scholarly journals Procedural interior generation for artificial intelligence training and computer graphics

2020 ◽  
Author(s):  
Egor Feklisov ◽  
Mihail Zinderenko ◽  
Vladimir Frolov
Keyword(s):  
Computer Graphics ◽  
3D Models ◽  
Automatic Generation ◽  
Machine Learning Algorithms ◽  
Single Person ◽  
Segmentation Algorithms ◽  
Realistic Scenes ◽  
Asset Creation ◽  
And Robotics ◽  
Semantic Database

Since the creation of computers, there has been a lingering problem of data storing and creation for various tasks. In terms of computer graphics and video games, there has been a constant need in assets. Although nowadays the issue of space is not one of the developers' prime concerns, the need in being able to automate asset creation is still relevant. The graphical fidelity, that the modern audiences and applications demand requires a lot of work on the artists' and designers' front, which costs a lot. The automatic generation of 3D scenes is of critical importance in the tasks of Artificial Intelligent (AI) robotics training, where the amount of generated data during training cannot even be viewed by a single person due to the large amount of data needed for machine learning algorithms. A completely separate, but nevertheless necessary task for an integrated solution, is furniture generation and placement, material and lighting randomisation. In this paper we propose interior generator for computer graphics and robotics learning applications. The suggested framework is able to generate and render interiors with furniture at photo-realistic quality. We combined the existing algorithms for generating plans and arranging interiors and then finally add material and lighting randomization. Our solution contains semantic database of 3D models and materials, which allows generator to get realistic scenes with randomization and per-pixel mask for training detection and segmentation algorithms.

Automatic generation of bas-relief on 3D models based on 2D images for rhinoceros

2018 ◽  
Vol 23 (6) ◽  
pp. 99-113
Author(s):  
Sha LIU ◽  
Feng YANG ◽  
Shunxi WANG ◽  
Yu CHEN
Keyword(s):  
3D Models ◽  
Automatic Generation ◽  
2D Images

scholarly journals ABGS: A System for the Automatic Generation of Building Information Models from Two-Dimensional CAD Drawings

2020 ◽  
Vol 12 (17) ◽  
pp. 6713
Author(s):  
Youngsoo Byun ◽  
Bong-Soo Sohn
Keyword(s):  
Computer Aided Design ◽  
Low Cost ◽  
3D Models ◽  
Automatic Generation ◽  
Information Modeling ◽  
Two Dimensional ◽  
Efficient Design ◽  
Information Models ◽  
Building Information ◽  
Floor Plans

Building Information Modeling (BIM) refers to 3D-based digital modeling of buildings and infrastructure for efficient design, construction, and management. Governments have recognized and encouraged BIM as a primary method for enabling advanced construction technologies. However, BIM is not universally employed in industries, and most designers still use Computer-Aided Design (CAD) drawings, which have been used for several decades. This is because the initial costs for setting up a BIM work environment and the maintenance costs involved in using BIM software are substantially high. With this motivation, we propose a novel software system that automatically generates BIM models from two-dimensional (2D) CAD drawings. This is highly significant because only 2D CAD drawings are available for most of the existing buildings. Notably, such buildings can benefit from the BIM technology using our low-cost conversion system. One of the common problems in existing methods is possible loss of information that may occur during the process of conversion from CAD to BIM because they mainly focus on creating 3D geometric models for BIM by using only floor plans. The proposed method has an advantage of generating BIM that contains property information in addition to the 3D models by analyzing floor plans and other member lists in the input design drawings together. Experimental results show that our method can quickly and accurately generate BIM models from 2D CAD drawings.

scholarly journals AUTOMATIC GENERATION OF ROAD INFRASTRUCTURE IN 3D FOR VEHICLE SIMULATORS

2017 ◽  
Vol 12 ◽  
pp. 79
Author(s):  
Adam Orlický
Keyword(s):  
3D Models ◽  
Automatic Generation ◽  
Quality Models ◽  
Road Infrastructure ◽  
Modern Methods ◽  
Interaction Interface ◽  
Vehicle Simulator

One of the modern methods of testing new systems and interfaces in vehicles is testing in a vehicle simulator. Providing quality models of virtual scenes is one of tasks for driver-car interaction interface simulation. Nowadays, there exist many programs for creating 3D models of road infrastructures, but most of these programs are very expensive or canÂtt export models for the following use. Therefore, a plug-in has been developed at the Faculty of Transportation Sciences in Prague. It can generate road infrastructure by Czech standard for designing roads (CSN 73 6101). The uniqueness of this plug-in is that it is the first tool for generating road infrastructure in NURBS representation. This type of representation brings more exact models and allows to optimize transfer for creating quality models for vehicle simulators. The scenes created by this plug-in were tested on vehicle simulators. The results have shown that with newly created scenes drivers had a much better feeling in comparison to previous scenes.

