CAD and the Metaverse

Announcer:

Today on Building the Open Metaverse.

Carl Bass:

We are moving to reality capture, where we can go out and scan the world and use that as input and merge it with digitally created things. And that world I think... I haven't seen any of the design companies do a particularly good job on the assets.

Announcer:

Welcome to Building the Open Metaverse, where technology experts discuss how the community is building the Open Metaverse together. Hosted by Patrick Cozzi from Cesium, and Marc Petit from Epic Games.

Marc Petit:

Hello everybody and welcome to our show, Building the Open Metaverse, the podcast where technologists share their insights on how the community is building the Open Metaverse together.

Marc Petit:

Hello, my name is Marc Petit from Epic Games and my cohost is Patrick Cozzi from Cesium. Hi Patrick, how are you today?

Patrick Cozzi:

Hi, Marc. Hi, everybody. Doing great.

Marc Petit:

So we're very excited to welcome to our show, Carl Bass today. Carl is a former CEO of Autodesk where he continues to serve as a director, and he also on the board of several other companies like Planet, Develop3D, Formlabs, Built Robotics. And in addition, he's a known investor in deep tech and heart tech startup and has been doing. He's been also a long time advisor to Google.

Marc Petit:

Carl, you seem to be everywhere and we're very glad that you could come to our show. Thanks for being there with us today.

Carl Bass:

Ah, it's a pleasure to join you guys this morning.

Patrick Cozzi:

So Carl, we love to start off the podcast by asking our guests about their journey to the Metaverse and in your case, you've had a very storied journey. Rewind us back to your time at Cornell and getting into computer graphics and starting ethical software, which was ultimately acquired by Autodesk.

Carl Bass:

Sure. Yeah, it's one of those things that wasn't planned, certainly. I was a student at Cornell. I was studying math. I actually dropped out of school for a while. I dropped out for about five years and went off and built all kinds of things and we'll return, I'm sure, to that topic of building physical things.

Carl Bass:

So yeah, time at Cornell was interesting. I was studying math. It was a time in which the math department really didn't believe in computers. I actually had a math professor say to me, "If you can't solve a problem with a pencil and a yellow piece of paper, it's not really a problem worth solving." And I eventually dropped out of Cornell for about five years, went off and built all kinds of things and we can return to that topic of building things. And I came back and the math department had a complete change of heart and all of a sudden in order to get a math degree, you had to study all this computer science. And so the last couple of years I was there, I just studied computer science.

Carl Bass:

In just one of those twist of fate, I used to play basketball at lunch every day and one of the guys I played with was one of the true pioneers in computer graphics. His name is Don Greenberg. He was a professor at Cornell and we were sitting there one day after, in the locker room, and he said to me, "You're studying math, right?" And he asked me a bunch of questions and he said, "I have this paper from NASA, I'm doing this algorithm and nobody has the mathematical background. Would you like a job trying to program this algorithm?" And I said, "Of course", because at the time I was making $2 an hour picking rocks out of a cornfield, which was my summer job. And he offered me three times that to actually work on something that was way more interesting.

Carl Bass:

And so through this crazy, random chance of playing basketball with Don every day, that was my beginning to computer graphics. I worked on that algorithm at that time. I met up with the person who was eventually the co-founder with me. We founded this company called Flying Moose Systems and Graphics. When we figured out that almost nobody at the time would buy from a company called Flying Moose, we went a little bit more mainstream and renamed it Ithaca Software and all of a sudden business just started to roll in. Still seems to me like Flying Moose was a perfectly good name, but-

Marc Petit:

Yeah, it would work now.

Carl Bass:

Today it would be great. It just didn't work out so well then and to the customers we were selling to.

Carl Bass:

So we started a company and it was all about computer graphics and we ran the company for maybe eight or nine years, did all kinds of interesting early work in computer graphics and then we probably started the company around '81 or '82 and we eventually completely sold it to Autodesk in 1993. And that was my beginning in computer graphics, which was really the background. And in '93 got into the realm of digital design and engineering and CAD and stuff like that.

