ODA Summit 2021 - Part 3: Beyond Data Interoperability

Summary

本视频讨论了3D扫描技术在建筑、工程和施工（AEC）行业中的应用，以及如何通过自动化提高数据处理效率。专家们讨论了从扫描到BIM的过程、激光扫描和无人机技术在项目中的应用，以及面临的挑战，如数据转换和模型创建。同时，介绍了ODA的可视化引擎，展示了其在工程设计中的作用，并通过案例研究和未来计划，展望了技术发展的趋势。

Takeaways

• 🌐 越来越多的公司使用Kodi进行可视化和网络应用开发，并寻求解决复杂行业问题的创新技术。
• 🏗️ AEC行业中扫描束（Scan to Beam）的应用正在增加，特别是在改造和翻新项目中。
• 🏢 公司在项目中创建模型的需求不断增长，尤其是在扫描到BIM（Building Information Modeling）的应用上。
• 📈 通过激光扫描和无人机技术，可以捕获大型场地的点云数据，提高项目执行的粒度视图。
• 🔄 将点云数据转换为模型是一个挑战，尤其是在项目紧迫和资源有限的情况下。
• 🤖 自动化在3D扫描中扮演着重要角色，未来可能会有算法帮助筛选和识别扫描数据中的相关元素。
• 🛠️ 行业正在寻求一个“魔法按钮”，以高效地将点云数据转换为可用的模型。
• 📱 激光扫描技术的发展使得数据消费变得更加便捷，如使用平板电脑和智能手机进行室内设计。
• 🎥 Oda的可视化引擎允许在任何平台上为工程应用添加专业的2D或3D图形。
• 🔄 通过集成Oda Visualize，MSI Design在其旗舰计算机设计应用中实现了性能提升和开发成本降低。
• 🚀 未来计划包括在Oda Visualize中添加反射平面支持、快速对象转换和亚像素形态学抗锯齿等高级功能。

Q & A

为什么越来越多的公司选择使用Kodi进行可视化和网络应用开发？

Kodi因其强大的可视化能力和灵活性，被越来越多的公司用于解决复杂的行业问题，包括数据互操作性之外的挑战。

在AEC行业中，扫描到BIM的作用是什么？

扫描到BIM在AEC行业中的作用主要是为了获取现有建筑的精确数据，以便在进行改造或翻新时，能够更好地理解现有结构和流程，并有效地整合新的设计和工艺。

Walsh公司在处理大型升级项目时，面临的最大挑战是什么？

Walsh公司在处理大型升级项目时，面临的最大挑战是如何在现有设施中整合新的管道、过滤系统和工艺流程，并确保这些新元素能够与当前的流程和程序顺利对接。

在设计和施工过程中，激光扫描的主要目的是什么？

在设计和施工过程中，激光扫描的主要目的是捕获现有建筑的实际建成状态，为后续的设计和施工提供准确的数据支持。

为什么自动化在3D扫描中扮演着重要角色？

自动化在3D扫描中扮演着重要角色，因为它可以帮助从扫描数据中过滤出不必要的信息，提高数据处理的效率，并且有助于将扫描数据快速转换为可用的模型。

ODA的可视化引擎AudioVisualize提供了哪些功能？

ODA的可视化引擎AudioVisualize提供了专业级的2D和3D图形添加功能，能够在任何工程应用程序和平台上使用，包括对点云、网格或BIM模型的高质量可视化。

为什么ODA的可视化技术对于AEC行业来说是一个重要的进步？

ODA的可视化技术对于AEC行业来说是一个重要的进步，因为它提供了一种优化的方式来处理和呈现复杂的模型数据，同时支持跨平台使用，有助于提高设计和施工的效率。

在集成ODA Visualize时，IMsi Design遇到了哪些挑战？

在集成ODA Visualize时，IMsi Design面临的挑战包括更新渲染引擎以提高性能、降低开发成本、改善客户体验，并且需要在不同的硬件平台上实现渲染的可扩展性。

ODA Visualize的哪些特性使得它在设计审查中特别有用？

ODA Visualize的高质量可视化、快速选择和高亮显示功能、以及对多种设计格式的支持，使得它在设计审查中特别有用，能够帮助用户有效地验证项目多个方面并进行沟通协作。

Open Cloud平台如何处理复杂的文件和数据？

Open Cloud平台通过提供参考API、可视化.js库、用户管理API、角色基础API、文件操作API和自定义作业API等工具，能够有效地处理复杂的文件和数据，支持从文件中提取几何或属性数据，并允许用户执行自定义作业来提取额外信息。

Outlines

00:00

🌐 Kodi的多元化应用与扫描到BIM的讨论

本段介绍了越来越多的公司正在使用Kodi进行可视化、网页应用开发，以及寻求解决复杂行业问题的创新解决方案。特别强调了扫描到BIM（建筑信息模型）技术在AEC（建筑、工程、施工）行业中的应用，并提到了一个专门讨论这一技术在行业内应用的小组讨论。提到该讨论的完整版本可在YouTube频道上找到，强调了其提供的深入见解的价值。

05:02

📊 3D扫描与模型转换的行业挑战

这一部分深入探讨了3D扫描技术在不同行业项目中的实际应用，尤其是在改造和维修项目中。讨论了从传统的激光扫描到使用无人机进行大范围扫描的技术进步，以及这些技术如何帮助从实际场地获取精确的数据。专家们分享了他们在使用点云数据和3D模型转换过程中面临的挑战，包括数据的大量性和处理这些数据的高复杂性，以及如何通过自动化和算法改进来解决这些问题。

