Democratize Visualization in the Studio

LUKAS FAETH: Welcome everybody to this presentation about democratizing visualization. My name is Lukas Faeth. I'm the Senior Product Manager for Automotive Visualization at Autodesk. And today, I'm going to talk about how you can leverage visualization and democratize it across the company so everybody can leverage it and you can make the best out of it.

So, before we start and dive into the specifics and the actionable items of the democratization process, let's take a look at the general value of visualization first. And let's quickly start with a quote from one of our customers. And just to put that in context, we're talking about collaboration here, right? And why visualization is important to collaboration and something that we'll discuss after I read the quote out.

"You'd think this is so much more complicated than a video call. But collaboration has been very reliable, that's only possible if we trust what we are seeing and the tools we are using. It amazes me that something so complicated is working so naturally and easily."

So, Mr. Guillaume, in this case, is talking about collaboration, virtual reality collaboration specifically in VRED, where, as you can see in the image, two individuals are reviewing a digital prototype of a car together. In this case, it's a Kia, beautiful car. And why is visualization important to collaboration? Because it is the base for the collaboration, the communication, the decision making, right?

So, this digital data set that they see, this digital car, is the base for them to take their decisions on, and also the base for them to collaborate on. Having collaboration functionality that allows people to work together across the globe, something that's just adding on top of the visualization element and providing even more value to the company.

So, let's dive into the value of visualization a little bit more. And I'm going to talk, or I'm going to show that in context of the automotive industry, because that's where I'm working in. But you can basically apply this to any other industry as well, any product design, or heavy machinery, or whatever you want. So, let's take a look at insights first.

You can use visualization, of course, and digital prototypes to generate insights early in the process. And in automotive, for example, you could simulate the in-car experience, meaning how will it feel for an end customer that buys the car when he's sitting in it, even before you build the first physical prototype.

Or you could explore material selections and combinations to make sure that you have the best interior set of materials that is fitting to the overall design of the car, and of course, lighting simulation. One example here would be ambient lighting in the interior, which plays a huge role in today's automotive.

Then the next part would be cost. You can reduce costs with visualization. So, of course, visualization capabilities, especially those of Autodesk VRED, are made to replace or reduce the usage of physical prototypes. In the automotive industry, those are very expensive. So, basically, a huge group of people will build a clay-based or a model out of clay and other physical elements.

And this is a very long and cost intense process, and the digital prototypes allow you to have quicker iterations. So, make more iterations and iterate more, decide more, change more, explore more, and thus, you will finally save money because you can do more in the same time. You have a quick turnaround. This is very different to the physical prototypes where it takes a long time to adapt.

In digital world, this is something you could even do ad hoc during the meeting. Just change something on the shape and get some feedback from your stakeholders as you go. And then, of course, we talked about that in the beginning, less and no travel cost. We have this quote from Gregory Guillaume from Kia and he's basically using VRED for collaboration.

And this saves him time to fly through the world to discuss the design changes on site with the individual team. So, yeah, there's plenty of cost that can be saved. Of course, one thing that is like the biggest aspect is finding errors early in the process. So, the sooner you will find the error in the process, the less costly it is to get rid of it.

If you find something in production phase that needs to be changed in the design on the engineering, this could have a huge financial impact on your project. That's why it's important to evaluate everything early on digital prototypes, see how it's going to behave, and then be sure in the later steps of the process to not run into those issues.

Time, of course, we talked about that a little bit already. Time when the traveling, for example, but also the iterative advantage that you have, you'll accelerate your decision making, which is a huge point, right? I think we touched that already. So, it's easier to take a decision because you can do more smaller iterations if you have good quality, reliable visualization.

You have efficient reviews, already talked about that as well, and you have a faster go to market, because the overall time to market is shrinking due to the speed advantage you have with digital prototypes and digital evaluation methods. And then the last point is quality. Something that is very, very important in the automotive industry is quality.

And visualization is helping you to achieve this level of quality. So, it may be the accuracy of surfaces and materials, which is extremely important to the automotive industry. As you can see that on the image in the background, which is a Ford. And you can see that there's a focus on high quality leather materials, with the stitching.

Everything is perfectly designed. And the same way it goes for the surfaces, interior and exterior surfaces, to build the ideal shape of the car and the ideal design. And then lighting and gaps. This is a very specific thing for automotive, like the bringing together all the pieces of the exterior is a very, very sophisticated process, in terms of including the aspects of the production of the end car and the tolerances that you need to take into consideration.