Marc Petit:

Great. You're actually known to build stuff with your hands. I was the lucky recipient of a gift, one of your first 3D printed pieces where... Actually, I remember it was in 2007 or 8 and I was blown away that you could actually print a high quality metallic pieces with 3D printing. So what are you designing these days?

Carl Bass:

So I work on a variety of stuff. Lately, I've just finished working on two electric vehicles. One is... This started because my kid built an electric go-kart, and so this is maybe seven, eight years ago just before he got his license, he built an electric go-kart. I got kind of fascinated with the fact that this little thing that he built could go like, 60 miles an hour. Scared the crap out of me. Then he went off to college, I made it autonomous, I actually crashed it autonomously, I repaired it.

Carl Bass:

And then I said, "I'd really like an electric vehicle" and I didn't really love any of the electric vehicles on the market. So I went and I bought a 1950 Chevy pickup truck and I converted it to electric. And so that's my daily driver. It's this beautiful old green pickup truck and I drive it back and forth.

Carl Bass:

And then a buddy of mine who's really into cars wanted to build a race car. And so we built, from a kit, a Shelby Cobra. And so we built an electric Shelby Cobra. That thing's a little too scary for me.

Carl Bass:

So I've worked on that. And then I've been working on a bunch of interesting projects with a number of the startups I'm involved with. I've mostly been involved with these startups that are some interaction between the physical world and smart software. And at some point, figuring that the same benefits that came to the world of computer graphics through Moore's Law and all the increased capabilities, it still left much of the physical world behind. And so I've been really involved in the intersection of these two things. And for a number of the companies I'm involved with, I fabricate and help design things. Usually at the point of building prototypes or prototypes as they go to scaling their production, somewhere in there I help out.

Carl Bass:

So I've been helping out one company that's growing meat from stem cells and they need physical containers to do it in. I've been doing some work on some aerospace stuff, making lightweight parts. Whole bunch of stuff. Actually, I try, I'm not always successful to document them and I stick them up on my website. I actually have a website at carlbass.com that's filled with the pictures. And mostly what I try to document is the process of doing it. I try to document the process and I try to document the failures because everyone thinks these things just come easily. And I actually find it's really through the mistakes and no matter how many times you've built things, you still make mistakes.

Carl Bass:

As a matter of fact, I'm sitting here in the shop this morning and one of my favorite parts of the shop is the wall of shame. And the wall of shame has parts that every one of us has messed up at one time or another, some more dangerously than others, but it is filled with a collection of things you wish you never saw in a shop. But it's really to counteract this kind of narrative that building things is easy and when you're good at it just comes naturally and other people should be scared and, no, you make mistakes, you learn from your mistakes. Unfortunately, sometimes physical mistakes are a little bit more dangerous and a little bit more expensive than software bugs. But that process of creating something and building it a second or third time to try to get it right is really very much the same.

Marc Petit:

Yeah. So what's your go-to design tool these days?

Carl Bass:

Oh, these days I mostly design almost everything in Fusion. So Fusion 360, which I was involved with when I was at Autodesk.

Marc Petit:

Yeah, that was your brainchild, it's fair to say it was your brainchild.

Carl Bass:

Yeah, I worked on that quite a bit. And so that is the tool I probably use the most. And it's mostly because I do both the design and the engineering.

Carl Bass:

I was actually looking through my drawer this morning for a microphone, and I saw this old copy of, actually, a DVD of SOLIDWORKS and I thought, "I think I really don't need that anymore." I don't even know why I had it.

Carl Bass:

But, yeah. So Fusion's the tool. Occasionally I'll use some of the newer tools. I'll use nTopology for different kind of design explorations. Been doing some work on generative design, so I'll try out some of the tools that do that. When push comes to shove, I still write code. So I'll write Python and use APIs. Occasionally, I'll use something like Maya, which Marc and I have in common from our past. So I'll occasionally use some other tools, but the one go-to tool every day is Fusion.

Marc Petit:

And I do remember the excitement around additive manufacturing and 3D printing where 15 years later, has that technology met your expectations?