10:04

🔍 在行业中推广3D扫描的挑战与机遇

讨论了在AEC行业中推广3D扫描技术面临的挑战，特别是高成本和高技术门槛限制了更广泛的应用。专家们讨论了如何通过技术创新和解决方案开发来降低这些障碍，例如通过开发新工具和平台来简化数据捕捉和处理流程。同时，指出了在个人和小规模项目中应用3D扫描的潜力，强调了行业内合作和知识共享的重要性。

15:06

🛠 ODA Visualize技术的实际应用案例

介绍了ODA Visualize技术在工程设计和CAD软件开发中的应用，通过IMS Design的案例展示了如何将ODA Visualize集成到他们的TurboCAD产品中。讨论了这一技术如何帮助改善渲染性能，提升用户体验，并支持多种硬件平台。此外，分享了开发过程中的一些关键考虑点，包括与ODA团队的合作，利用源代码访问和不同版本库的同时使用。

20:14

🌟 ODA Visualize的动画和测量功能

这一部分重点介绍了ODA Visualize的动画API和测量功能，展示了如何在视觉化场景中对模型进行动画处理和精确测量。详细解释了这些功能如何帮助设计应用程序用户以更直观、更灵活的方式呈现和分析模型数据。同时，讨论了高级自定义和性能优化技术，如使用字体缓存和减少GPU调用来提升渲染性能。

25:16

🚀 ODA Visualize的未来发展方向

讨论了ODA Visualize技术未来的发展计划，包括增加反射平面支持、改进对象变换处理和引入子像素抗锯齿技术。此外，展望了新的文件格式VSFX的开发，该格式旨在减小文件大小并支持高效的数据流式传输。提到了基于强烈市场需求，开发针对Open IFC Viewer的插件系统的计划，以便更好地支持AEC行业用户。

30:16

📈 ODA在云基础设施和WebAssembly应用中的使用

介绍了ODA技术在云基础设施和WebAssembly应用开发中的实践案例。首先，通过Lunar Engineering的EDMS系统展示了如何使用ODA技术管理、审核和协作处理工程文件。随后，详细介绍了Open Cloud解决方案，强调了其在处理复杂CAD和BIM文件、优化数据存储和执行定制作业方面的能力。最后，通过Google V8团队的介绍，探讨了在Chrome浏览器中调试和性能分析WebAssembly应用的方法。

Keywords

💡可视化

可视化是指将数据或模型通过图形界面展示出来，使得用户能够直观理解复杂信息。在视频中，可视化技术被用于工程设计和建筑信息模型（BIM）中，以帮助专业人员更好地理解和分析项目。例如，通过使用ODA的可视化引擎，可以将复杂的BIM模型以2D或3D图形的形式展现给客户和团队成员，从而快速理解空间布局或进行简单测量。

💡激光扫描

激光扫描是一种测量技术，通过发射激光束并测量其反射回来的时间来确定物体的形状和位置。在视频中，激光扫描被广泛应用于建筑、工程和施工（AEC）行业，用于创建现有条件的详细模型，这些模型可以用于设计、施工和维护等多个阶段。

💡点云数据

点云数据是一种由大量三维点组成的数据集，每个点都有自己的位置坐标。这些点通常由3D扫描设备捕获，用于表示物理对象或环境的三维模型。在视频中，点云数据被用于记录建筑信息，尤其是在激光扫描和无人机测绘中。

💡建筑信息模型（BIM）

建筑信息模型（BIM）是一种数字化工具，用于表示和管理建筑和基础设施项目的物理和功能特性。BIM模型包含了建筑物的详细信息，如材料、构件尺寸和成本等，这些信息可以在设计、施工和运营过程中使用。

💡设计审查

设计审查是指在项目开发过程中对设计方案进行的一系列评估和验证活动，目的是确保设计满足所有技术和功能要求。在视频中，设计审查涉及到使用专业软件工具来检查和分析BIM模型，以验证项目的多个方面。

💡WebAssembly

WebAssembly（简称Wasm）是一种为网络浏览器设计的低级字节码格式，它允许在网页上以接近原生性能运行编译后的代码。这种技术使得C、C++等语言编写的程序能够高效地在网页中运行，而不需要解释执行。

💡模型转换

模型转换是指将一种格式的三维模型数据转换成另一种格式的过程。在AEC行业中，这通常涉及到将扫描得到的点云数据或BIM模型转换成其他软件或平台能够识别和处理的格式。

