aspect, the present invention provides a computer device, including a memory, wherein the memory stores program instructions, and when the program instructions are executed, the three-dimensional Gaussian scene color rendering method described above is executed.

[0035] The beneficial effects of the present invention are:

[0036] The present invention separates primary colors and lighting effects through the HSL color space, thereby reducing data redundancy.

[0037] The present invention adopts spherical radial basis functions to model illumination changes, thereby achieving efficient and controllable color expression.

[0038] The present invention supports selective transmission based on viewing angle, thereby improving data transmission efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of using spherical radial basis functions to model illumination changes and achieve efficient and controllable color expression according to an embodiment of the present invention;

[0040] Figure 2 It is a flow chart of a three-dimensional Gaussian scene color rendering method provided by an embodiment of the present invention;

[0041] Figure 3This is a flow chart of a user accessing a 3D Gaussian ellipsoid resource online provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0043] The present invention provides a three-dimensional Gaussian scene color rendering method, such as Figure 2 As shown, including:

[0044] S1, converting the Gaussian ellipsoid of the RGB color space into the HSL color space, and extracting the basic hue of each Gaussian ellipsoid;

[0045] S2, solving the color when observing the Gaussian ellipsoid from different angles, using the basic hue, and the saturation and brightness of the observation point fitted by a combination of spherical radial basis functions to represent the color;

[0046] S3, the user activity space is divided according to the fixed length, width and height, and the index of the Gaussian ellipsoid at different positions and angles in the space block is obtained through calculation preprocessing for the divided space block, and the index of the SRBF in the space block is recorded at the same time;

[0047] S4. When users access 3D Gaussian ellipsoid resources online, they are calculated, transmitted and loaded in real time and finally rendered to the users.

[0048] Specifically, Figure 3 As shown, step S4 specifically includes:

[0049] S401, initialization stage;

[0050] Initialize the client, initialize the rendering engine, and initialize the data receiving buffer;

[0051] Initialize the server, load 3D scene data, load pre-calculated visibility tables, and initialize the priority transmission queue;

[0052] S402, first scene loading;

[0053] The client sends an initial request to obtain the current position coordinates and viewing direction, and packages the obtained data to send to the server;

[0054] Determine the spatial block where the user is located: The server receives and processes the initial request sent by the client, calculates the spatial block index based on the location coordinates, and obtains the pre-calculated visibility data corresponding to the spatial block;

[0055] Calculate the transfer priority:

[0056] The first priority is the Gaussian ellipsoid in the viewing cone. The Gaussian ellipsoid in the viewing cone and the SRBF list related to the user's viewing angle are obtained and sent to the transmission queue according to the label of the first priority.

[0057] The second priority is the Gaussian ellipsoid visible in the current space block. The visibility table is queried, and the visible Gaussian and its related SRBF are read according to the table, and sent to the transmission queue according to the second priority label;

[0058] The third priority is the Gaussian ellipsoid visible to the adjacent spatial blocks. The adjacent spatial block list is calculated, and the visibility table is queried to obtain the visibility data of the adjacent spatial blocks, which are sent to the transmission queue according to the third priority label;

[0059] Data Packaging:

[0060] According to the priority in the transmission queue, read the header information, and query whether the basic information of the Gaussian ellipsoid has been sent through the unique Gaussian ellipsoid index. For the first transmission, pack the basic data of the Gaussian ellipsoid, including: basic color tone, position, direction and scale;

[0061] According to the unique Gaussian ellipsoid index, query the SRBF parameters related to the viewing angle. If the relevant SRBF parameters have not been sent, send the relevant SRBF parameters, including the relevant weight and direction parameters;

[0062] Record the packaged data and identify the basic information and SRBF parameter information that have been sent according to the unique Gaussian ellipsoid index to prevent repeated transmission.

[0063] Send the packaged data to the client;

[0064] S403, client rendering loop;

[0065] Data reception and analysis: The client receives the data packet sent by the server, analyzes the Gaussian ellipsoid data, analyzes the SRBF parameters, and updates the scene cache;

[0066] Rendering process: Get the current viewing angle parameters, calculate the SRBF color, render the Gaussian ellipsoid and display the picture;

[0067] Position and view monitoring: Regularly send position and view changes to trigger update requests.

[0068] Scene update loop:

[0069] The client sends an update request and the server processes the update request.

[0070] Client exit:

[0071] The client terminates rendering, sends an exit request, clears the cache, and exits the program.

[0072] The server receives the termination request, cleans up the sending queue and cache (3D scene data, loaded visibility table), and terminates the conversation.