Carl Bass:

I think in some ways it's succeeded and in some ways it's fallen short. So I think in the place it's exceeded is in the geometric complexity that you can print. So even 10 years ago, that bowl that you held up, kind of phenomenal. No other easy way to do it. You couldn't cast it, you couldn't machine it. And so 3D printing is relatively agnostic to geometric complexity. It doesn't care if it's a cube or something as complicated as that. And people are finding all kinds of uses. And most of the uses for it seem to be, or the best uses, are the multi-physics things where you have a metal part and some fluid running through it or air running through it and you need channels. So there's a handful of applications where it's really superb and does something that nothing else can do.

Carl Bass:

And I'd say on the beginner side, I think it's awesome that there are sub $500 printers. I can't tell you the number of people who call me and say, "What printer should I get for my kid?" And people are printing out soldiers or figures or whatever. And it is a great introduction to that thing of designing and making.

Carl Bass:

So in those ways, it's succeeded. The places where I really think it's fallen short is it's still relatively slow. It is not a mass production tool for almost anything despite what people want to say. Sometimes the overall throughput is faster because, for example, you don't need to create a mold to cast something. So making ones or twos or threes are incredibly fast, but it mostly hasn't made it out of prototype or limited production. Handful of places where it has, but that's really where it's at its best.

Marc Petit:

By the way, we're going to have Ping Fu on the podcast. I'm sure we are going to go back on the topic and have her perspective on that.

Marc Petit:

So you mentioned generative design. During your tenure at Autodesk, you drove a lot of innovation, procedural and rules-based design. Remember dynamo? And generative design was an early form of machine learning. So how about... Is generative design impacting design the way you thought it would be?

Carl Bass:

I think we're still early. So just to back up a little bit, here was the inspiration or insight, is that generally people design things and they do multiple iterations and at some point they run out of time or money or patience or interests and they go, "Good enough." And sometimes in the engineering realm, they double up the size of this, they add in a safety factor.

Carl Bass:

And many of the things that people design are very similar to things that have been designed before, sometimes even by them. And so the inspiration was that computer-aided design or CAD, the computer really didn't do much aiding. The computer was really just a documentation tool. It was a way to record. And so you sit there and you mouse around and it essentially records what you are trying to build. And instead, it was this idea that the cost of computing by Moore's Law was getting cheaper and cheaper and in the limit, it was approaching free.

Carl Bass:

And so we talked a lot about infinite computing, this idea that it was going to be really available and really cheap, and that's come to pass. You look at a CPU hour, whether it's on Azure or Amazon, and it's ridiculously cheap. And the idea was that we could optimize design by having the computer run multiple iterations, and by multiple, I mean thousands or tens of thousands or hundreds of thousands. And what you would do is you would specify to the computer what was important. So for example, I want to build something that's about this size, these are the load conditions on it, needs to have two bolt holes here. Please find, and this was a simple optimization problem, find me the lightest structure that actually does these loads.

Carl Bass:

And where I'd say we are right now is there's a lot of people doing it, it's made for some innovative design. I don't think it's completely polished yet, but many of the designs that comes up are really great inspirations that are stepping off points for use for design.

Carl Bass:

And particularly in places where there's complicated physics, multiphysics going on or in places where weight is a big concern, like in aerospace applications, you're seeing some great uses of generative design.

Carl Bass:

So still early, there's more to be done where people are starting to add constraints to the generative design, if not only these are the loads, but things like the manufacturability. And I think that really increases the utility if you say, "I'm going to make it on a three axis or a five axis machine", or, "I'm willing to do additive manufacturing." Those are the kind of things that help. I still don't think there's enough.

Carl Bass:

And this becomes this interesting philosophical question about, how do you talk to a computer about aesthetics? So if you notice many of the shapes that come out of these generative design tools, quite organic and quite lovely. Sometimes they're a little bit irregular, they're not quite symmetrical, and designers have other things in mind. And so I think in the next go-round, I think you'll see people add controls and ways to inform the computer that these are the kind of properties you're interested in. But I think it has a lot of legs and in the future, many of the things will be designed by the computer as opposed to being documented by the computer.