💡自动化

自动化是指使用技术来执行原本需要人工操作的任务。在视频中，自动化特别指的是在3D扫描和模型创建过程中减少人工干预，通过算法和软件自动处理数据和执行任务。

💡开放BIM标准

开放BIM标准是指一系列公开的、用于建筑信息模型数据交换和协作的标准规范。这些标准允许不同软件平台之间共享和利用BIM数据，促进了行业内部的协作和数据互通。

💡云计算

云计算是一种通过互联网提供计算资源（如服务器、存储、数据库、网络、软件、分析等）的服务模式。它允许用户按需访问这些资源，而无需在本地部署和维护硬件和软件。

💡性能优化

性能优化是指改进软件或系统以提高其运行效率和响应速度的过程。在视频中，性能优化涉及到对WebAssembly应用程序进行调试和分析，以发现并解决性能瓶颈。

Highlights

越来越多的公司今天正在使用Kodi进行可视化，他们正在使用Kodi进行Web应用开发，并且他们正在寻找解决其他复杂行业问题的解决方案。

本次会议的这一部分致力于超越数据互操作性的早期技术。

扫描到BIM在建筑信息模型中的应用正在增加，每个项目现在都有某种形式的扫描。

在建筑生命周期中，人们希望了解扫描的前端设计和施工过程的动机。

激光扫描工作过去主要集中在捕获建筑物的实际情况，例如大型机场项目。

由于无法信任建筑物的历史建筑文档，人们希望通过扫描来确认建筑物的实际状况。

在小型空间的3D扫描中，无人机的使用正在增加，但目前还没有一个工具能够帮助小型项目进行3D扫描。

自动化在3D扫描中的角色正在发展，尤其是在处理扫描数据和转换模型方面。

目前，行业面临的挑战是如何有效地将点云数据转换为模型，这是许多人的障碍。

ODA正在尝试提供填补这些空白的核心技术，而不是等待每个单独的供应商和行业部门自己解决。

可视化在工程设计中起着关键作用，无论是点云、网格还是BIM模型。

ODA的可视化引擎允许在任何平台上为任何工程应用添加专业的2D或3D图形。

通过集成ODA可视化，MSI设计提高了其TurboCAD系列产品的渲染性能和质量。

ODA可视化引擎的集成使得在不同的图形系统上运行时都能提供最佳的用户体验。

ODA可视化引擎支持多种渲染设备，这意味着客户端应用程序可以运行瓦片打印或PDF导出，同时与屏幕渲染设备一起使用图形系统缓存以获得最佳性能。

ODA可视化引擎现在支持在GDI设备中进行透明度处理，这可以用于打印和渲染。

ODA可视化引擎的未来计划包括添加对反射平面的支持，以及为动画和快速对象变换实现子像素形态学抗锯齿。

Transcripts

[Music]

more and more companies today

are using kodi for visualization they

are using kodi for web application

development and they are looking to idea

for solutions to other complex industry

problems this last part of our summit is

dedicated to early technologies that go

beyond data interoperability we'll start

this section from a panel discussion

about the role of scan to beam

in the aec industry in the scope of our

summit you can see a part of this

discussion with bright insights but i'd

like to assure you that it's worth

watching a full version available on our

youtube channel

[Music]

