Realize your designs anywhere, anytime – with new V-Ray Cloud rendering

ETIENNE DEGROOT: All right, thanks everybody. Can you guys hear me OK in the back? Awesome. All right. I think this is the last session of the day, hopefully. Right? We're going to go ahead and get going here, since we're at the bottom of the hour. Just a quick introduction.

So my name is Etienne DeGroot. I'm at Intel, and I'm part of our Intel Data center group. So I manage our visual cloud, responsible for what we're doing from a cloud graphics and cloud gaming perspective. And with me today, I've got Phil Miller, he's the VP of Products from Chaos Group. And we're going to give you a presentation talking about how to maximize your workflows from a rendering perspective.

So with that, we'll kick it off. And some quick introductions, here. So from the flexibility designers are looking for flexibility across their workloads. And even for the most demanding designs, they want to take full advantage of their hardware investment. And there's a lot of optimizations to make it happen that are done in both hardware and software.

So not only can you tackle today's challenges, but you can be ready for what tomorrow brings forward from a platform perspective. And what we're doing from a VR and AI perspective, you saw a lot of that stuff in the keynote. Bringing your maker visions to reality, we can do that on common infrastructure. We'll talk a little bit more about that as we go forward.

This is a build, sorry about that. From a workstation perspective, some of the tools of the trade. There's three basic buckets. There's the mobile and entry, that is basically our Icore product, and our Xeon product. This is more entry level, including Mobility. Mainstream that is across both Icore products as well as Xeon. That is more of the volume opportunity. And as we get into the expert, that's more of the dual socket workstations. There's a variety of offerings there from a variety of programs that you guys have seen. And as we go to cloud, obviously, there's other choices than having powerful workstations. As things are rendered in the Cloud, you can do this across multiple devices as you guys are probably aware.

So from an empowerment perspective, there's a variety of things that are happening. So to maximize the utilization of what you guys are doing, running data center in AI workloads on common flexible architecture has a lot of values. You can run multiple workloads, and take advantage of kind of that initial investment. And then you can add acceleration to that as you go forward.

From a memory perspective, breaking those barriers. There are limitations to what you can put on a physical card. So a lot of the things that are being done from a rendering perspective on CPUs is really embracing a full system capability, full system memory. And there are multiple systems say, in the Cloud environment, you get to take advantage of all that capability. So that's kind of the direction going forward on cloud usage models.

How many of you guys have heard of Embree or what we're now calling Intel rendering framework? This is basically software libraries that are open source. And Autodesk amongst others, embrace this, and they put this into the rendering engines. And what this allows you to do is take advantage of what I just mentioned, full system memory across, whether it's a workstation or multiple servers. And you can actually take advantage of this, and run models that are beyond GP memory in a physical card.

And this allows you to have that flexibility and scalability where, if you've got to do a job and you've got a client coming in, you could render across all of the cores in the network versus what's just in the pedestal system as an example. And this allows you to do very interactive ray tracing kinds of capabilities. So you can actually realize your designs as you go forward in that environment. There's a variety of applications that are out there. Digital content creation, Hollywood, a lot of movies are made this way that are made on standard infrastructure.

The bottom left image is a Moana one. There is another one movie that just came out, The Grinch, that's also been taken advantage of this as well. And there's a host of others. So this is across multiple segments, not just Hollywood, but whether it's high performance computing, energy like oil and gas, manufacturing, et cetera. And Phil will talk a little bit more about that as we go forward.

As far as an ecosystem perspective, Autodesk and many others are embracing this. This is something that's not widely known. But a lot of people are taking advantage of these software, this kernel that's put out there, and a lot of people are taking that and putting that into the rendering engine. And as Autodesk users, you guys are enjoying that with a lot of the hero apps that they have here.

As far as the reason for this, people are taking advantage of Moore's law. That doubling performance every two years continues. And this is kind of a map of Xeon over the last, say, 10 years here. And as we go forward, we're actually in the Sky Lake mode now, second from the top. And this is per socket, so if you have a dual socket workstation or a server, you're basically going to double this. So the amount of core's that are out there to render in your infrastructure, or your IT environment are pretty pervasive going forward. And this is not slowing down anytime soon. So with that, I'm going to turn it over to Phil, and talk a little bit more about Chaos Group.