Marc Petit:

Fascinating. So let's switch from the real world to the virtual world, and the topic of the metaverse. So I think we all agree that data and models are incredibly important and people tend to think if you control the data, you control the ecosystem. And it seems to me that in the design space, the value shift seems to have shifted from CAD data to digital twins, which seems to be the representation that has the most value. And you're still the director at Autodesk, so you're still a-

Carl Bass:

Oh, I'm no longer a director.

Marc Petit:

Oh, you're no longer a director.

Carl Bass:

I'm no longer a director, but I'm still closely tied.

Marc Petit:

So do you see this as well, and what does this mean for CAD companies, that mindset shift around owning the data and control as well as the shift to the digital twins?

Carl Bass:

Yeah. So look, let's just back up a ways. Go back to those early days of computer graphics. And it was funny, I was listening this morning to the podcast you did with Ed Catmull and it was a trip back in time. As I remember the early days, we were simultaneously working on graphics, algorithms, things that are fundamental about shading polygons and hidden surface and hidden line algorithms. And we were doing three polygons at a time, never imagining the complexity that would be possible.

Carl Bass:

And so you go back to that time... So my beginning and the training I got at the time is we were trying to make things as lifelike as possible, to mirror reality and the physics of the real world. And at the time there were really two uses of what people were doing with computer graphics. One was in the realm of entertainment. Ed spoke artfully about all the things that were done in the world of entertainment and certainly was a leader there.

Carl Bass:

And then there's the other side that was more on the design and engineering and science side. And they were both important, but the requirements were somewhat different. Interestingly enough, these two paths are kind of merging now. As we now represent things with pretty lifelike qualities to the point of, it's really hard to tell. If you see a still image, is that real or is that not? Is that really the building or is that a picture of the building? Some of us who are well-trained can see it, and even some of us who are well-trained are easily fooled these days. And we're getting to the place where it will be pretty indistinguishable by anybody. So we're now at that point where we can represent stuff.

Carl Bass:

So back to the topic of economic value. Look, there is huge value in creating these designs and the tools that are needed for design exploration. I think that will continue to be valuable. That there is... And the biggest value is not just the design. It's, how do you take that design and bring it into the real world? And bring it into the real world because you have to manufacture it or you have to build it.

Carl Bass:

There's this entire workflow where you need to communicate with literally dozens, if not hundreds or thousands, of people what the designers' intent is. And that aspect of it is really important.

Carl Bass:

Now, there's an idea of what is that digital replica? And the closer we can get to capturing these things as true digital twins, the more value moves to that side of the equation. And you've certainly seen it in the beginning was photorealistic renderings of people's designs. We are moving to reality capture, where we can go out and scan the world and use that as input and merge it with digitally created things. And that world, I think, I haven't seen any of the design companies do a particularly good job on the assets.

Carl Bass:

But when I think about the business side of this, I always used to... The hair on my neck would go up when people said, "This company owns the customer." No company owns a customer. And in the same way, I don't think owning the data... The data really belongs to the customer, and they will choose what to do with it that best serves their needs and they will choose a set of tools that do it.

Carl Bass:

So if I look now, there's probably less than a dozen tools in the world that are used to create 90% of constructed or manufactured products in the world. I don't see that disappearing or becoming less valuable. What I think what companies have managed to do is cut off the next step in the chain of how these things are used in the future. And I think it's a different set of companies that are doing that and doing it considerably better.

Carl Bass:

And so I think most of the design engineering companies will head into, how is this thing made? How do I communicate with the people who make it? And I think there will be partnerships between those companies who can really maintain the digital twins and do it justice and the original designs. And I think there'll be a handful of new companies who come into fill because trying to understand the difference between the original design and the physical manifestation is really important. So whether that's as-built, here's what the blueprints say we're going to build, here's what we actually built. And if you can understand the differences, it's incredibly important.

Carl Bass:

There are also uses where I don't think the CAD data is particularly the right level, but I've also seen companies... So one of the companies I'm involved with is actually an interesting merge of both, doing generative design and doing downstream use. The company's called Higharc and what they're doing is generating residential homes, designs, but it includes things like renderings and walkthroughs and everything like that. And I think it's particularly an interesting thing because in order to create the designs, number one, you're no longer, as I joke, mousing around. You're not sitting there clicking and making walls and doors and stuff. You're talking about it at a higher level of abstraction. You're saying, "I want a three bedroom house with an attached double garage," and generatively the computer actually makes that design. At the same time, it can make blueprints, it can make walkthroughs, it can make renderings. And I think that, to me, is a window into the future of all design, where we will be specifying things at a higher level and having the computer do more of the work and have the output be more complete than it is right now.