So, there it's very important to have reliable, accurate visualization as well. And something that's very obvious is better perceived quality. So, if you can check the quality of the final vehicle soon in the process, you can change it or improve it more and have a better final result.

But all this value needs to be unlocked and there are challenges to unlock this for a wider field of audience. VRED is used by currently, or other visualization tools as well are used by specialists, but eventually, you want everybody in the company to have access to this value. So, let's take a look at some of the challenges companies are facing when trying to democratize the value of visualization.

There's basically four aspects that are challenging. One is accessibility. So, allowing everybody to access the visualization system or the visualization data. Then there is the barrier of hardware, because this is often the limiting factor because for rendering, you need specified or very capable hardware, which not everybody in the company has.

Then there is expertise, we already talked about that. So, you need to have the skill to run a visualization product, to be able to benefit of it. And scalability, which sometimes, for very specific, very expert use cases, you need more than one machine to compute the image in real-time.

So, let's take a look at how VRED Core addresses those challenges and the VRED product line, and how we can solve it. So, we just talked about that already. Let's take a look at how it looks around accessibility in the past, right? So, we have all those different personas, and as I said, we are not just focusing on the visualization expert. We are focusing on everybody else.

We want to provide everybody access to this visualization data. And currently, it looks like that the whole company needs to go through the expert to access the data set. And in the future, or you could do this as of today with our product, you can shortcut this and provide everybody access to the data set themselves. Because as I said, there are the barriers, like the hardware or the expertise.

And we are removing the both of them, and I'll show you in the next slides a little bit how we can do that. And we're removing all those barriers to allow everybody to directly access the pool of visualization data. And this could be through an Ultrabook, mobile devices, XR devices, or even if they have it, VRED capable hardware or very performant workstations, for example.

But it's not necessary, right? So, with VRED Core and with the whole setup that we're talking about today, you could also access the visualization data, the same visualization data through your mobile device without a problem from wherever you are in the world.

So, as you can see, this is an example of having real-time access to this data set through a tablet. Through streaming, we are able to reach everybody. We can provide the visualization result, as you can see, in very high quality to everybody. It's very easy UI as well. It's easy to use device that not only visualization specialists have, but everybody in the company would have access to.

And this is a bit more sophisticated, but going down to the same problem. This is streamed XR [? even. ?] So we talked about streaming is going to enable the accessibility. This is a more sophisticated approach where XR streaming, so we're streaming AR mobile application.

And the cool thing here in this example is we're even doing this in collaboration. So, this mobile device is in a collaboration session with two HMDs. So, as I said in the last slide, it's independent of which device you have, you'll be able to access, collaborate, communicate, and take your decisions on the same data set, if you use VRED visualization capabilities.

The next limiting factor is hardware. And in the past, because as we already said, you need capable hardware, so either this is very powerful workstations or even more cluster servers that are on site. There's limitations to that as well, of course, right? So, if you have them on premise, again, there's the maintenance cost.

Only a limited amount of people can access it because it's expensive and it's not easy to operate. So, you will have to have somebody who's capable to operate the whole server set up and everything, again, limiting the usage for a specific group of people and not providing it to everybody. With the cloud, we have the capability to allow everybody to use resources on demand, right?

Whenever they like to use it to an extent that they need to use it at this point in time. That's why we have designed VRED Core specifically in a way that it works on premise server, but also on the cloud, so we can achieve this accessibility, or remove the hardware barrier, and allow our customers more flexible ways to leverage as much computation power as they need from wherever they are. Not location bound to where the hardware is located.

And one example here can be seen in this video. This is cloud streaming. And you can see, again, we have a high quality rendering image. This is a collaboration session. So, both our colleagues are looking at the same data set. And you can see that in this case on the left, it's manipulated through mobile device even. So, there's more than one device collaborating here.

We have two desktops and two mobile devices being powered by the cloud. So, you can either use the cloud power to render one big output, or as we just saw, have several streams to different people at the same time to collaborate or to provide a end user experience to hundreds of people at the same time.

Then something we touched on as well is expertise. And I think this is a very serious barrier to visualization or has been in the past, because let's be honest, visualization software is not the easiest to be used. It's not super tough and you can get into it, but not everybody is also willing and has the time to learn it, right? So, let's think about a senior decision maker, like a management person in an automotive company.

It wouldn't be smart even for him to learn a visualization tool just to get access to it, right? Because there's other things that he has on the plate, I guess more important stuff that he needs to take care of than learning, for example, VRED Pro interface to operate it. So, in the past, again, we had only experts, as we saw in the first slide, with the accessibility problem.