PHIL MILLER: Oh, that was quick.

ETIENNE DEGROOT: It was.

PHIL MILLER: Hi, everybody. So I thought it was a great opportunity to talk with a Intel about both the cloud, as well with Intel because it's all about scalability. So first of all, how many people here are familiar with Chaos Group, or I should say Brayrender? OK. Well it sounds like everybody is, a couple people aren't.

So a little background, Chaos Group, we've been around for about 20 years. We probably have the most widely adopted professional renderer in the industry. We're used by about 92 out of the top 100 architectural firms of the top ones. Basically they depend upon Bray for all of their visualizations. And then we're, of course, used in visual effects and the like, as well.

So you may have heard us or seen us the most, actually, in all the movies that we've been used in. We're very popular in motion pictures, especially, for when the scenes are actually very architectural. So a lot of the 3D backgrounds and everything that you see in the scene movies, are all done in V-Ray. For example the last Black Panther, almost all the almost all the Marvel movies, all coming to you from V-Ray. And for this, we were awarded an Academy Award, a couple of years ago.

So while you might be most familiar with us, in shots like this major visual effects movies. We're also used heavily in styling, in design, industrial design. And when those designs come to the screen in the form of commercials, or even when those come to form in print, like this. Or in product design, product presentation, and marketing. And then, of course, coming back to architecture, where we're used quite a bit.

A lot of the things that we put in V-Ray we actually do with architects in mind. And actually, that migrates to Hollywood, rather than the other way around. But we do it both ways. Also V-Ray's very important, not just from a presentation standpoint, but also for understanding what your design really is. Especially how it interacts with light. Or imagining what can be. These are some competition images done by some of leading architects like Gensler on the left.

Or imagining what will never be. Complete flights of fancy. And a lot of times, these worlds combine. Like for example, architecture and motion pictures all coming together, like in a movie like this from San Andreas. So we give you access to all this great imagery through basically plug-ins that are added to your most popular products. The tools that you use every day.

So at Autodesk, your primary ones are right here. We have these four tools here from Autodesk, and used quite a bit. Also this week, and what we're showing here on the show floor is V-Ray for Unreal. So V-Ray for Unreal actually connects to all the other products that you see up there. So you can actually work within Max, Maya, Revit, and bring your things over into Unreal very easily. You can see that in the epic booth all week.

So our mission isn't just about rendering. What we really seek to do is we want to solve hard problems so you don't have to. That's kind of our mission. And so in doing so, we're always looking for what will make the most impact to you. Not just in your image, but in your workflow, and how you make decisions, what you can do for your customers. So this allows us to play with a lot of the latest hardware, explore new possibilities, and talk about things with you in terms of like what you think should be possible, or what ifs. We love that question, what if?

So if you're new to rendering, and many people are. The world kind of starts out as a really basic model. Such as this, this is just geometry. This is actually a rendering of a basic model. That's where the shadows look so good. But the rendering process really adds life to that by bringing in what we call shaders or materials, and the true lighting. And I think this image here, is very appropriate for talking about scaling and the cloud, because it's really about when you have a time problem.

When you run out of resources, you have a deadline, you have to get it done. That's when the watch comes in. And you're very, very conscious about how much time it takes. And so that's where it kind of taken about managing time. And time, from a rendering standpoint, is all about scaling. And it's about efficiently using the hardware. And so, this is one of the things we take the most attention to when we design our rendering software V-Ray is how well we scale. How do we get every inch of performance out of every single core? Out of every single socket?

And by scaling it means how efficient are you across those cores? We're in the very, very high 90s, like 98, 99th percentile when we go across processors. So it's nearly linear. So if you add more cores, we're going to go that much faster. It's pretty easy math.

And retracing itself is a very parallel problem. And you would think it would be really easy to multi thread this, and to make it go fast. But there's so much else going on, it's not that easy. So there's so much you're dealing with, the scene, and file IO, and everything else, that the entire process of rendering, you have to be very careful to scale this well. And that's what we try to do.

We also allow you to use your choice of processors. So we have another mode called V-Ray GPU, where we can run on Intel processors, or in video GPUs. Or use them both together. And this allows people to pick the best of both worlds for what they need. Often what happens is that artists will use the GPUs in their workstation. And then when they go to do production rendering on their farm, or in the cloud, it's always on CPU. And what we do is we make sure the results are exactly the same.