Marc Petit:

And it is guaranteed constructability as well, because constructability is built into the design at the heart of the design.

Carl Bass:

And you can test for it, you can compare against as-builts versus as-designed. So I think there's a lot of benefit to come from that. And I think those guys are one of the early ones doing it, but there's no reason why this can't apply to commercial construction or many of the manufactured, cars, planes, consumer products.

Carl Bass:

One of the things that I think has historically been one of the limiting, I don't know, visions or something, but everybody's always talked about digital twins and they think about it through the one sense of what they see. And what I've always been interested in is this idea of, how do we experience more of the things we're designing when it's on the other side of the glass? It eventually comes over to this side and it's different. It has other properties. It, first of all, getting scale and proportion right on the far side of the screen is difficult.

Carl Bass:

But what I long for is, for example, taking the many facets of a design and being able to see them work together. So I've always had this idea that we should be able to take an object and, for example, turn it on. And we should see lights flash and you press a button and you understand how it operates. You should hear what it sounds like.

Carl Bass:

And so when I think of the future, I think many of the views of digital twins are incredibly limited to just the visual. And while that's certainly incredibly important, if I was a designer, I want to really understand the scale and proportion. I want to understand what it feels like, I want to see what happens as I interact with it. So these products that we design and make are so much richer and certainly compared to where we started this conversation 40 years ago when we were drawing a polygon at a time, the renderings of these objects are phenomenal and I don't want to discount the value in that, but it's really not all there is to it. And there's really a lot more that needs to be done so that digital twins become as rich as the products themselves.

Patrick Cozzi:

Yeah, Carl, and we couldn't agree more. A lot of the conversation on this podcast has been about the digital twin needs to be so much more than the visual.

Patrick Cozzi:

And you also said something recently that really resonates with us. You said that "the customer owns the data," and we explicitly call this podcast "Open Metaverse", and the idea that the customer owns the data and it's interoperable as a key point.

Patrick Cozzi:

So we wanted to pivot a little bit to talk about digital twins in the context of real estate. So it seems like owner operators will want the digital as much as the physical, because it's key to managing the performance and operations and building, yet it seems right now the construction companies remain "hard hat." Do you think they'll ever get into the opportunity for creating, delivering, and managing digital twins?

Carl Bass:

So there's a lot to unpack in that question, Patrick, but let me try to do my best.

Carl Bass:

So the first one, as somebody who has built tools for the construction industry for decades, the first thing, I think the construction industry gets an entirely bad rap. And when we talk about the construction industry, it's a multi-trillion dollar industry and most people's view on the industry is the guy who showed up with his dog in his pickup truck to build their deck. That is not the construction... It's a tiny little piece of it. And it's a little bit like talking about the computer industry and saying, "My nephew has a PC and he writes Python code." Yeah, that's interesting, I'm glad he does, but that's not the computer industry.

Carl Bass:

So this thing about the construction industry being knuckle draggers is just I think a little bit off base. What I've seen is the way they build buildings today looks nothing like how buildings were built 20 years ago or 40 years, or certainly a hundred years ago. The first things that go in at sites are total connectivity. There are screens and phones and tablets on every commercial construction site.

Carl Bass:

And so the first thing is, I think the understanding of the construction industry really comes down to they are technically, very sophisticated. They use materials, tools, and processes today that they didn't use a generation ago. On the other hand, unlike much of the computer industry, this is a low single-digit margin business. It does not have the luxury that many of the tech businesses have. A single project can make or break a construction company. And what this leads to is this M and M practicality.

Carl Bass:

If your business... When we were at Autodesk, we had profit margins 20, 30, 40%. If you're running a construction company, it's low single-digits. It's 2, 3, 4%. So you can't afford for a project to go wrong. So before you adopt new technology, you have to be sure that it's going to work and that it's going to add value. And so in that way, it's a much tougher buyer, but it's not speculative.