And we already saw that through streaming we can reach every persona as our content reaches the systems they are used to and comfortable working with. With their level of complexity and functionality tailored to each personas' needs. So, we can work around the problem of having to be an expert by automating and providing custom interfaces that are just easy to use and intuitive to use.

Let's take a look at the automation first. And this is a bit of a visual example, but I hope we still get the point. So, what we see here is automated script for the data preparation of an alias file into VRED. So, what's happening is nobody's doing anything, and you can see the progress on the lower right of the screen.

And you can see that the script runs through a set of commands and completely prepares the data set from a Cat dataset to a high quality visualization dataset, and then even render some images out that you could use. So, with no expertise at all, you could just drop a dataset into an automated system and get a fully prepared visualization dataset back that you could use afterwards in one of the applications we are talking about on the next slide.

Or you could also just get your assets back, like, for example, automatically created images or videos. Let's look at an easy to use interface, because this is, I think, the most challenging thing, right? You don't want to operate a software. With VRED, we are delivering this interface that you can see, which is automatically populated with all the variants you have of your dataset and it's easy to use as an app on your mobile device.

You can just access it from any device, because its browser-based, and then you can run through it, play through your configurations, play animations, collaborate, set annotations, and a lot of more things that you can do with it. This is our out of the box streaming app that we are shipping, but of course, we are shipping the source code of this as well. So, you can use it as a starting base for your own custom interface as well.

Or you could connect VRED to an existing interface that you have in your company already, that people are used to using already, because we're basically just accessing the stream and sending some commands to VRED to switch to variants. And this is very, very easy HTML code that you could write yourself as well, if you want to customize it for a specific group of people, or as I said, integrate it into a system that you might have in your company already and want to power this system with high quality visualization with our VRED product line.

So, let's look at the last challenge, which is scalability. And it's not a big problem as long as you just have one person that wants to do something, an easy visualization task on one machine locally, that's not a problem at all. Because you can just use your notebook or your workstation to achieve or to do the task. But it's getting more complicated the moment you want to really visualize something that is very demanding in terms of performance.

Because then it could be the case that you want to connect several machines to compute one result, which is called clustering or scalable rendering. And for this, there are challenges, of course, because we saw that in the beginning already. You either have to have the hardware on site, which is connected to maintenance cost, specialization, and expertise, or you access the cloud.

And I think the cloud here is a great new addition to the workflow, because not only is it simple to access from anywhere, right? So, you don't have the location dependency, but also, you can scale it up and down on demand. So, if you need, I don't know, 100 machines that compute one task at a time at one day, you can just spin them up, have them compute it, close them down.

If you want to do, I don't know, like, need 500 machines the next day to compute one result for 1/2 an hour or whatever, however long the design review goes, you can do that as well, or with the flexibility to adjust that to what's your needs. Another aspect of scalability might be not having several machines computing one result, but having several machines streaming several results, different results to different people in the world.

And that's possible with the scalability of the cloud as well, and with VRED Core, right? So, you could spin up an experience for hundreds of customers, different ones, tailor made to each persona or each person, and then send it to those customers. And they can look at it simultaneously around the world wherever they are.

So, the cloud provides us huge amount of flexibility and VRED has the possibility and capability to efficiently scale on this environment, on this hardware environment. So, it's perfectly suited together, VRED Core and cloud, goes together very nicely because it is designed to be used on the cloud.

A quick comparison video or quick video that shows some of the aspects that we're just talking about. So, we have different notebooks here. And you can see just a relative scale of how good they, or how fast they compute the same task. You can see we're past, we're at a local server with 8RTX GPUs. Then this is a cloud instance, this is 40 GPUs.

Again, 40 GPUs, and then we're going to 160 GPUs. So, you can see that in the cloud, you can scale that up tremendously and adapt it to whatever task you have. And this is not possible, it's just simply not possible, if you have fixed amount of computation power on site, right? Because at a certain point, you'll just not have enough machines to scale it up more and you need to buy them, which is a lengthy and costly process.

So, yeah, using the cloud resources according to the demands and needs of your visualization is something that we use to solve the challenge of scalability for our customers. How could the result like that look, right? For everybody who didn't see a cluster result yet, this is a great example. We have a Porsche Taycan here.

This is fully, physical, accurate rendering, which is called full global illumination. It's powered by a set of GPUs in real-time, as you can see. So, you can navigate around, evaluate your car, evaluate aspects of it. And this would be one example of a result that you want to look at in real-time, where you need more than one machine to compute it.

