Will Carmakers Switch Clay For Computers?

The 3D printing revolution has transformed a lot of industries, but according to [Insider Business] the car industry still uses clay modeling to make life-sized replicas of new cars. The video below shows a fascinating glimpse of the process of taking foam and clay and making it look like a real car. Unlike the old days, they do use a milling machine to do some rough work on the model, but there’s still a surprising amount of manual work involved. Some of the older film clips in the video show how hard it was to do before the CNC machines.

The cost of these models isn’t cheap. They claim that some of the models have cost $650,000 to create. We assume most of that is in salaries. Some models take four years to complete and a ton of clay.

A film over the clay makes the material look more like plastic or sheet metal. Modern model makers do use 3D printing to make pieces like headlights or logos. They also use 3D scanners to send the model’s shape to designers for feedback. Of course, if the designers make a change, the model must also change.

You’d think computer modeling would be sufficient, but apparently, there’s nothing like seeing the real thing. They also put the clay models in wind tunnels and use smoke wands to study air flow around the proposed vehicle shape. Maybe one day, they will use pure CAD rendering and VR goggles, but today, they are still paying clay modelers to make replicas.

Concept cars have been using 3D printing lately, too. Of course, if they ever do 3D print the entire car, they will have to assemble it from parts or get bigger printers.

47 thoughts on “Will Carmakers Switch Clay For Computers?

  1. Why do they even bother modeling them in clay, or any other medium? They’re all going to look the same anyway. Monotonous, stultifying, demoralizing uniformity. They aren’t legally permitted to build them any other way now. Every manufacturer, every model. Here’s your hybrid hatchback, and don’t worry they’ll take that away too.

    1. They all look the same because they all are optimized for the same drag physics. Also because the US market buys mostly small SUVs so that’s what manufacturers make.

      In other words you are complaining about government regulation when the real driver of car design is the laws of physics and capitalism. Ironic

      1. > Laws of capitalism

        I’m not so sure the consumers are driving the demand for small SUVs as much as it is the manufacturers leaving sedans and hatchbacks out in the dust

        1. And various historical legislation (not sure if it still applies) where “light trucks” were exempt from some kinds of emissions amd safety legislation, which often made them cheaper, or the design more flexible.

    2. There’s nothing in the law that defines how your car looks. You’re just conspiracy theorizing away here… this is all driven by very usual market forces. If you want to foster diversity in design, you might need to introduce anti-trust legislation and regulate the car market much more.

      1. Other than the laws of physics. Government regulations require a certain efficiency. The only way to achieve that is reducing the drag coefficient. There are only so many ways to reduce the drag coefficient of a box on wheels.

        Same goes for rockets and airplanes. There is a reason these groups are all generally similar.

        1. Drag coefficient for cars usually refers to aerodynamic drag divided by area. Making the car smaller reduces the area, thus reduces drag. Making the car smaller also makes it lighter, which reduces tire drag. Using lighter materials, which are more expensive, reduces tire drag. Low rolling resistance tires help, but there are compromises involved.

          More efficiency can also be achieved through sophisticated means like advanced engine technologies (too numerous to mention, mostly expensive), running under conditions that improve Carnot efficiency (superchargers, turbochargers, disabling unneeded cylinders, all expensive), and tuned free-flowing exhaust. Better transmissions also help, but there’s not a lot of room for improvement there.

          The alternators in cars used to be horribly inefficient. I guess they’ve improved; I don’t know. The move to LED headlights helps a little.

      2. yes it does…specifically the US mandates big bumpers. Then there’s the windshield, roof or roll bar, a given set of lights, the noise and emission regs need internal space, crash safety needs t h i c c A-pillars and more internal space for crumple zones and airbags, auto-brake and lane-keeping needs sensors in specific places…

      1. One off specials can be made street legal through loopholes, like not requiring airbags because the base model of the car you’re modifying did not have airbags when new.

        For new production cars, the requirement for various mandatory safety equipment, structures, fuel economy and emissions, there’s not a whole lot of room to make cars that look any different. They’re all wedge-shaped soap boxes with long sweeping A pillars and a short stubby back.

        1. I see the debate about drag coefficient and what not…and that doesn’t really matter. The truth is that in the US the trend has been towards bigger cars because of the “safety” feeling. No one wants to be on a mini when all the cars around you tower you and in a crash you would go under their engines. It’s that simple. Specially in suburban areas, where 1 in 100 that own a pickup ever uses it to carry any load.