you know looking at your practices in

general

how often is your company seeing

projects where there is a need to create

models from these scans and the point

cloud data and there is definitely an

increase in

demand for the use of this it's to the

point that basically every project now

in the last couple years has had some

form of scanning whether it's

to um

scan to bim

the last project was a substation in a

basement uh

with complex floor and ceiling and wall

condition where we see the greatest

value is in those retrofits those

renovations um

walsh has three different verticals it's

our building group civil group and our

water group and

water time and time again when you're

massive upgrades how are you

retrofitting this existing facility with

new pipings new filtration systems new

processes so

really having a granular view on how

this is going to integrate into the

current process and procedures is very

difficult and you know can go out there

and take a tape measure and measure it

but you know that really doesn't provide

um

great hindsight uh when you're throwing

these models together um as well as how

you're gonna get this information or

this equipment into these spaces

yeah for us it's it's very similar

we've

deployed point cloud technology for the

last decade

video via either laser scanning directly

or increasingly with with drones to

capture the larger type sites and

i tend to look at it from the building

life cycle perspective so i always like

to start with why

why do people want to do the scan

in the front end of the design

construction process you have the as

built condition

so a lot of the

front end laser scanning work that we've

done in the past has has been around

capturing the ass built condition for

example we did mem international airport

which had an extremely large footprint

you know as an airport of course and it

was built

where nobody really trusts the as-built

documents because they go i need 10 20

30 generations you know they're like

you know

you got you want the 1975 version or you

want the 1982 or do you want the 1994

and they recommend you look at all of

them because they all have different

pieces of information on them but none

of them have the truth

so

because you can't trust the as built

documents that we're inheriting from

the history of that building

and people want the confidence to truly

know what's out there we've been

scanning to document the as built

condition something for us that that is

important is the scale that we have some

projects that you know we have towers

that we want to retrofit we're

definitely going to do that but there is

also a niche in in a a practice area

that we call workplace

and and there is so many smaller spaces

and uh and we see the drones as gene

mentioned

we see the point clouds we see all type

of 3ds kind but there is no um a tool

that help

that small ditch when you have to 3d

scan something you don't want to send

your team there

but you need to capture the space so

there is a lot of opportunities in there

in the in the field we have seen

increasingly um i would say probably

a hundred percent of the projects that

we have that are existing conditions

they're doing 3ds kindly maybe you can

uh help also describe what is your

process i mean are you

are you going out and doing the laser

scanning yourself with your own units

are you hiring somebody third party to

do it are you then getting the point

cloud data and and translating that

yourself or is that also a third party

service what you know what kind of

things are you relying on depending on

the complexity of the project

the availability of our scanners you

know that's that's always a

difficult thing to approach we have

multiple scanners within our

organization but

you know we have a retrofit project

comes up you know that's that's just

like a full-time fte that's sitting

there and that's being leveraged on that

project

one of the banes of our existence is

when we hear we need to perform laser

scanning on that project and then the

first question that comes to mind are we

creating the model oh no the architect's

creating the model no problem we'll go

scan that that's we're completely fine

with scanning that project because

that's the easy part scanning and

registering those points no problem uh

the complexity of really developing that

model that's time consuming and uh

unfortunately these days in the industry

it's you're closing up one project

starting up another project

simultaneously and bidding a project in

between the two of those so

it's very difficult to kind of stop do

all those iterations and go through it

from the architect's standpoint um we've

seen a couple different variations we've

never done our own scanning we're

usually contracting out it's either with

the general contractor

or it is coming from an outsource

firm of some sort

and then from that point once we've done

the scan

we have both modeled in-house and we've

also outsourced the model or the general

contractor has provided the model from

their scan a big point it's it's

something we've toyed with as far as the

the model conversion

it's something that we used to

handle all of that in-house

the software years ago has been

developing where there's

automatic feature extraction

but what we found is

particularly with piping type systems

the software will create those elements

but then you spend just as much time qc

and making sure everything fits properly

that

we weren't really seeing the the

efficiencies to make that process go and

that's that's kind of why we're here

talking about this the

industry's been chasing this magic

button where point cloud

allows us to capture more data than

we've ever had on the construction

site and then with drones as well

but the ability to efficiently convert

that to a model is

is something that really

holds a lot of people back

so what will be the role of automation

in 3d scanning because we all face the

same challenge when we go and we scan

we get a lot of information that

we probably don't need so for example um

i get a 3d scan i come with a bunch of

piping that has they're not really

relevant for the

let's say the scope of work that we're

doing i wonder if in the future some

kind of algorithm is going to start

understanding

what is piping what is a wall you know

it should be something that

start filtering those things we we have

seen some automation with you understand

what the what type of geometry is there

and say okay this could be a wall and it

and it throws a wall there

um sounds like cost and the time

associated with a making the scans and b

making those conversions is a very big

deal

uh you know obviously not every

not every type of stakeholder you know

if you're an architect and you're

focusing on design

that may not be feasible within your

firm or you know sounds like on the

contracting side you figure out ways to

make it work because

that is a part of your process that is a

fundamental part of what you do and

being able to do the scanning but maybe

conversion isn't always necessary or if

you could get conversion on top of that

as a

as an easy gain then it would be much

easier to justify

you know so so oda is is

you know in this this realm of trying to

provide sort of these core technologies

that fill these gaps um but what do you

think about that approach i mean does it

does that seem to make sense to you

rather than sort of waiting for

you know each of the individual vendors

and sectors to just sort of figure it

out

well yeah i think there's a need so it's

interesting because um one part of this

is how do you consume the scan data

right

one of the issues that we've always had

of course you know when you talk about

lasers kind of an airport you can

imagine how difficult this is to consume

in every sense hardware

just transferring the information making

accessible making it available bringing

it into other systems so we obviously

need more optimized ways of being able

to make that data available in a wide

range of platforms in a wide range of

environments everything

from uh you know tablet computer all the

way up to your smartphone you're seeing

movement on the laser scanning side of

things you know it's everyone's got the

newest latest and greatest cell phone or

ipad and they now all of a sudden have

laser scans on them nobody knew what the

heck they were going to do with it you

know it was for the ar so you can have a

little dancing rhino on your desk but

then people in our industry started

being like wait a second i can now

work with an interior designer i could

scan my apartment send it off to

interior designer and get a beautiful

decoration of my apartment well

if we could do that in our personal

level why why aren't we trying to

holistically approach this in our

industry to make it easier if i can go

out there with an ipad all my sub or all

my formative ipads out there let's start

scanning this let's start tying this

into different third-party solutions uh

procure not procurement uh production

tracking things like that so

i think the sky's the limit when it

comes to developing solutions for this

workflow

because it's an underutilized and the

burden of entry is so high that not

enough people are playing here yet so

it's kind of a

majorly under underutilized uh principle

in our industry and i i think it's it's

worthwhile and it's worth the investment

to start understanding what it could do

for you and your projects to your point

i i think that's definitely somewhere

where the industry as a whole would

definitely benefit there are a lot of

different groups out there that have

different tools and

i'm sure like most of you we we have to

be pretty agnostic we have to use a lot

of different tools there's a lot of

different conversions

to get what you need

each has their strengths have their

weaknesses

and we're constantly going through that

process on

on the scanning and software side just

to see

okay where's the industry at who's who's

bringing new

new technologies new potential to it but

the ability to

bring all those collective ideas

together in a simple

efficient solution i i think is

an amazing challenge but definitely

something that that would benefit the

whole industry i will follow up with

what you're saying james

and i see the same thing so one would be

the part of visualization that gives

access to clients and everybody in the

team to really quickly understand the

space or even take a simple measurement

you know

now to be able to access that data it is

pretty heavy and it is hard to to

anybody in the team to

just open and take some quick

measurements

and the other thing would be translation

so how you take all that point point

cloud and you make it accessible to

whatever platform you're going to use it

you know we cannot restrict our

our users or whoever is is working with

this data to only work in one specific

brand

software so we cannot say well you if

you get this one you use this tool to

translate the data and now you need to

use auto a revit you know or you need to

use um archicad so it needs to be

something called like ifcs you know that

is is it can work across any platform

and you can open this this file in

sketchup if you want a rhino and you can

get the same precision

[Music]

visualization plays a key role in

engineering design

either your data is point cloud air mesh

or a beam model

now it is time to take a closer look at

oda's professional visualization engine

[Music]

audiovisualize allows you to add a

professional 2d or 3d

graphics to any engineering application

on any platform

let's let's take a closer look at what

this technology can do for you and i

will start from a success story provided

by one of our oldest founding member msi

design

last year they have integrated our

visualize to their flagship computer

design application

and in the context of these samples

i would like to draw attention to the

creative approach that allows to emulate

feature which is only in our future

plans here i mean reflection plane

next demos illustrates how visual styles

can create an impressive visual

representation even for a simple model

changing of a few visual style options

like h model or age crease angle can

significantly alter the rendering and

add different effects to the final model

representation now msi design

will tell you more about the experience

with oda visualize

[Music]

hello i'm tim olsen vice president

development at imsi design i've had the

opportunity to be involved with cad

development since the early 80s close to

40 years now

and throughout that time visualizing cat

data such as text dimensions line styles

wireframe and 3d facet data have been a

constant challenge

it's been a challenge because of the

changes we've seen in the underlying

software and hardware interfaces

and driven by our customers ever

demanding need to support larger and

larger files

today i'm going to share with you our

experience of integrating oda visualize

with our turbocad family of products

imsi design has been developing desktop

and mobilecad products since 1983.