So it doesn't really matter what type of processor you're using, you get to choose what makes the most sense for you. And most of the time, that's either high performance for when you're making decisions, and economy when you're doing production. That's usually the two trade offs that people make.

And I want to come back to this slide of Etienne's because we love this slide. We use the Embree architecture that he talked about-- and we've used it for quite a long time-- to get as much as we can out of the processors. And for us, what is most important here, is it just works.

We can take software that we shipped back in the early 2000s, and it still works in the latest Intel processor without us having to do anything. OK? And that's just wonderful. And when it comes to scaling, it is just brain dead simple. We get exactly the same linear performance as we add cores and processors. So that becomes really important as we start looking at scaling in the cloud.

But first, let's talk about scaling outside of the machine. So first of all, you look at how fast you can make your machine. Then you look around the office, and decide how many machines do I have? When they're not being used at night, how many machines do I have? And you start to put together as much of a render farm as you can. How many people here network render? OK, how many people here have never network rendered? OK, couple of honest people.

So to do that, you are really sending a job to other machines, and letting them do the work for you. It relieves you from having to lock up your machine for rendering. And you can get on to doing more creative tasks. You do need something called a Queue Manager, or a resource manager. The one that comes in the box, right here, is like deadline. Sorry not deadline, is back burner, comes with the obvious products. You can get a more sophisticated one that works with a lot of products, called deadline, that comes from Think Box software, which is part of Amazon.

Once you put all that together, you can harness all the machines in your office. And you can get some animations done much more quickly than on a single machine. And there's really two approaches to this, you can either export-- you can either run your machine in what's called headless mode, you put Max Maya, et cetera, on all the different machines, and you have it process it just like you do your local machine. Or you can export to an intermediate file, a rendering file, and have it run in standalone mode. So those are the two approaches, and they both have pluses and minuses.

So with headless network rendering, it's really simple concept. What was working on the artist station? That's what you put on the farm, that's easy. And usually, this works really well, because the host application usually, is completely free to do this. You're allowed to do this on as many machines as you want to according to your license agreements from Autodesk. So it works really well. But it's not that easy.

So there's a term called asset wrangling, which means you have to get your lasso, and figure out all the assets that were used in that project, and make sure that those assets get on to every single machine. And you have the same version of all the software on every single machine. Down to the funds, so it is-- There's a lot to gather up. So jobs can go bad basically, because files didn't get copied, network paths were wrong, different things. Sometimes you don't have permissions from the right machines, little things like that can get you.

So the act of wrangling, if somebody in your office is doing this, that's can be time consuming to do it right. Or the most common thing, often, is sometimes a cot next to the render farm, because the person is babysitting the job. People are laughing because they know what I mean. OK. The next step approach is called standalone rendering. And here, the really good thing about it is that the renderer has all the resources. It's not sharing it with anybody, the renderer is entirely greedy, it takes it all over. It doesn't share it with Max or Maya, for example.

And there's no asset wrangling, because everything has been packaged from the artist machine and sent off to the farm. There are some cons though. Exporting can take a little bit of time. And those files can be big, because we're processing it all and baking the final result from what was done in Max or Maya as an example. And then, there can be sometimes when you're using third party plugins, that those are not respected in the standalone mode, because they have a special type of map or material that only that plugin knows about.

We are working with all of those companies. It's one of the reasons that we take usage information from you, when you use our renderer, so that we can tell what is most commonly used. So that we can then work with those companies to make sure that they get supported in our standalone mode. But you only have so many machines in your office. And at a certain point, you don't want to buy any more machines. Especially when they're coming out so quick, and you're like, well, that seems old already. So

This is where the cloud comes in really nice, because it's when you really have a time problem. You've exceeded the resources at your own office. And you need to get the job done. This is really the main reason people come to the cloud first.

Now the beauty here, is it can scale to anything. OK? You pick your cloud provider, maybe it could be Amazon, Azure, Google, take your pick. They have so many resources, you are not going to run out. Just think of it that way.

Then you only pay for what you need and when you need it. That's the beauty. So if you've got a crunch job, you're only going to pay for those resources at that point in time. You don't have to buy them forever or even for a long time. And this basically means you have a supercomputer at any machine that you want to. You could be at a laptop in Starbucks and basically have the same rendering power as a major motion picture studio.

