3D scene mesh generation and editing device

The space generation and editing device addresses 3D modeling inefficiencies by modularizing tasks and integrating specialized algorithms, improving precision and reducing distortion in 3D models.

KR103047810B1Active Publication Date: 2026-11-11SOGANG UNIV RES & BUSINESS DEV FOUND
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
KR1020240103661
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2026-11-11
Estimated Expiration
2044-08-05

AI Technical Summary

Technical Problem

Conventional 3D modeling technologies suffer from significant distortion due to combining individual camera poses into a whole model without proper modularization of algorithms, leading to inefficiencies in space generation and editing.

Method used

A space generation and editing device that breaks down tasks into detailed modules, employs specialized algorithms for each task, and integrates them using a framework, including a space providing unit, object unit, coordinate generation, and image providing units to enhance precision.

Benefits of technology

Improves 3D model quality by modularizing algorithms and integrating them effectively, reducing distortion and enhancing the accuracy of space generation and editing processes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112024084732497-PAT00001_ABST
    Figure 112024084732497-PAT00001_ABST
Patent Text Reader

Abstract

A space generation and editing device according to an embodiment of the present invention may include a space providing unit and an object unit. The space providing unit may provide a three-dimensional space based on a first instruction among a plurality of instructions. The object unit may place an object in the three-dimensional space according to a second instruction among the plurality of instructions. The space generation and editing device according to the present invention can improve problems occurring in existing 3D models by breaking down space generation and editing into detailed tasks, modularizing algorithms specialized for each detailed task, and building a framework that integrates them.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to a space generation and editing device. Background Technology

[0002] Conventional technologies combined parts of space generated individually from various camera poses into a whole 3D model, resulting in significant distortion in the resulting 3D model. Recently, various studies have been conducted to address this problem. Prior art literature

[0003] (Korean Registered Patent) No. 10-2272650 (Registration Date: June 29, 2021) The problem to be solved

[0004] The technical problem that the present invention aims to solve is to provide a space generation and editing device that can improve the problems occurring in existing 3D models by breaking down space generation and editing into detailed tasks, modularizing algorithms specialized for each detailed task, and building a framework that integrates them. means of solving the problem

[0005] To solve these problems, a space generation and editing device according to an embodiment of the present invention may include a space providing unit and an object unit. The space providing unit may provide a three-dimensional space based on a first instruction among a plurality of instructions. The object unit may place an object in the three-dimensional space according to a second instruction among the plurality of instructions.

[0006] In one embodiment, the space generation and editing device may further include a coordinate generation unit. The coordinate generation unit can generate coordinates corresponding to the three-dimensional space.

[0007] In one embodiment, the space generation and editing device may further include an image providing unit. The image providing unit may provide a plurality of basic images corresponding to the three-dimensional space.

[0008] In one embodiment, the image providing unit may further include a selection unit. The selection unit may select from the plurality of basic images and provide a selected image.

[0009] In one embodiment, the first selected image among the selected images may be a border image corresponding to the border of the three-dimensional space formed according to the coordinates of the three-dimensional space.

[0010] In one embodiment, the second selected image among the selected images may be a depth image that expresses the depth of the three-dimensional space formed according to the coordinates of the three-dimensional space in grayscale.

[0011] In one embodiment, the third selected image among the selected images may be a semantic image that distinguishes the floor, walls, and windows in the three-dimensional space according to color.

[0012] In one embodiment, the space generation and editing device may further include a plurality of modules. The plurality of modules may provide the selected images corresponding to the plurality of instructions.

[0013] In one embodiment, the space generation and editing device can learn based on the result of compressing the coordinates corresponding to the three-dimensional space in a predetermined manner.

[0014] In one embodiment, the space generation and editing device can learn based on the square of the absolute value of the difference between the loss value and the result value generated from the space generation and editing device.

[0015] To solve these problems, a space generation and editing system according to an embodiment of the present invention may include a space providing unit, an object unit, and a correction unit. The space providing unit may provide a three-dimensional space based on a first instruction among a plurality of instructions. The object unit may place an object in the three-dimensional space according to a second instruction among the plurality of instructions. The correction unit may further correct the three-dimensional space and the object according to a third instruction among the plurality of instructions.

[0016] In one embodiment, the space generation and editing system may include a coordinate generation unit and an image providing unit. The coordinate generation unit may generate coordinates corresponding to the three-dimensional space. The image providing unit may provide a plurality of base images corresponding to the three-dimensional space.