    3. Having been a life-long Detroiter and an engineer (3rd gen BTW) in the automotive industry: looks may say one thing, but the feel make a huge difference. What is created on the computer may be visually appealing, but could lack when ‘touch’ becomes involved, and vice-versa. This is why full-scale modelling is part of the vehicle development process. If the customer doesn’t like the feel of the vehicle, it may not sell.

  2. I can see them using only VR amongst the engineers and designers, but when it’s time to show it off to the executives and investors, I don’t see physical models going away any time soon. I can totally see the clay go away in favor of large format 3d printing, though. Humans, especially artists, are expensive and notoriously finicky instruments.

    1. “This large scale print is covered in ugly layer lines. If only there were some kind of malleable sculptable substance we could cover it with to smooth everything out perfectly”

    1. One must realize through deduction that the events in 1984 can be no less then the third time around in that closed time-like curve.

      See, the first time 1984 happens, there could be no John Connor, as Reese didn’t exist in that “time” until “later”, when he could be sent back the first… time, without knowledge of John but subsequently getting it on with Sarah… for some reason, and generating John.

      The second time 1984 happens, there is still no John until Reese gets it on with Sarah so that John can deliberately select Reese for the mission.

      The third time through 1984, as depicted in the documentary, Kyle has knowledge of John, gets it on with Sarah, and ensures the continuum’s integrity.

      This is an absolute minimum. Who knows how many times around it was until Reese finally, through pure entropy, gets it on with Sarah, generating John who just happens to be the right person to lead humans against the machines in the future. Perhaps thousands of loops.

      Sarah was probably always a target because through the lore, it has been shown that Skynet has the ability or either simulate or communicate with other timelines, and if one iteration lost to John Connor, it is likely that other iterations may cover that possibility from ever reaching fruition.

  3. > there’s nothing like seeing the real thing

    There’s two things: scale illusion, and the fact that real-time rendering still doesn’t quite do photorealistic surfaces right. Since the VR rendering is optimizing to look good, not look exact, and the surface is only a texture with depth approximated by a bump-map, it does not accurately capture the real curves and contours of the shape.

    But the scale illusion is the worst. Looking at objects on a monitor, or even in VR, royally mess up human depth perception and as a result, sense of accurate scale. This is not something easily fixed, since the problem is with the optics, not the rendering.

    1. I’ll also say that I guess these designers are humans, and humans are actually kind of… handy with their hands? Clay is a very intuitive medium, compared to a lot of VR controllers, still.

      Want to add a carveout in the 3D model? Easy, press the circle-menu button on your left-hand controller, and move the controller to select the carve-out tool; regulate its size by tilting your hand until the circle is the right size; then apply the tool with your right hand to the model. If our system is really good, your glove will give you pressure feedback depending on how deeply you go into the material

      *grabs a medium-sized tool from belt and carves out a line in 0.75 s*

      1. Human hands are good at feeling slight imperfections and undulations in the surface shape, that would not be visible to the eye except when the light catches it just right, which may or may not be captured by the rendering in the first place.

        What’s the VR equivalent of sandpaper?

      2. Having played around in VR enough I’d suggest the difference will actually be very low with the right control setup and enough user practice.

        However the Clay medium has the major advantage that even if you have never worked it before you have used pens, pencils, paintbrushes, knives etc – its very very akin to a whole slew of skills you already know, not much learning to do before you can make a start. Where as nobody grows up in VR (yet anyway) you are stuck with a steeper learning curve and interfaces that have to make the choice between really speedy for the experienced user but nearly impossible to learn or a bit more ‘helpful’ but slower to use.

        1. Though as Dude rightly points out we can feel in the real world rather tiny imperfections, no such thing in VR (yet anyway). In many ways it won’t matter in VR as you can just tell the computer this is perfectly smooth etc. But it is worth bearing in mind.

    2. VR gives you enough depth information for that not to be a problem (assuming you are one of the folks with ‘normal’ enough eyesight for it). The problem with VR will be fidelity – even the best headsets don’t have enough pixels to really let you have a good simulation at all viewing distances.

      1. It’s about the fact that the VR goggles do not and cannot provide you with real accurate depth perception the way your eyes work directly. It is always a compromise.

        Sure, if you spent weeks inside the VR environment, you’d get the same sense but then you pull the goggles off and your world is distorted again. Everyone who wears fairly strong eyeglasses know this effect: when you get new frames and lenses, it takes a day before you stop feeling sea sick. For the first few hours, driving a car for example feels pretty harrowing.