Then there's one other thing that is really kind of cool, and it's what I call the economy magic within the cloud. And it starts with this, which is how much does it cost. Now it's weird, but this is actually the exciting part for cloud rendering, which is where the economy comes in because there's a couple of magic tricks here.

Number 1 is the cost per core. As I add more cores, I pay exactly the same price. What that means is I can pick two choose-- I can basically-- as long as the render is scaling the same, as long as I scale 96 times if this goes a factor of 2 and this goes a factor of 96, I'm paying the same amount for going 48 times faster. When does that ever happen in the world? So that is one magic trick that's really great in the cloud.

The other thing is that as I pick these bigger machines to render on, look what happens to the memory. The memory keeps getting bigger and bigger, and I'm not paying any more for it. Another great magic trick.

So pick the biggest instance. You get-- it's not going to cost you anymore, and you have more memory than you could ever use. It's a great combination.

And then there's the final magic trick, and you have got to be a pretty good magician on this one. And it's about using what are called pre-emptable or spot instances. This is when the instance can go away when you don't necessarily have it the whole time. But the savings are huge. So suddenly I'm paying one fifth as much as I do on what's called a reserved instance. So this is really cool. If I can figure out how to get my renderer to pause to actually stop if I need to stop and pick up later, I save a lot of money.

So how do I get that magic? Well, first of all, you're entirely dependent upon the software. There's nothing that you can do if the software doesn't permit a lot of this stuff. The software has to scale well in order to get that linear economy, and then it also has to be able to be what-- we call it resumable rendering so the ability to pause.

So this is a feature we put in v-ray a couple of years ago. People who are like it's cute to pause. I can pause and come back, but this is what it was really for so that we can use machines that cost one fifth as much in the cloud. And so we can pause and resume all the time. That is really important for you in the economy for cloud rendering.

Now so I'm going to use the cloud-- got me excited. How do I do it? Well, you can't do it entirely on your own, and this is actually the most cost effective because you're entire-- have complete control over your expenses, and there's nobody in the middle. So without a doubt, using the likes of AWS, GCP, Azure, and the like, you can totally do it yourself and it will cost you the least.

How many people here have ever tried to use a cloud rendering service or anything in the cloud other than Autodesk 360? So you actually did it.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: So it was actually a cloud service. We'll get to that next. So tell-- how many actually set up a service yourself? You actually went and said I want this instance, I'm going to provision the instance, I'm going to do everything? I'm going to install the software and everything. No one's done that yet. That's what this means. That's what this means.

So if you-- and the larger you are as a company, the more this is interesting to you because you probably have a full-time person that can do it. This is what the studios will do because they have full-time people because this is hard. Because this is the cons. The cons are you have to figure out how to do it, and it's not for the faint of heart.

You have to learn all of the mechanisms of how to interact with the various providers. They're similar, but they're different. You have to learn what's called provisioning an instance, which means you have to basically create a virtual machine and copy that up and get it installed on all of the machines. And sometimes that may even include the right graphics driver. And you have to figure out your licenses.

So the good news when you're doing it yourself is that you don't have to worry about max or mile licenses. You're allowed to go and do this on your own. It's called bring your own license. That's OK, but everything else probably does need a license. Renderers, plug ins, sometimes even scripts need licenses. So you have to figure out how to get all those machines that you rented suddenly to get licenses. That's tricky.

And then, of course, asset wrangling, you have to get all the stuff that you were doing on your local network up into the farm as well. And you have to upload each time because once you stop paying for those machines, your data is gone unless you're also paying to store it up there, which you can. And then you can get it-- minimize how much you're uploading each time.

The other thing is that there are a lot of little charges along the way. So you have to bandage all those, too. There are charges for upload, download, storage, long-term, short-term storage, and, of course, compute although the compute part, that is going to be the lion's share. That's going to be over 90% of your costs because this is all about crunching numbers on CPUs.

And then there is a bit of risk. You're going to be paying the provider whether it worked or not. Because you were cranking out those machines, it doesn't really matter. You got the use of the machine. The software, that was your side. So that's a little tricky.

One thing to note here is that the deadline software that I mentioned earlier from Think Box, they help a lot with some of this. They make it so it's easy to get instances, and they make it easy so that you can actually get short-term licenses for the software that you need. So they've made this a lot easier than it was just a couple of years ago.