[0073] The present invention first separates the essential attributes of color and the lighting effects through the HSL color space. Figure 1 As shown in the figure, HSL stands for Hue, Saturation, Lightness, which is a method of representing points in the RGB color model in a cylindrical coordinate system. It is one of the most widely used color systems. The HSL space uses three components, hue (H), saturation (S), and brightness (L), to represent color, where hue represents the dominant wavelength, saturation represents color intensity, and brightness represents the degree of lightness and darkness.

[0074] For each Gaussian unit, define its color as:

[0075] ;

[0076] in is the hue, which is a fixed value. are reference saturation and brightness values ​​under neutral lighting conditions, respectively. In the method of the present invention, spherical radial basis functions are used to express them.

[0077] like Figure 1 As shown, the present invention uses spherical radial basis function (SRBF) to express saturation and brightness values. A single SRBF is defined as:

[0078] ;

[0079] in is a unit vector on the sphere, is the center direction of SRBF, is the shape parameter.

[0080] The linear combination of multiple SRBFs can approximate any spherical function :

[0081] ;

[0082] in is the weight coefficient.

[0083] The present invention adopts the MP (Matching Pursuit) matching pursuit algorithm to solve the optimal SRBF parameter combination. The MP algorithm is an iterative greedy algorithm that approximates the target function by gradually selecting the best matching basis function.

[0084] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art shall not deviate from the spirit and scope of the present invention, and shall be within the scope of protection of the claims attached to the present invention.

Claims

1. A three-dimensional Gaussian scene color rendering method, characterized in that: include: S1, converting the Gaussian ellipsoid of the RGB color space into the HSL color space, and extracting the basic hue of each Gaussian ellipsoid; S2, solving the color when observing the Gaussian ellipsoid from different angles, using the basic hue, and the saturation and brightness of the observation point fitted by a combination of spherical radial basis functions to represent the color; S3, the user activity space is divided according to the fixed length, width and height, and the index of the Gaussian ellipsoid at different positions and angles in the space block is obtained through calculation preprocessing for the divided space block, and the index of the SRBF in the space block is recorded at the same time; S4, when the user accesses the 3D Gaussian ellipsoid resource online, the Gaussian ellipsoid and the SRBF index are combined, and the 3D Gaussian ellipsoid is calculated, transmitted and loaded in real time, and finally rendered to the user; Step S4 specifically includes: S401, initialization stage; Initialize the client, initialize the rendering engine, and initialize the data receiving buffer; Initialize the server, load 3D scene data, load pre-calculated visibility tables, and initialize the priority transmission queue; S402, first scene loading; The client sends an initial request to obtain the current position coordinates and viewing direction, and packages the obtained data to send to the server; The server receives and processes the initial request sent by the client, calculates the spatial block index based on the position coordinates, and obtains the pre-calculated visibility data corresponding to the spatial block; Calculate the transfer priority: The first priority is the Gaussian ellipsoid in the viewing cone. The Gaussian ellipsoid in the viewing cone and the SRBF list related to the user's viewing angle are obtained and sent to the transmission queue according to the label of the first priority. The second priority is the Gaussian ellipsoid visible in the current space block. The visibility table is queried, and the visible Gaussian and its related SRBF are read according to the table, and sent to the transmission queue according to the second priority label; The third priority is the Gaussian ellipsoid visible to the adjacent spatial blocks. The adjacent spatial block list is calculated, and the visibility table is queried to obtain the visibility data of the adjacent spatial blocks, which are sent to the transmission queue according to the third priority label; Data Packaging: According to the priority in the transmission queue, read the header information, and query whether the basic information of the Gaussian ellipsoid has been sent through the unique Gaussian ellipsoid index. For the first transmission, pack the basic data of the Gaussian ellipsoid, including: basic color tone, position, direction and scale; According to the unique Gaussian ellipsoid index, query the SRBF parameters related to the viewing angle. If the relevant SRBF parameters have not been sent, send the relevant SRBF parameters, including the relevant weight and direction parameters; Record the packaged data and identify the basic information and SRBF parameter information that have been sent according to the unique Gaussian ellipsoid index; Send the packaged data to the client; S403, client rendering loop; The client receives the data packet sent by the server, parses the Gaussian ellipsoid data, parses the SRBF parameters, and updates the scene cache; Get the current viewing angle parameters, calculate the SRBF color, render the Gaussian ellipsoid and display the image; Periodically send changes in position and view angle to trigger update requests.

2. A computer device comprising a memory, wherein the memory stores program instructions, characterized in that: When the program instructions are executed, the three-dimensional Gaussian scene color rendering method as claimed in claim 1 is executed.

Images