we have distributed over 16 million

products during that time targeting

consumer as well as professional cad and

aec users on windows mac android and ios

platforms

our products include turbo cat design

cad floor plan and our home design

architectural series our turbocad for

windows product is our first application

to integrate visualize

this product has extensive 2d drafting

annotation surface and solid modeling

geometric constraints

parametric design and an extensive suite

of interoperability solutions turbocat

is used by our customers on a wide

variety of platforms

ranging from integrated gpus

to high-end dedicated boards such as the

nvidia geforce

we need our rendering implementation to

scale with the platform providing all of

our customers with the best possible

experience

our goal with visualize was to update

our rendering engine to improve

performance reduce our development costs

by consolidating rendering engines and

to improve our customers experience

our first product to integrate visualize

with is turbocad

turbocad previously supported four

graphic engines including gdi opengl

lightworks and red sdk

our render manager wraps these engines

under one layer our approach was to add

visualize over a 12-month development

period as a fifth rendering engine

through two development phases

phase one we were to duplicate the

current functionality into the turbocad

2021 release

and phase two exposed new features in

the turbocad 2022 release turbocad 2021

with visualize was released last spring

with positive feedback from our

community of users

we observed two areas of notable

improvements in performance and quality

rewarding performance we saw up to 10x

an order of magnitude performance

updates for wireframe and 3d facet type

data on high-end graphics ports

and 2-3

x performance on lower end boards we

also saw some unexpected quality

improvements regarding anti-aliasing and

ambient inclusion

regarding anti-aliasing we are able to

expose methods and parameters to

fine-tune those jaggies out of both

wireframe and 3d facet type data

and with respect to screen

space ambient occlusion we are able to

expose some simple lighting setups for

users to quickly get more realistic

renderings

next up are four things we believe

helped us integrate visualize into our

turbocad product

first the oda visualized team was very

supportive in answering our questions

providing suggestions and listening to

our suggestions

i encourage anyone getting involved with

oda components to review the example

code use the timeline help and get

involved with the user form second

as a founding member we have access to

source code

source code provided us with the ability

to step through the debugger inside the

example code as well as through our code

and provided us with additional insight

into the workings of visualize

third the ability to use two different

versions of odi libraries together in

one application this enabled us to

update visualize in a monthly manner

without affecting the base oda library

which is modified source requiring more

time to update for us

and lastly our lead engineer vlad veslov

was able to leverage his prior

experience with rendering engines

towards visualize

we found implementing visualize

consistent with other rendering engine

concepts with turbocad 2021 we

duplicated our existing rendering engine

to include oda visualize with turbocad

2022 we extend our rendering engine with

new features based on visual styles

new visual styles such as x-ray

conceptual and gray along with

customizable parameters will allow more

ways for our customers to view complex

data

in closing our original goal was to be

as fast or better than our competitors

with our rendering engine while

improving the overall customer

experience

with the help of oda visualize we

believe we have accomplished that goal

going forward it is our intention to

leverage visualize into more of our

products including our macbase products

i look forward to our development team

focusing more on creating unique product

features

where oda components such as visualize

are a key component of that strategy

thank you

now we are moving forward with our

presentation there are many interesting

things ahead of us

[Music]

the animation api allows objects within

a visualized scene to be animated

entity animation allows translation

rotation and scaling to be applied to

the objects in the model

objects in oda visualize are stored in a

hierarchical structure and the animation

api takes into account this hierarchy

for example we can specify one animation

for an entity and another one for its

sub-entity and the resulting sub-entity

animation will be a combination of both

its own and its parent animations

view animation introduces a new camera

object that contains properties similar

to the view

using the camera object you can create

things like walkthrough animations for

complex building models

now let's talk about measurements

the ability to select and measure parts

of a model is a popular feature for many

types of design applications

for convenient management visualize

supports options like nearest point

center point starter endpoints and so on

also we have added measurement 2 to our

sample applications which can be used as

a base for the measurement support to

your client applications

[Music]

let's talk about highlighting

customization and sun graph highlighting

customization provides a way to control

edge and face style color masks for

highlighted objects transparency

displaying of the lines or contours and

drawing order of the highlighted

geometry can be also controlled

in addition we can use different

highlighting styles in a rendering stand

different parts of the model can be

highlighted using different styles to

meet the needs of even the most complex

design applications similarly this

technique can be used to highlight

different kinds of objects by most fit

highlighting style the next block of our

presentation will be devoted to the

visualized sun graph

it uses background processing of

geometry data to minimize number of gpu

calls and optimize rendering performance

we actively developing and enhancing

this feature visualize uses a triotype

fonts cache and block references cache

to optimize vectorization performance

and reduce memory usage for complex

models

in the past this reduced memory usage

has come at the cost of slower rendering

performance this year we have improved

our sun graph to work seamlessly with

our font and block cushion giving

reduced memory usage without any serious

loss in the rendering performance

another critical some graph bottleneck

is overall memory usage we have reduced

some graph memory usage by eliminating

intermediate geometry data which was

previously kept to facilitate selection

and highlighting processes

this table shows the benefits of this

work for different models

[Music]

during class here we have significantly

improved the screen space ambient

occlusion as sao it is a rendering

technique which allows to add an effect

of global illumination to the scene

first of all now we are using enhanced

blur to avoid shadow noise

in addition the ssao can now be adjusted

for the different zoom levels

either manually or in automatic mode

when realizing kfc in many cases the

full ifc model is divided into a set of

separate files electrical mechanical

plumbing

open fc viewer is now able to load all

these files into a single scene

all these modules can be manipulated in

object explorer pane

we can control visibility performance

action and many other operations like

for example collision detection between

objects from different models

for example we can test that analytical

equipment

doesn't collide with plumbing

also we have added the new functionality

to the collision detection pallet

to allow the transparency of the

collided objects to be customized this

year we have continued to work on apple

metal renderer fios and mac platforms

since metal device invokes the similar

code base with opengls 2 device

we have work on a more deep unification

of the data providing for these two

devices

also we have add new functionality to

the metal device

like support of the substantial

attachment support of the index geometry

cutting planes

another important feature is the frame

buffer support

one of the benefits from the frame

buffer support is the correct order

independent transparency even in

multi-target mode

our future planes with metal device

includes the transferring of the

openjlis to geometric shaders to the

metal as far as implementation of the

advanced features like screen space

event occlusion

fast proximity anteriorizing and more

[Music]