Now the other way to go is to use a cloud rendering service. And this is more like what you were talking about. Yeah. And this is more common, especially for a smaller company that doesn't have a full IT person to go do that. And examples of that are like Zync, Render, which is part of Google now, Rebus Farm is a very popular one in Europe. There's many of them out there. We actually service v-ray to know nearly two dozen of these services around the world, so we're pretty familiar with what they need and how they're working.

So the nice thing is is that the service now handles all the asset wrangling. In fact, when you pick services, that's kind of how they differentiate. They are really like how easy am I making it for you to get your project to the cloud and reliably rendered. That's what it's all about. Can I get it done? Do I get the result back? And so that's really how they differentiate besides price is how easy is it.

And then they're also then handling all the licensing. That's not your job. One thing to note, though. While you got to use max [INAUDIBLE] free in the cloud, the services don't. So they're going to have to pass that price along to you. Just one thing to realize.

So the cons, well, it is a bit more expensive because they need to get paid. The services obviously, they're providing extra value to you, and they-- it's going to be more expensive.

I've mentioned that they often have to pay for host application. That goes into the extra price. If you're using standalone, though, if you're working with a service that you stand alone, then that's not an issue at all. You do have to upload each time. That's one of the basics of it all and usually don't know the results until it's done. So it's like you ship it off and you hope it works and you examine it at the end. Not too many of them allow you to preview it as it goes.

And then neutral you still need to know what you're doing a bit. It's not for every person in your office. More likely you'll pick one person that gets to know how to do this, and they'll-- it'll just be part of their job. But it does-- and then some of them also-- it sounds liberating, but maybe it's not. They actually have you pick the exact hardware you're rendering on. Some people do want to do that. Some people would rather just have that taking care of for them, but a lot of them do let you get very detailed in what you're picking.

Well, getting back to one of our mission statements, which is we're here to solve problems for you. We thought that things could be easier, and we started thinking about what if it was just built in. What if it was just one click? What if you didn't have to deal with any of the details other than seeing how the job was doing? And, of course, let's make it really efficient and make it cost effective.

And so that's what we're doing right now. So starting last August, we started a public beta of something we're calling v-ray cloud. And we do believe we set out with that mission to make the world's simplest cloud rendering. We want to make it super simple while still being really, really useful.

So that was the major tenant. This thing had to just work. It just had to be so easy. And so what we've done is we've added a simple cloud button into all of our products. So if you have things like Revit or through 3ds Max or Maya, your latest versions will now have a cloud button, and the same render settings that you use locally just get sent to the cloud.

You don't have to do anything special. You use the same interface you always have. If you know how to set it up in Revit, fine. You're just doing it the way you always have in Max and Maya exactly. You don't have to learn anything new. Everything's included. When we roll this all up, it will all be one charge. Nothing else will be-- you won't see anything else. It'll all be all self-contained.

And we designed it to be super flexible and efficient. We'll get into some of those details later. But the-- but it starts with-- everything is-- we're running on preemptables, so we're using the lowest cost machines possible. And we're running on headless, so we are in charge of all the licensing. So this reduces your costs-- I mean our costs so that we don't have to charge you as much.

So to give you an idea how this works, here's an example of a production file in Maya. And so that's actually what it looks like in Maya viewport. You send-- you basically click on that little render cloud button up there on the top, and in about 15 minutes, that's the image you'll get back, which will be the rendered image coming from v-ray cloud.

So v-ray cloud itself is all browser based. It is-- which means that all of your control can be viewed from any device. You're not locked to your workstation. This means you can monitor your job even on your phone while you're away from the office and see the progress. You can even-- some people will-- and there's no problem to pause a job. You can just pause it if you don't like it or cancel it. And you can do whatever you like.

You can also-- once the job is up there, you can actually copy it and make all your own changes to it. You could submit a low resolution image, for example, see how it went. It went fine. Copy the job and now put it up to high res all from your phone.

So some of this is just easier to show you live. So if you don't mind, I'll play you a demo. I couldn't rely on the internet connection, so apologize.

Thought I had it right here.

Oh, I know what happened. Sorry about this.

I apologize. When the PowerPoint went full screen, it didn't allow the video to show.