Carl Bass:

In the way there's all kinds of stuff you can sell from high-tech company to high-tech company and people will just kick the tires for a million dollars. No construction company's going to spend a million dollars to kick the tires.

Carl Bass:

So let's just put that to the thing of, they're technically sophisticated willing to change, but it has to have commercial value. So that's the first part of it.

Carl Bass:

The second part of it is the contractors who build these buildings really do what's paid for. So you have to look at the other economic structure of the industry and I think you nailed it correctly, Patrick. I think it's the owner operators who are the ones who will do it. The people who build buildings and flip them, fill them up with tenants and flip them, they have no interest in the ongoing operation and maintenance of that building. But the people who do, and whether that's a chip fab company or somebody who's building hospitality or retail like a Home Depot or a Starbucks where they only construct their own stores, they're really interested in the lifespan and understanding the lifecycle of the things in those buildings and they need to do the maintenance on them. And in that way, I think there's just incredible value of tying real world data to these models.

Carl Bass:

And whether it's maintenance and repair information, real estate information, this should be the tool that people use. This should be the foundation, is I have a model of the building, I understand what the HVAC system's like, I understand the electrical system, I understand where the desk and chairs are, and when I go to remodel or rebuild or reconfigure or just maintain and operate, this is the tool that's used to do that.

Carl Bass:

And I think we will get there. And whether it is some form of augmented reality or just the next extension of what we do on our mobile devices, to me it's the natural extension.

Patrick Cozzi:

Yeah, well said. And good insights into the industry.

Patrick Cozzi:

So now let's talk a little bit about digital twins-

Carl Bass:

Yeah, I just had to defend my construction friends. They always get beat up for doing hard work.

Patrick Cozzi:

Oh, we meant no insult or bad will to them.

Patrick Cozzi:

So I'm going to switch gears a little bit and talk about digital twins for consumers. So it's not hard to imagine a future where we have digital twins for everything and potentially platforms like Fortnite or Roblox or Matterport could become a platform where we monitor our homes or we host virtual events. Do you think this kind of future will happen?

Carl Bass:

Yeah, let me say this. I feel like I have some insight and intuition into what professionals and in some of the industries I've built tools for over the years. One of the interesting things is when you're a tool builder, people in the industry share with you, as I'm sure both of you know, their insights. I always thought of it as you're an instrument maker to a great musician. And so they always share their secrets in the hopes of you can somehow help them fulfill their creative dreams.

Carl Bass:

And so I've spent years and years talking to people and feel like I have really good insights around many of the commercial people using it. When it gets to consumer, I got to say, I just feel like there's a big ice cream cone planted on my head. I am often confused and more often just wrong about what consumers will do.

Carl Bass:

So as you point out, is there value there? Absolutely. I was just thinking about this the other day as my home is slightly become more digital, and I'm pouring through digitized analog records of, when did I buy this, and what's this furnace and where's the blueprints for it or fixing the grill or the whatever or where in the house is something?

Carl Bass:

It's kind of nuts that we still have this really archaic way and whether... I just dug up my yard to put in some fence posts and "call before you dig" is the most archaic thing. I had like, 14 gas and electric trucks out there before I could dig a hole. Then I connected up on a tool with an electric current to figure out where the abandoned irrigation pipes ran. I mean that, to me, for example, is just a perfect use of having a digital twin of reality. I want to know where the current stuff is, I want to know where the abandoned stuff is and whether I'm drilling a hole or digging a hole, I want to be able to do it without the worry that I'm going to blow up the neighborhood.

Marc Petit:

Well thanks. Yeah, that makes sense. So last topics wanted to discuss with you, Carl, today is about open standards. That's a big topic of ours. So last season, we invited Raji Arasu, the CEO of Autodesk, and Dana Colella on the show. We had a good chat around model creation and open standards. So many of the models that we can interact with in the metaverse, originating applications that were built by Autodesk and many under your stewardship, actually. So while you were there and to the extent that you can speak to it now, what are your views on open standards? And in hindsight, what's your take on those standards?