[0017] In order to solve these problems, in the method of operation of a space generation and editing device according to an embodiment of the present invention, a space providing unit may provide a three-dimensional space based on a first instruction among a plurality of instructions. An object unit may place an object in the three-dimensional space according to a second instruction among the plurality of instructions.

[0018] In the method of operation of a space generation and editing system according to an embodiment of the present invention, a space providing unit may provide a three-dimensional space based on a first instruction among a plurality of instructions. An object unit may place an object in the three-dimensional space according to a second instruction among the plurality of instructions. A correction unit may further correct the three-dimensional space and the object according to a third instruction among the plurality of instructions.

[0019] In addition to the technical problems of the present invention mentioned above, other features and advantages of the present invention are described below, or will be clearly understood by those skilled in the art from such description and explanation. Effects of the invention

[0020] According to the present invention as described above, the following effects are achieved.

[0021] The space generation and editing device according to the present invention can improve problems occurring in existing 3D models by breaking down space generation and editing into detailed tasks, modularizing algorithms specialized for each detailed task, and building a framework that integrates them.

[0022] In addition, other features and advantages of the present invention may be newly identified through the embodiments of the present invention. Brief explanation of the drawing

[0023] FIG. 1 is a drawing showing a space generation and editing device according to embodiments of the present invention. FIG. 2 is a drawing showing a coordinate generation unit included in the space generation and editing device of FIG. 1. FIG. 3 is a drawing showing an image providing unit included in the space creation and editing device of FIG. 1. FIG. 4 is a drawing for explaining a selection unit included in the space creation and editing device of FIG. 1. FIGS. 5 to 7 are drawings for explaining a selected image applied to the space generation and editing device of FIG. 1. FIG. 8 is a drawing for explaining embodiments of the space generation and editing device of FIG. 1. FIG. 9 is a drawing showing a space generation and editing system according to embodiments of the present invention. FIG. 10 is a flowchart illustrating the operation method of a space creation and editing device according to embodiments of the present invention. FIG. 11 is a flowchart illustrating the operation method of a space generation and editing system according to embodiments of the present invention. Specific details for implementing the invention

[0024] It should be noted that in assigning reference numbers to the components of each drawing in this specification, identical components are given the same number whenever possible, even if they are shown in different drawings.

[0025] Meanwhile, the meaning of the terms described in this specification should be understood as follows.

[0026] Unless the context clearly defines otherwise, the singular expression should be understood to include the plural expression, and the scope of rights should not be limited by these terms.

[0027] Terms such as "include" or "have" should be understood as not excluding in advance the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0028] Hereinafter, preferred embodiments of the present invention designed to solve the above problems will be described in detail with reference to the attached drawings.

[0029] FIG. 1 is a drawing showing a space generation and editing device according to embodiments of the present invention, and FIG. 2 is a drawing showing a coordinate generation unit included in the space generation and editing device of FIG. 1.

[0030] Referring to FIGS. 1 and 2, a space generation and editing device (10) according to an embodiment of the present invention may include a space providing unit (100) and an object unit (200). The space providing unit (100) may provide a three-dimensional space (SP) based on a first instruction (INS1) among a plurality of instructions (INS). For example, the plurality of instructions (INS) may be text entered by a user, and the first instruction (INS1) among the plurality of instructions (INS) may be "create a three-dimensional space (SP) with a width, length, and height of 4.5m, 3.3m, and 2m respectively, in which gray tiles are placed on the walls, and which has two windows and one door." In this case, the space providing unit (100) may create a three-dimensional space (SP) based on a pre-learned Large Language Model (LLM) model.

[0031] The object unit (200) can place an object in a three-dimensional space (SP) according to the second instruction (INS2) among the plurality of instructions (INS). For example, the second instruction (INS2) among the plurality of instructions (INS) may be "place furniture in the three-dimensional space (SP)," and in this case, the object unit (200) can place furniture in the three-dimensional space (SP) based on a pre-learned LLM model.

[0032] In one embodiment, the space creation and editing device (10) may further include a coordinate generation unit (110). The coordinate generation unit (110) can generate coordinates (JP) corresponding to a three-dimensional space (SP). For example, among a plurality of instructions (INS), the first instruction (INS1) may be "create a three-dimensional space (SP) with a width, length, and height of 4.5m, 3.3m, and 2m respectively, in which gray tiles are placed on the walls, and which has two windows and one door." In this case, the coordinate generation unit (110) can generate coordinates (JP) corresponding to the three-dimensional space (SP).