Sorry about that. Yeah. Even with the video, you can mess up.

Yeah, so here's an example-- in this case, we're starting off in SketchUp. You set the rendering resolutions as normal. Basically we send it to the cloud. What actually happens is that the file gets exported, and it goes through an interface. And your only job is to basically give it a job name. That's your only job in between-- that you would normally have to do from a standard rendering.

So here it is. It's been already uploaded. You not give the-- you now give the job, and you can even change your mind from what you did export if you wanted to. Jobs go into projects. You can start organizing them. One of the things we're working on is actually giving you the ability to organize things in teams or as a corporate account level as well.

And you can see the progress of the rendering as it works, so you don't have to wait until the end. You can actually see it's progress at any point, and then you can just download your files. And the files stay up there for you for quite a while, so you can share the links and everything with other people.

So that was that demo. So some of the things, when I said we tried to make it smart and flexible and everything, is first of all something we call Smart Vault. When you upload your projects, we actually keep them in persistent storage in the cloud. We do this for you for free. It's all part of the service.

And that means your data is already there. That's why you can change your idea on a job even a month later. A month later you can resubmit the job without ever having to go and re-upload.

The other thing that's really great-- let's see. Yeah, so that basically is it about Smart Vault.

The next thing is Remote Control, and that's what I was talking about. You can do it from your phone. You can do it from a browser. And it allows you to pause resume in the, like I said, cloning, which allows you to copy jobs without having to resubmit it. Very powerful. Many people are doing it by submitting low resolutions or submitting just a couple of frames and then going up and saying now we want the whole animation. So many ways that you can see how things are going before committing everything to it.

And then finally, some of the magic happening is something we call Smart Sync. And it works with all three of these. And what we gain by having your data in the cloud is that we never send out the same thing twice. So if you have ever sent up that mesh or that bitmap, we don't care what it was called, what file it was in, what folder was in, if you've ever sent up that chair, that chair will never go up again. It's already there.

And this also has the benefit of auto instancing. You know what instancing is. In AutoCAD, they're called blocks.

Well, let's say you didn't plan on it, and every one of these things in your file is an explicit chair even though they're identical. Well, we'll detect that they're identical and only send up one. All the other ones get automatically instanced for you. You'll get the exact same result.

The benefit here is that your uploads are really small. So one amazing statistic is that of all the uploads that we did last month, the average upload time for an entire job was under 2 minutes. And that's pretty amazing considering how big many of your files are. So the first time you ever do a project, your file may take 10, 15 minutes.

But then as you work on the project and make changes to it, only what changed goes up. So it really works in an iterative design fashion. As you make changes in everything, the cloud just becomes that much faster for you.

So the beauty here is that is is currently free to try. So we encourage you to come and try it. You can come register on our website, and you'll get free Cloud Credits. And with that, you can start rendering for free. I think we have at least one-- how many credits have you gone through so far?

How many? 2,000 credits. So that's about 60,000 core hours that he's used for free so far. And we love you for it. Keep going. But it just shows you that we're serious about this.

And some of the statistics, the bait is going really well. And in just last month alone, we did 3 and 1/2 million frames. That equates to about 45 hours if you were to actually render that out as a single animation, which is about 20 full-length feature films. And that took about 20 million core hours. So people are pounding on this, and the success rate is extremely high. We're having hardly any failed frames. How many-- how's been your success so far?

95%? That's pretty good. We want to get it higher, but I'm glad you're finding it that way.

And this is just an example from a studio in LA. This is A52. They're famous for doing some of the-- like the Game of Thrones intro and things like that. And as a benchmark, they just tried one of their test scenes, and it rendered 4 and 1/2 minutes locally. And then they submitted 100 frame animation, and it took seven minutes to get it up there and back for all 100 frames.

And that's where you start to see the magic of the cloud because we go up and we use the biggest instance we can get, 64 core instances. We're using them as-- and we basically will use-- for every frame you upload, we're going to start up a machine. So when he uploaded that, we lit up 100 machines and did it as fast as we could.

And the beauty is, it cost us the same as if we had used one machine for 100 times longer. It's kind of an intelligence test. It-- you really-- it's the beauty of the cloud rendering.

So again free beta, come to our website. It's on our home page. You can't miss it. We encourage you all to try it. It should work fine from Max, Maya, or Rivet as well as many of our other products.