audio visualize now supports multiple

rendering devices for a single database

this means that client application for

example are able to run tile printing or

pdf exporting with optimal performance

using graphic system cache together with

on-screen rendering device previously

this was not possible another small

enhancement is ability to draw

background texture on faces in hidden

line mode

this enhancement can be used to create

an effect when background is visible

through geometry in contrast with

classic hidden line where a constant

background color is used to fill

geometry the next item involves raster

images previously

we could select a raster image only by

clicking on its border highlighting is

also applied only to raster image

borders

and now audio visualize provides an api

to enable selection and highlighting of

the full raster image contents a popular

question on oda technical support forum

is how do i

render a raster image with transparent

background

and now client applications is able to

request rendering of

32 bits per pixel raster images with

transparent geometry for any oda

vectorization

[Music]

model now we will mention a few final

visualization features

and the first one is a transparency

support in our gdi device which can be

used as for printing and for rendering

the next two options are related to the

import part of the oda visualize the

first one allows to generate a so-called

3d view during import of revit files if

such view is absent

the second one is the ability to import

data from frozen layers in dwg files

the last topic that i want to mention is

the viewcoup implementation

vcoop is a clickable object that allows

you to change the current view

it is a quite popular tool in modern

computer edit design applications

[Music]

now it is time to discuss our future

plans

first of all we will plan to add the

support of the reflection planes

the next topic is a fast object

transform

this is a very important feature for an

nd animation

for the cases when many objects are

involved in animation

like for example is an explode tool

also we will plan to add the

implementation of the subpixel

morphological anti-aliasing as maa

we should solve the problem with fast

approximate analyzing when one pixel

lines appear to blur

we are now working and we will continue

to work on the new file format for the

visualize vsfx

this new format will reduce the size of

the file up to 2.5 times

and will support efficient streaming

during last year

we have seen a strong interest to our

application open ac viewer

and as a part of this interest we have

received a main request for the creating

of the plugin system to this application

and we plan to work in this direction

before we move on to next topics let's

see an interesting member case

demonstrating a complex use of different

audi tools in cloud

including visualize and publish

[Music]

hi there i'm sean the product lead at

lunar and onset design coming to you

today from melbourne australia and

thanks for tuning in in this talk i'll

run through a basic introduction of the

lunar engineering document management

system or edms for short and dive into

some of the key oda technologies that

make our product possible

onset design is a small company based in

australia and we specialize in

everything related to engineering

document management

we've been an oda partner for a number

of years and we assist companies

implementing edms to manage their cad

files

one of the key challenges faced by our

customers has been receiving and

reviewing large sets of drawings from

their contractors and we've historically

used the oda libraries as a part of a

portal product to implement validations

on receipt of drawings

this allows us to automatically check

title block metadata quality among other

things and speeds up the qa process

allowing as built to be received and

processed more efficiently

luna is a cloud-based edms hosted on

microsoft azure that grew out of this

original portal product

it's designed to allow smaller companies

to get their edms up and running fast

without the need for a complex design

and config process

the key features of lunar document

upload revision review and collaboration

we use a key set of oda technologies to

make these features possible

[Music]

to show how the oda tech fits into our

application architecture i'll introduce

you to something that we call the lunar

document processing pipeline or the

journey that documents go through after

being uploaded to luna firstly as

documents are uploaded to luna they're

automatically published using the

publish sdk to thumbnail and pdf

thumbnails are used for previewing

documents in our list view whilst pdfs

are used for document review and markup

following this

we push the documents through the

drawings.net sdk

to extract the various document metadata

properties

and store the title block and xref data

next to the document in azure blob

storage as json

this can be used for validation metadata

review and more later on in the drawings

workflow

when models are uploaded we also push

them through the open cloud file

converter automatically generating vsf

files which can be used to view the icf

model ifc models in the browser as our

customers are starting to work with more

model formats we've seen an increased

demand for the users to be able to view

and navigate models in luna without the

need for an installed component after

looking around at the various model

viewing options available we finally

landed on open cloud

using opencloud we've been able to

enable customers to view ifc models as

well as native dwg and dgn files

natively in the browser without the need

for first translating them to pdf the

next step will be to use the bim rev rv

revit sdk to enable this functionality

for revit models as well

[Music]

whilst there are alternative methods to

achieve some of these features for

example 3d model viewing they typically

come with trade-offs like requiring a

model to be uploaded and stored in a

separate cloud repository which can be

problematic when it comes to data

sovereignty

or per model or document rendering fees

which can drive up the total cost of

ownership for customers over time by

using the oda libraries we can maintain

control of where our data is hosted and

keep our prices comparatively low by

avoiding rendering fees

so for us it's been a huge step forward

and what's next well going forward we

aim to further use the ifc and rapid

sdks to unlock some of the smart data

captured in models making it available

to all you lunar users

that's all for me today thanks for

tuning in enjoy the rest of your day

[Music]