[0033] FIG. 3 is a drawing showing an image providing unit included in the space creation and editing device of FIG. 1, FIG. 4 is a drawing for explaining a selection unit included in the space creation and editing device of FIG. 1, and FIG. 5 to 7 are drawings for explaining a selection image applied to the space creation and editing device of FIG. 1.

[0034] Referring to FIGS. 3 to 7, in one embodiment, the space generation and editing device (10) may further include an image providing unit (220). The image providing unit (220) may provide a plurality of base images corresponding to a three-dimensional space (SP).

[0035] For example, the image providing unit (220) may include a plurality of modules, and the plurality of modules may include a first module (221), a second module (222), and a third module (223). The first module (221) may provide a first basic image (BI1) corresponding to a first instruction (INS1), and the first basic image (BI1) may be a border image corresponding to the border of a three-dimensional space (SP) formed according to the coordinates (JP) of a three-dimensional space (SP). Additionally, the second module (222) may provide a second basic image (BI2) corresponding to the first instruction (INS1), and the second basic image (BI2) may be a depth image representing the depth of a three-dimensional space (SP) formed according to the coordinates (JP) of a three-dimensional space (SP) in grayscale. Additionally, the third module (223) can provide a third basic image (BI3) corresponding to the first instruction (INS1), and the third basic image (BI3) is a semantic image that distinguishes floors, walls, and windows in three-dimensional space (SP) according to color.

[0036] In one embodiment, the image providing unit (220) may further include a selection unit (230). The selection unit (230) may provide a selected image by selecting from a plurality of basic images. For example, the plurality of basic images may include a first basic image (BI1) to a third basic image (BI3), and when the selection unit (230) selects all of the first basic image (BI1) to the third basic image (BI3), the first selected image (SI1) among the selected images may be a first basic image (BI1) which is a border image corresponding to the border of the three-dimensional space (SP) formed according to the coordinates (JP) of the three-dimensional space (SP), and the second selected image (SI2) among the selected images may be a second basic image (BI2) which is a depth image representing the depth of the three-dimensional space (SP) formed according to the coordinates (JP) of the three-dimensional space (SP) in grayscale. Additionally, among the selected images, the third selected image (SI3) may be the third basic image (BI3), which is a semantic image that distinguishes floors, walls, and windows in a three-dimensional space (SP) according to color. The space generation and editing device (10) according to the present invention can provide a panoramic image of a three-dimensional space (SP) indicated by G in FIG. 8 according to a pre-learned LLM model based on the first instruction (INS1), the coordinates (JP) of the three-dimensional space (SP), and the first selected image (SI1) to the third selected image (SI3).

[0037] In one embodiment, the space generation and editing device (10) may further include a plurality of modules. The plurality of modules may provide selection images corresponding to a plurality of instructions (INS).

[0038] FIG. 8 is a drawing for explaining embodiments of the space generation and editing device of FIG. 1.

[0039] Referring to FIGS. 1 through 8, in one embodiment, the spatial generation and editing device (10) can learn based on the result value obtained by compressing coordinates (JP) corresponding to a three-dimensional space (SP) in a predetermined manner. In another embodiment, the spatial generation and editing device (10) can learn based on the square of the absolute value of the difference value, where the difference between the loss value and the result value generated from the spatial generation and editing device (10) is the difference value. For example, the coordinates (JP) of the three-dimensional space (SP) can be compressed into a one-dimensional value according to a Long Short-Term Memory (LSTM) corresponding to a predetermined method, where the result value may be the compressed one-dimensional value and the loss value may be the difference value between the output image of the spatial generation and editing device (10) of the present invention and the correct image corresponding to the output image. The spatial generation and editing device (10) according to the present invention can be learned according to the formula shown at the bottom of FIG. 8.

[0040] The space generation and editing device (10) according to the present invention can improve problems occurring in existing 3D models by breaking down space generation and editing into detailed tasks, modularizing algorithms specialized for each detailed task, and building a framework that integrates them.

[0041] FIG. 9 is a drawing showing a space generation and editing system according to embodiments of the present invention.