We have a little bit time, or we are running out? OK, OK. So also up on the website, you'll find a case study.

Now I got to do my--

Let's do a case study that I thought was very applicable to people here at Autodesk University. It was done-- company was Till Pixel. They were specializing in architectural visualization. This entire-- this project was done entirely in v-ray cloud on the beta. And it's pretty impressive for what they were able to do in a very short amount of time.

But when you're doing architectural visualization, you often concentrate on stills. And you forget how convincing even just small pans of animation can be to convincing-- to really show what your design is. And so this is like, well, if we gave you 10,000 machines, what would you do. Well, let's just really explore the design. And this is just an example of what these people were able to do very easily by just rendering on the cloud entirely and exploring just a lot of animation pans that they probably would not otherwise be able to do.

[INAUDIBLE]

I think they talk about some of the statistics about how long it took. But the competition was only open for a month, so they turned it around pretty quickly. But you can read all about it here. We try to post-- there's a couple other ones up there as well, but this one was specific of what I thought was of keen interest for what you guys do.

Etienne, back over to you.

Oh, absolutely. Sure. Go ahead.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: Yes.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: The question is when will it-- when will-- what we have to charge for it and when-- how do you actually plan for what you use in the cloud versus what you have on prem-- on premise? Both great questions.

The first one is easy. Thanks to the beta going really well, it looks like we will be in full commercial deployment in first quarter. So sometime in the first quarter, we should be live and then full featured.

We will be competitive on pricing. We're not announcing any pricing yet. We will be competitive. But we're not-- we don't believe that we're going to be-- you're going to be choosing us for price. You're going to be choosing us because we think we're going to be the easiest to use-- easiest, most reliable to use. We think so. So we probably won't be the cheapest as cloud services go.

As far as planning for what you should purchase versus not purchase, it really comes down to how you run your business. It-- many people start out by-- it's only for time crunches. And you pass that cost directly along to the client who's paying for it.

So there was a deadline. It was a rush job. It had to get done. You pass that along. That's pretty easy to figure out.

What we find out was over time, people get really comfortable doing this, and they start doing it all the time. And when that happens, that's when it starts to impact how much hardware you're buying locally because you're just using it so often, you're not doing it as much locally anymore. It's both ease of use. The machines in the cloud are often much more powerful than what you have locally.

And so sometimes you just can get-- it just becomes more streamlined. You get used to it. It becomes part of your behavior. You don't even think twice about it. You're just-- you're working on your station. You fire after the cloud. You keep on working. It's all in the background.

So it shifts. So take studios, for example, who are very serious about where they use resources. And they will schedule things. They will do all like their internal tests locally, but then when it comes down to final production, they'll burst for of the cloud.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: Yeah. It's really like--

AUDIENCE: [INAUDIBLE]

PHIL MILLER: So most people will equip their local if it's sustained, if they're using it for a certain amount all the time. And then that-- the economics are going to be better. You're going to be better off having a local if you've got the floor space and everything else. But then it should be more relaxing now because you always have the parachute to go to the cloud.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: Yeah, sure. Go ahead.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: Yeah, it does. Yeah, it does. So basically if it's been set up for ren-- so the-- all the things necessary for v-ray to understand your desires are in what's called the VR scene file. And that is all that's necessary to render a frame. So it's just a self-contained unit.

Now to double check ourselves, one of the features that we put into the last 3ds Max version was a cloud check, which basically lets you run standalone locally. Believe it or not, we install standalone all the time for you. You may never use it, but it's there. And so what we did is we automate it so that you can do a verification locally before even bother with the cloud. And that actually also helps you if you're using standalone on your own farm, too.

So there can be little hiccups, but usually they're plugin related. They're usually not v-ray settings related. Although to be honest, we've been catching a few those in beta, and we've been fixing them. But by the time-- we encourage you to come in with your most challenging setups and put it in there.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: Yes.

AUDIENCE: [INAUDIBLE]

PHIL MILLER: Yeah, all the final results are baked. So the-- so, for example, if it was a procedural object in Max, it's going to become a mesh. If it was a-- if it was a plug-in that did something, only the end result is shown. All that. Yeah, so it's all self-contained. That's why we don't have to run plugins or the software in the cloud. It's already been done.