new features added to our digital kits

and especially to ada visualize

give us the ability to provide the

solution for another important task a

thorough design review of aec models to

validate multiple project aspects

this solution is a design review engine

that evolved from a set of specific ode

technologies visualize publish open

cloud common data access and various bim

formats support

the major new capabilities we had this

year to support a complete design review

vlog 4 are

support for federated b models in

visualize and open cloud and two-way

conversion between visualize toolkit

and navisworks database

for now ode design review engine is a

solution with advanced set of

functionality import natively supported

multiple design formats ifc revit files

navisworks files dwg files with

architecture civil and map data

investigate an aggregated model with

high quality visualization

fast selection and highlight

get access to all model attributes

validate the model with advanced 3d

clash detection

use 3d markup tools to communicate

project needs

collaborate basing on open beam

standards such as bcf

and prospectively open cde

publish entire model of

or part of it

to pdf

exchange the model data with navisworks

files

this design review engine features are

available to extend your existing

application

or to create full featured design review

tools

[Music]

when developing the open cloud solution

we try to find a balance between

stability system security and speed of

adding new functionality

to do this we deploy our solution to

demo infrastructure to test new

functionality

and improve our stability for the past

year

several thousands of users were

registered in our

application

which give us the opportunity to test

our backend unreal data set

in the same time we were able to test

our new functionality in real conditions

open cloud support a wide range of file

formats

for card it's a dwg djn dxf and some

others

for the beam it's ifc revit at nineties

work

also we support some

common 3d file formats

some files use additional file resources

this can be a font file textures

materials and even external reference

to work with such complex files we

provide a reference api

you can upload all your files to the

backend and link them together through

the reference api

this mechanism is flexible enough so you

can link your file multiple times

it can be used for example to create

your own font library for rendering

files in the browser we provide a

visualize.js library

this is a cross-compiled version of

visualize project into a web assembly

format

it provides a rich api to work with the

geometry data to manipulate with them

or to create a new one

beside this we provide an api for

different

tools like a measurement tool a slice

panel tool

and some others we are constantly

improving functionality of our

visualize.js library from the latest

improvements we speed up the performance

of loading geometry data

and we change the way of loading this

data

which now is loaded the closest to the

camera

[Music]

open cloud server provides an api to

working with the users

administrator can create new user and

specify his specific settings

and configuration

for example you can create a user

specify his name his position

number of the project that the user can

create and the amount of storage that

the user can use

for the convenience these users can be

combined into groups

quite often on the real project

it is necessary to specify the same

project access for the set of the users

to do this we provide a role based api

you can create a custom role

change

per user permissions and assign this

role to your users or your groups open

cloud provide the class detection

functionality

you can use this functionality to detect

intersections between objects in the

scene and it provides an api to

configure this

intersection detector

with the help of the visualize.js

library you can highlight the

intersected objects and show them

another feature of the server is the

file assemblies

this type of endpoints allow you to

merge

several files into a single common

structure

and then you can work with this

structure as if it were a single file

[Music]

open cloud provide a simple set of file

operations you can extract geometry data

from the file or properties data

but sometimes it is required to execute

custom jobs to extract additional

information

to do this open cloud server provide an

api to run custom jobs

for example

you can create such custom job to

extract xdata from the dwg file and send

it to a separate storage with a

notification by default open cloud

server use file system to store user

files

this is a reliable and secure solution

but it requires additional

administration and configuration

if this option doesn't fit you

we provide

access to the additional third-party

storage systems

you can connect server to the amazon s3

storage or azure blob storage the server

is powerful enough to withstand the load

of

a large number of users

during stress testing

uh

we use the server instance with one

single core and two gigabyte of memory

the server in such configuration was

able to handle 18 000 requests per

minute with one and a half requests per

second for one active user the server

can withstand the load of

200 users for the convenience of working

with the open cloud system we provide an

open cloud cli tool

you can use this to easily install and

configure solution on your environment

for example you need just only two

commands to run our solution in docker

system

[Music]

welcome to our session on web assembly

debugging my name is emmanuel ziegler

and i work in the va development team of

google v8 is our javascript and

webassembly engine that is used in the

chrome browser microsoft edge and

node.js if you are new to webassembly

you need a little refresher on it don't

worry we will quickly present the

knowledge required to understand this

talk

the focus will be on how to debug and

profile a webassembly application right

in your chrome browser

this can be useful if you have a bug or

performance issue that is hard to reduce

in native code of course you may also

just use chrome out of convenience for

not having to reproduce the error in

another environment

in last year's talk i gave an

introduction to webassembly or wasm for

short and how you can use it to bring

your cc blusbus code to the web i

recommend this talk if you want to know

how to get started with western

development

for now all you need to know is that

wasm is a low level language that allows

you to safely and performantly ship

binary code on the web

instead of a machine executable western

provides an intermediate representation

that will be compiled just in time in

your browser

wasm is available in any major browser

but i will focus today on google chrome

for convenience

to create a business module from cc

blastbus code we will use the

inscription toolchain wasm also supports

other languages such as rust or

assemblyscript with more to come

we will start with a simple example that

simulates the collets conjecture which

claims that if you triple the value and

add 1 for odd numbers and half the value

for even numbers repeatedly you will

eventually arrive to a loop consisting

of the sequence to one over and over

again provided that you started with a

positive integer

if you're confused about what this does

don't worry the details are not

important for our investigations we will

be running this with different random

numbers over and over for 500 million

times the problem with this

implementation is that it does not work

because it gets stuck in an endless loop

we now want to debug the algorithm to

find the mistake and we can add

debugging symbols and switch off

optimizations using the minus g switch

just like with gcc or client for

convenience i am creating an html file

together with the watson module and its

javascript code

using a special extension which is still

in beta testing chrome can reach dwarf

debugging information embedded in the

module and we can navigate the c codes

directly you can find the extension

under the link in the slide

for the extension to work we also need

to enable an experimental feature in the

chrome developer tool settings

you can do that by opening the developer

tools clicking the little gear icon in

the upper right corner and then enabling

the web assembly debugging devtools 12

support in the experiment section

i have now opened the debugger in

devtools and set a breakpoint in the

main function after reloading the

program pauses at the main function as

expected

[Music]

you can see this from the pause in

debugger message on the top of the page

as well as in dev tools paused on

breakpoint and the breakpoint that i'm

currently passed on is highlighted in

the list and in the source code view

the scope on the right hand side

contains the local variables the local

variable is still undefined at this

point but once i start stepping through

the function it gets initialized with 0

and then with its actual value from the

time function

[Music]

once i reach the scope of the for loop

the run counter also becomes visible

run is initialized with zero and i not

only can see this in the scope but also

when i hover over the variable

[Music]