[0042] Referring to FIGS. 1 through 9, to solve these problems, a space generation and editing system according to an embodiment of the present invention may include a space providing unit (100), an object unit (200), and a correction unit (300). The space providing unit (100) may provide a three-dimensional space (SP) based on a first instruction (INS1) among a plurality of instructions (INS). The object unit (200) may place an object in the three-dimensional space (SP) according to a second instruction (INS2) among a plurality of instructions (INS). The correction unit (300) may further correct the three-dimensional space (SP) and the object according to a third instruction (INS3) among a plurality of instructions (INS). In order to perform more detailed tasks on a three-dimensional space (SP) and an object, a third instruction (INS3) among a plurality of instructions (INS) may be used, and the third instruction (INS3) may be "paint the walls white and the floor wood color," and in this case, the space creation and editing system according to the present invention may edit the three-dimensional space (SP) in detail according to the third instruction (INS3).

[0043] In one embodiment, the space generation and editing system may include a coordinate generation unit (110) and an image providing unit (220). The coordinate generation unit (110) may generate coordinates (JP) corresponding to a three-dimensional space (SP). The image providing unit (220) may provide a plurality of base images corresponding to the three-dimensional space (SP).

[0044] FIG. 10 is a flowchart illustrating the operation method of a space creation and editing device according to embodiments of the present invention.

[0045] Referring to FIGS. 1 to 10, in order to solve this problem, in the operation method of a space generation and editing device according to an embodiment of the present invention, a space providing unit (100) may provide a three-dimensional space (SP) based on a first instruction (INS1) among a plurality of instructions (INS) (S100). An object unit (200) may place an object in the three-dimensional space (SP) according to a second instruction (INS2) among a plurality of instructions (INS) (S200).

[0046] FIG. 11 is a flowchart illustrating the operation method of a space generation and editing system according to embodiments of the present invention.

[0047] Referring to FIGS. 1 to 11, in the operation method of a space generation and editing system according to an embodiment of the present invention, a space providing unit (100) may provide a three-dimensional space (SP) based on a first instruction (INS1) among a plurality of instructions (INS) (S100). An object unit (200) may place an object in the three-dimensional space (SP) according to a second instruction (INS2) among a plurality of instructions (INS) (S200). A correction unit (300) may further correct the three-dimensional space (SP) and the object according to a third instruction (INS3) among a plurality of instructions (INS) (S300).

[0048] In the method of operation of the space generation and editing system according to the present invention, space generation and editing are broken down into detailed tasks, algorithms specialized for each detailed task are modularized, and a framework integrating these is constructed, thereby improving the problems that occur in existing 3D models.

[0049] In addition to the technical problems of the present invention mentioned above, other features and advantages of the present invention are described below, or will be clearly understood by those skilled in the art from such description and explanation. Explanation of the symbols

[0050] 10: Space creation and editing device 100: Space provider 200: Object part

Claims

Claim 1 A space providing unit that provides a three-dimensional space based on a first instruction among a plurality of instructions; and an object unit that places an object in the three-dimensional space according to a second instruction among the plurality of instructions, and learns according to the following [Equation 1] based on a result value obtained by compressing coordinates corresponding to the three-dimensional space in a predetermined manner and a loss value generated from a space generation and editing device, and [Equation 1] Here, is the result value, A space generation and editing device characterized by being a loss value. Claim 2 A space generation and editing device according to claim 1, characterized in that the space generation and editing device further includes a coordinate generation unit that generates coordinates corresponding to the three-dimensional space. Claim 3 In paragraph 2, the space generation and editing device is characterized by further including an image providing unit that provides a plurality of basic images corresponding to the three-dimensional space. Claim 4 A spatial generation and editing device according to paragraph 3, wherein the image providing unit further comprises a selection unit that selects from among the plurality of basic images and provides a selected image. Claim 5 A space generation and editing device according to claim 4, characterized in that the first selected image among the selected images is a border image corresponding to the border of the three-dimensional space formed according to the coordinates of the three-dimensional space. Claim 6 A space generation and editing device according to claim 5, characterized in that the second selected image among the selected images is a depth image that expresses the depth of the three-dimensional space formed according to the coordinates of the three-dimensional space in grayscale. Claim 7 A space generation and editing device according to claim 6, characterized in that the third selected image among the selected images is a semantic image that distinguishes floors, walls, and windows in the three-dimensional space according to color. Claim 8 In claim 7, the space generation and editing device is characterized by further including a plurality of modules that provide the selected images corresponding to the plurality of instructions. Claim 9 delete Claim 10 delete Claim 11 delete Claim 12 delete Claim 13 delete Claim 14 delete

Citation Information

Patent Citations

  • Rendering device

    KR102673660B1