Quaternion epipolar decomposition for camera pose identification and animation

2013 ◽  
Vol 21 (1) ◽  
Author(s):  
W. Skarbek ◽  
M. Tomaszewski
Keyword(s):  
Computer Vision ◽  
Measurement Error ◽  
Computer Graphics ◽  
Quadratic Form ◽  
Essential Matrix ◽  
Virtual Camera ◽  
Camera Systems ◽  
Camera Pose ◽  
Pose Identification ◽  
And Robotics

AbstractIn the literature of computer vision, computer graphics and robotics, the use of quaternions is exclusively related to 3D rotation representation and interpolation. In this research we found how epipoles in multi-camera systems can be used to represent camera poses in the quaternion domain. The rotational quaternion is decomposed in two epipole rotational quaternions and one z axis rotational quaternion. Quadratic form of the essential matrix is also related to quaternion factors. Thus, five pose parameters are distributed into three independent rotational quaternions resulting in measurement error separation at camera pose identification and greater flexibility at virtual camera animation. The experimental results refer to the design of free viewpoint television.

Consistent Mesh Segmentation Based on Shape Diameter Function and EM

2014 ◽  
Vol 1049-1050 ◽  
pp. 1417-1420
Author(s):  
Hui Jia ◽  
Guo Hua Geng ◽  
Jian Gang Zhang
Keyword(s):  
Computer Graphics ◽  
3D Model ◽  
Shape Function ◽  
3D Models ◽  
Gaussian Mixture ◽  
Mesh Segmentation ◽  
Shape Similarity ◽  
Research Focus ◽  
Model Segmentation ◽  
New Research

3D model segmentation is a new research focus in the field of computer graphics. The segmentation algorithm of this paper is consistent segmentation which is about a group of 3D model with shape similarity. A volume-based shape-function called the shape diameter function (SDF) is used to on behalf of the characteristics of the model. Gaussian mixture model (GMM) is fitting k Gaussians to the SDF values, and EM algorithm is used to segment 3D models consistently. The experimental results show that this algorithm can effectively segment the 3D models consistently.

Automatic generation of computer graphics languages

1976 ◽  
Vol 11 (6) ◽  
pp. 113-122 ◽  
Author(s):  
Gregory J. Suski
Keyword(s):  
Computer Graphics ◽  
Automatic Generation

scholarly journals AUTOMATIC GENERATION OF BUILDING MAPPING USING DIGITAL, VERTICAL AND AERIAL HIGH RESOLUTION PHOTOGRAPHS AND LIDAR POINT CLOUDS

2016 ◽  
Vol XLI-B7 ◽  
pp. 171-176 ◽  
Author(s):  
W. Barragán ◽  
A. Campos ◽  
G. Sanchez
Keyword(s):  
High Resolution ◽  
Point Cloud ◽  
Spectral Band ◽  
Point Clouds ◽  
Automatic Generation ◽  
Aerial Photographs ◽  
Band Combination ◽  
Segmentation Algorithms ◽  
Digital Processes ◽  
Overall Effectiveness

The objective of this research is automatic generation of buildings in the interest areas. This research was developed by using high resolution vertical aerial photographs and the LIDAR point cloud through radiometric and geometric digital processes. The research methodology usesknown building heights and various segmentation algorithms and spectral band combination. The overall effectiveness of the algorithm is 97.2% with the test data.

scholarly journals Comparison of methods and tools for generating levels of details of 3D models for popular game engines

2020 ◽  
Vol 17 ◽  
pp. 412-416
Author(s):  
Michał Tomecki
Keyword(s):  
Computer Games ◽  
3D Models ◽  
Automatic Generation ◽  
Load Test ◽  
Comparison Of Methods ◽  
Simplified Models ◽  
Game Engines

Computer games have come a long way since their inception using increasingly advanced graphics. With so many models present on the screen at the same time, simplified models are needed that will be replaced when they are further away from the camera in the game. The aim of the study was to compare the tool for automatic generation of model detail levels in Unity and Unreal Engine. The article presents the results of the equipment load test and the time needed for generation, as well as a survey checking the opinion of people.

Sign in / Sign up

Export Citation Format

Share Document