[Music]

i will now put a breakpoint here so we

can pause every time a new call to run

example is made

i can either deactivate the other

breakpoint or remove it completely as we

don't need it anymore

[Music]

we are now going to step into the

function but of course the first

function to be called is rand and we

therefore end up in there

as you can see this function does not

have debug information and we therefore

are presented with a raw vessel code

which you can also see from the stack

trace showing this little warning icon

for the c main function i generated the

debugging information while for this

automatically generated main wrapper

there is also no information

below that we find the javascript trains

you can see that the scope now shows

information on the local variables in

javascript

if i go back into rand we can see that i

can step through this low-level web

assembly just like the rest of the code

i can look at the local variables and as

i keep stepping through it the

information slowly gets populated

including the low level stack

if you are interested in this low level

behavior here you can just step through

your functions and see how it works

[Music]

we are currently not interested in this

so i'm going to step out of this

function and step into run example

we can watch the value of i change as

i'm stepping through it

[Music]

[Music]

do

[Music]

we now want to figure out why our

algorithm does not complete and i

therefore continue until we reach the

point where it fails

can see that the breakpoint is not hit

again even though we should have

completed one example in fractions of a

second

so this means that we are still in the

first call of this function

let's pause and see where we are

we are still in one example as expected

and if i go one level up we can confirm

that this is indeed still run 0 and we

really didn't hit the breakpoint

let's go back into the top scope and see

what's happening

there let's take a good look at the

number i and remember it for later

as i'm stepping through the function the

number keeps changing as expected

[Music]

[Music]

[Music]

but then at this point in time i end up

with the same number that we saw

originally and this means that this loop

is going to continue forever and i will

always toggle in between these two

numbers and never reach the exit

condition as i never becomes one

[Music]

if you remember the collets conjecture

you'll notice that something is not

right here

we have an odd number here and we should

be multiplying it by three and add one

but we end up in the branch that should

divide by 2 but as you can see from the

shift operator it's actually pointing in

the wrong direction here

[Music]

[Music]

so we got two errors to fix

switch the if and then bodies and fix

the shift operator

this should then be an accurate

implementation of the coreless

conjecture and we should exit each loop

properly

let us quickly fix the issue with the

wrong condition by exchanging the if and

else bodies

if we compare the performance to that of

the native version compiled with clang

we see that we are substantially slower

this only sometimes happens and if we

simply reload the page the performance

is on par with the native version

sometimes westminster is still a little

slower but it should come close to what

you see in native performance

so why was our first run slower

in last year's talk i explained that v8

uses a two-tier compilation strategy

this allows faster startup times while

ensuring a high peak performance

in the beginning it can happen though

that the top tier has not yet completed

its compilation when the program is

executed and because there is no on

stack replacement it can happen that it

stays in the baseline here for the whole

execution

this is usually not an issue because if

the program keeps returning to a

function higher up the stack and then

calls the inner function again the new

version will be called

in benchmarks this might not happen

though because of their simplicity and

the forced inlining of the run example

function certainly does not help here

there can even be implicit inlining the

compiler notices that the function is

not used outside the code

what works well in native code can

therefore be a problem with wasm

so you might consider forbidding

inlining of benchmarking functions that

are called

repeatedly but why did it work the

second time

that's because chrome caches compilation

results of the top tier so they don't

need to be recompiled again and again

on repeated runs you will therefore

likely start with the top tier right

away

now prevent inlining of the function and

we can see that the performance is now

much more predictable

or so it seems at least

if i load the program with devtools open

debugging will actually force vh to use

the baseline tier leading to bad

performance

this is especially confusing if you

decide to output your benchmark results

to the console and therefore keep the

console open

it's better to run the benchmark first

and then open the console to check the

output

v8 will then run at full performance and

you still have access to all the

messages

[Music]

if you want to run a proper function

profile of our code we should recompile

it with a profiling command line flag so

function names which are normally

stripped from the reson binary are

included

this also enables some other useful

changes to the module but if you do not

want that and you prefer just having the

function names put in the module you may

just use the profiling funcs flag

we can record function level profiles in

webassembly using the performance tab in

devtools

you can choose to press the recording

button to start the recording

immediately or reload and then start

recording

i will reload so my program gets

re-executed

once we start recording the module will

be tiered up so we can record the

regular performance and not the one of

the baseline

here a few seconds of data suffice for

our needs and when the profile is done

we can see that the majority of time is

spent in script execution

[Music]

[Music]

looking at the timeline we see that we

are mostly skipping in between two

functions the original main from the c

file and the run example function

the gaps where run example stops

executing is the time spent in main or

the rand function

we can also take a look at the bottom up

call graph which features the functions

in which most time is spent on the top

as to be expected the majority of the

time is spent in one example where the

inner loop is followed by original main

with the outer loop and occasional calls

to rent

the other parts only contribute to a

negligible amount we can also check from

which functions this one was called so

we understand the call pattern

we can also look at the top down graph

where we can easily see the path the

program takes and on which branch it

spends the most time

all of this works exactly like it does

with javascript for more complex

programs this provides useful insights

for optimization targets that really

help speed up your application

thank you for watching this little speed

run through webassembly debugging and

profiling

if you work with webassembly please try

these features and let us know what you

are missing most

have a nice day and happy coding