This is the gaps example that we were talking about in the beginning where automotive is very focused on. You really want to see how the shadows and the light is interacting with the different elements, gaps of the car, with the shapes as you can see here. With the reflections, or creating reflections. And this is a very demanding visualization task because the data set is very heavy, and also the level of accuracy is extremely high, and you want to still achieve this in real-time.

And this is where scalability is necessary, where you need more than one machine to compute it. And as we said in the last slide, with the cloud usage, this is possible and very convenient and saves money compared to the traditional ways of having the hardware on site.

So, let's summarize quickly. We took a look at the challenges, and how we solve them to democratize virtualization to allow everybody to access the data sets from anywhere. Let's recap what and how we solved it. We have the problem of access, or we provide access to everybody, allow everybody to leverage high complex visualization data through easy to use interfaces.

So, making it easy, removing the expert needs. That can be flexibly adjusted towards the end user in the application area. Then something that's very important, reuse data sets for many tasks throughout the whole lifecycle of your product development. So, this is something specific to VRED as well. We are able to leverage the same data set used to prepare it once.

Might be automated, as we saw, and then you can use it for virtual reality, you could use it for the cluster review that we just saw. You could render images, movies, whatever you like, right? So, it's one dataset. You you're saving the time to prepare the dataset specific to your use cases, because VRED is so versatile in the way it is capable to provide the output types.

Then removing the bottleneck, like getting everybody access by removing the experts needs. So, it's not about getting rid of the visualization experts. It's more about creating more time for them to do the actual work they should do instead of trying to help others to get access to visualization data. And then, of course, leverage VRED on the cloud.

We have the automation piece that we saw, which allows non-experts to prepare the data completely without anything they need to know or any time they need to spend. It's completely automated on a server or on a machine. And the data is converted, prepared, and you can even generate your renderings or any movie assets you want to create automatically.

Then we have the customization aspect, which is very important. You need to be able to customize or we are able to customize the interface to tailor it to whoever is in need of using the system, making it easy for them. And just provide the information that they need to fulfill that task and not overwhelm them with a UI that might be complicated to operate.

Or integrate the stream into a custom system that you have already. And then we have the scalability aspect as well, that we just talked about, using the cloud to scale up and down the hardware that you need to fulfill your visualization tasks, depending on your needs. And all of that is enabling a huge amount of benefit, as we talked about, or value, as we talked about in the beginning.

We're saving time, money, and we have more insights, and also, this is going back to the very initial quote that we had, this is enabling collaboration, communication, and decision making because the visualization data set, or the visualization that you're using, will be the base. Only a vehicle or something that helps to facilitate a conversation or that is the base for a decision. So, this is something that is necessary for the basic work that every company is doing on a daily basis. And it's going to become more and more important in the future as well.

One thing here as well, which is coming back to the initial statement, is something that's very important. You need to be able to rely on the visualization. So, it needs to be accurate enough because we are actually replacing physical prototypes, which are physically accurate, because they are in the real world, right?

So, you need to be able to trust what you see. I think VRED has proven to be able to do that. Some of our automotive customers are not even building any physical prototypes anymore, as I said initially as well. So, we already proved that our visualization is accurate enough that you could produce cars with it and find real life arrows with digital evaluation methods. And with that being said, let's take a look at a quick video of very simply explained what VRED Core is and how it can be used.

[MUSIC PLAYING]

PRESENTER: Visuals are powerful. They bring your ideas to life, so you can design smarter, collaborate better, and produce faster. But in today's automotive industry, you need your visualization tools to do more. You need them to work for every person, on every device, at every step in your development process.

Now, they can with Autodesk VRED Core. VRED Core takes the industry-leading visualization power and quality of VRED and delivers it in a new server-based solution, that can run in the cloud or on-premises. Through the VRED Core API, you have access to the full functionality of VRED, which means you can attach VRED's high-quality rendering stream to any front end to customize your entire visualization workflow, based on your needs.

You can automate every step in your process to let VRED Core take care of repetitive tasks so you can focus on more important things, like creating products customers love. VRED Core kicks your rendering performance into high gear, giving you instantaneous feedback and access to your photorealistic images, animations, and real-time presentations.

Because anyone can access VRED's high-quality streaming from anywhere, on any device, you can collaborate, review, and showcase it all in real-time and high quality. So, your visuals are always powerful enough to bring your ideas to life, precisely. Discover the power of VRED Core. Contact us today.

LUKAS FAETH: And with this video, that's my last entry for the topic of democratized visualization. I hope you liked the presentation. I hope it was helpful to you. If you have more questions, just let me know. And yeah, thank you very much for your attendance.