Carl Bass:

I think there's always two levels of ways of representing. I think there's one that's really close to the application and embeds a lot of the logic and the internal thinking of the application. It's pretty close to the internal data structures and just the overall kind of gestalt of how the application works. Those are often difficult to share. When you get to the manifestation, the artifact that's created, it should be easy to share and every company should be willing to share those.

Carl Bass:

So for example, in the world of mechanical CAD, nowadays, most people who are doing mechanical CAD have parametric feature based modeling of some sort or another. The way that's implemented in the half dozen or so different modelers out there is different. And nobody's ever done a great job of being able to really move those models with the complete history around. And a lot of effort can be spent trying and I'm not sure it's actually worth it. We could debate that and I'm curious what you guys think.

Carl Bass:

However, in the end, all the parametric feature based solid modelers essentially make models that have boundary representations. They're filled with prismatic shapes and BREP surfaces and those kind of primitives. I will say, nowadays, that I'm interacting with lots of other people and exchanging data, I can get a model from almost anybody for the purpose of using it downstream.

Carl Bass:

So I often get a model, we're going to make a few tweaks and then we're going to machine it or printed or something else. And that exchange of information seems robust and high fidelity enough for many applications. It's not great for collaborative design. I can't go back and get the benefits of parametric design and it's sometimes a little bit frustrating, but the cost of doing it is high.

Carl Bass:

So I'm a believer. When I was at Autodesk I tried to encourage. It was interesting, we had a number of competitors who were willing to exchange data openly and when we could, we made those agreements. It was always interesting that there were a number who thought it was proprietary. I always thought that was kind of fool hearty. It was shortsighted and it was a little bit insulting, truthfully, to the customer. You know, "You can only use my tools and my tools are the best." Look, nobody makes the best tools, they are just tools in the toolbox and people should be able to take that data where they need to use it and use whatever tool they feel is best for it.

Carl Bass:

And so I think we're at a point where these internal representations don't make it into the world. I don't know what other people who can use, for example, a grasshopper graph. I don't know what to do with a grasshopper thing. The thing it makes, I can do a lot with. I don't know what to do with the internal Maya representation, but I do know how to use assets that are created in Maya in all kinds of downstream things.

Carl Bass:

I wish there was some kind of bridge where we got more of the capabilities and more of the expressive abilities that are in these things without burdening the progress that's made in the creation tools. And I don't know quite how to dance on the head of that pin well enough. But any company who thinks they're going to keep the data proprietary, it will only last for a short time and it will not lead to commercial success over the medium to long term.

Marc Petit:

Yeah, we're thinking shift in the... I mean, powered by the technology developed by Pixar on USD, I think we're seeing that internal presentation, the authoring representation and the publishing of distribution representation. I mean, this converging the flexibility of an architecture like USD is alter natural representations, environments may allow that actually we're in conversation with the Adobe teams and the Maya team. They want to move their persistence mechanism on top of USD. There will still be the private data that you mentioned, but it gets carried around and it will allow an authoring tank and multi-applications workflow. So I think the needle is starting to move and I think we owe it to Pixar, this amazing USD architecture and-

Carl Bass:

No I think things like that are good. And when we go back to the product realm, you got to think about, what do the next set of users, the downstream users want to do? So we talked a little bit about home construction, home design and construction. As a homeowner, I probably don't want to move the studs in the wall, that's not my job. But the things I want to do are, for example, open the doors. Or I want to see it at nighttime, I want to see at daytime, I want to see it at various times of the year and see what the shadowing looks like. Those are the kind of user activities that certainly can be enabled with downstream data.

Carl Bass:

Same thing in manufacturing. I want to machine it, or even in my goal of eventually getting to being able to turn it on. I want to turn it on, it doesn't mean I need to change the parametric definition of that thing to turn it on.

Carl Bass:

And so when you look at what Pixar is doing, and there's a number of companies who are trying to do this, I think the more we enable it, the better off we'll be. And I think the mistake that companies make is thinking that it's zero sum.