        And for fidelity, aside for the resolution of the headset, a human can see down to microns. The surface texture of the model does not have information down to that level, so when you go up close to look at it, it loses detail.

        1. Depth in VR really doesn’t take long to get used to, and is really damn close to the real world experience if your one of the folks with sufficiently normal vision… First time you put it on and reach out with a motion capture controller you will find your sense of depth for the object you wanted to interact with was correct enough that it just works and feels pretty normal, practically every game in VR is leveraging that your body position and depth perception still match up well… The bit that is disorienting about VR for most folks is the eyes and inner ear disagreeing or bumping into something real you can’t see – one is mostly solved by having powerful enough hardware so the frame rate and motion tracking updates never dip too low and the other by designing the VR experience to keep you in whatever bounds are set…

          Also you are comparing to a clay model here – its not got the final finish level of detail anyway, its still just an impression of what will be. So you really don’t need to worry about the down to the microns level of detail. But VR as it stands now would struggle to render fairly normal sized text crisply enough to read easily – it is just not that sharp and smooth and generally sharpest only in some smaller section of the FOV, so reposition your head all the time relative to the virtual object. For the more macro object getting a sense of the car’s presence as you walk around it, or for how much of a reach the controls are from the ‘seat’ it will be absolutely fine, it will fall down on details like being able to easily tell which control does what unless the legend on them is larger than usual in real life.

    3. I think this is the only way the Pontiac Aztek was allowed to come into existence. It probably looked “cute” on a monitor. I’m guessing GM demanded a full scale clay model before green-lighting any new model after that.

  4. also at the lower reynould numbers that cars operate
    it is more or less impossible to do decent computer
    simulations of the airodynamics
    or as one famed airodynamisist proclamed”low reynould
    numbers should be banned”
    so full sized models,real air,and visual confermation
    of what is going to happen

  5. The article conflates producing a physical prototype with the means of producing it.

    There are few if any products that are done at scale which do not have a physical prototype produced at some point in the process. Consider it as the “final exam” for the design and production process, since you’ll find conflicts caused by things that looked good on a screen and added up in a spreadsheet but are untenable at full scale. This is especially critical where manufacturing systems and aftermarket maintenance are important – and horror stories about slip ups between design and production go back to the dawn of the industrial age and are unlikely to go away anytime soon.

    1. Some people are willing to take risks and do the prototype round when you have promised to start delivering actual products. Then you gotta learn and figure out solutions really fast. This is also the part where cash gets burned in large heaps. The cost of prototype is usually small fraction of it.

      When you design things long enough, whole team gets completely blind to obvious issues in the design.

      And simulation is always as good as assumptions used during building the simulation. It never is something actual in the real life but might be pretty accurate if all assumptions are selected correctly.

  6. While getting a life size accurate representation to see how it looks is an important aspect of this, it’s mainly for wind tunnel testing. Our understanding of fluid dynamics and chaotic systems has improved dramatically (and continues to) but the real thing quite often does not match the computer simulations and the full size often has different characteristics than the small scale models. I saw a case where the radio antenna (although very small) caused a huge amount of turbulence which was fixed my moving it a mere 3/4 of an inch. The computer simulations did not accurately model the turbulence from the antenna nor the dramatic effect of moving it slightly. (This tale from my uncle who was one of GM’s top engineers, recently retired)

    1. Hm. Anything that can be fixed in the wind tunnel by moving something 3/4 inch, is probably still in trouble in the real world, where pure laminar flow rarely exists. A slight crosswind, other cars, or flow under an overpass, and there’s your turbulence again.

  7. Sometimes, the clay model is faster to design with, its not a “print” of the design, but the actual working medium for the design. An artist and design in clay faster than a person on a computer. As a product designer I still sometimes start my designs on paper sketches as its so much faster than using solidworks, which I love and move to after I have a design I like. – Back to the clay, and artist can pull a bump/line up, down, remove walk by the car and “feel” the hight of the lines etc. much more real life.

  8. $650k ain’t nothing to sniff at, but if it takes 4 years to produce that’s “only” $162k/yr for one person to do the whole thing. I assume
    They have more than one skilled artisan and designer working on it. Hard to imagine two people producing it gettin foams $80k a year each. So I’d say this cost is cheap, cheap cheap especially in the co test of automobile marketing and also R&D budgets. Like, massively cheaper than even a relatively simplistic VR and computer assisted drafting system plus the entire team of skilled engineers to use it.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.