Carl Bass:

The one thing I can say after doing this for decades, the industry is huge by comparison of where it's... I think of the first SIGGRAPH where I went to with several dozen people. And the industry is huge in comparison with that now. And I think that zero sum game mentality is really limiting. It's limiting for the customers and it's limiting for the people who run businesses and the more they think about, what's the potential to grow the size of the pie and do better, the better off all we'll all be.

Patrick Cozzi:

So Carl, we've recorded about 35 episodes of this podcast so far and I think pretty consistently we've heard about folks who want to grow the pie for everyone with respect to the Metaverse. We haven't heard a lot about folks wanting to keep data proprietary. So yeah, I think the industry as a whole is very much sharing your sentiment there.

Patrick Cozzi:

And then also with your downstream use cases and those getting more robust, it's interesting. We are starting to see that now in the standards with glTF from The Khronos Group where it's starting to look at futures like behaviors and composition. Then Marc, mentioned USD from Pixar. So we're at a very interesting time to hopefully try to help shepherd the future.

Carl Bass:

Yeah, and look, what I hope is the companies that have said they will endorse this, really see it through. And it's not... I've seen standards before that get a lot of lip service and I just hope the companies really follow through on this and then in return reap the benefits of it. So I think we all understand where we want to go and what's possible and we will be in a much better place if the companies do that.

Marc Petit:

Actually, I think we are in a good place because your friend Jensen Huang from NVIDIA has actually invested quite significantly and prove it out to all of us that the potential technology developed by Pixar goes way beyond its original intended use for film production. So yeah, it's happening.

Carl Bass:

Yeah, absolutely. No, it's an exciting time. It's just hard, like I said, that trip down memory lane, it's hard to imagine as I think back to single polygons at a time being drawn on terminals connected to mainframes. That the machine in front of me with the GPU in it, what is possible today and this world of software that's built on top of it is absolutely phenomenal.

Patrick Cozzi:

Well said. So Carl, we like to wrap up with two fun questions. The first is, we covered a lot of ground, but is there anything that we didn't talk about that you'd like to?

Carl Bass:

I think we had a pretty wide-ranging conversation.

Marc Petit:

And finally, is there a person, an institution, an organization that you would like to give a shout out to today?

Carl Bass:

I mean, I think there's a couple. So the first one, I certainly owe much of my career to both Don Greenberg and the work that was done at Cornell. It was one of the institutions, back in the day I heard Ed talk about Utah. There was work at North Carolina and work at Brown. And so I mean certainly a shout out to the people who pioneered this at those four universities because I think they've made a huge difference. And if you were to look at the genealogy of all of us who came down through that, it's kind of staggering.

Carl Bass:

So I think they did great and I would certainly give a shout out to the folks at Pixar, John Lasseter in particular, for what I thought... I still remember the moment when I first saw the Luxo lamp for the first time, the animated Luxo lamp and said... The technology that we all had a part in creating when it was in the hands of a real filmmaker, an artist, what it was possible of doing. And so I think John and that team deserve huge credit as well as all of the academic pioneers of this.

Carl Bass:

And Ed kind of hinted at it, but it took more than a decade for computer graphics to even be accepted into the realm of computer science. It was something no one wanted anything to do with. That's cute, you playing with pictures. But it's been amazing what's possible. So not only the people who had this vision and insight, but also had the perseverance to see it through in spite of their colleagues disdain for their fooling around with pictures. So...

Marc Petit:

No, indeed. And a little bit of a shout out to the guy who put the power of Autodesk behind Maya to make it what it has become, as well. So Carl, thank you. Thank you very much for that.

Marc Petit:

So everybody, thank you very much. Carl Bass, it's been a pleasure having you on the podcast today. Lovely perspective. I think you being at the frontier between the physical and the digital is a bit of unusual conversation for us around the Metaverse, but I think it's very... That linkage is absolutely important and I hope people have enjoyed your perspective. So thank you very much, Carl, it's been a pleasure.

Carl Bass:

Okay, thank you, Patrick. Thank you, Marc. Pleasure talking to you guys.

Marc Petit:

Thank you to everybody out there. Keep on hitting us on social, let us know what you think about the podcast, who you want to hear from. And Patrick, thank you so much. And everyone, thank you very much. Bye.