Slashdot Mirror
Elon Musk Shows His Vision of Holographic Design Technology
MojoKid writes "Tesla CEO Elon Musk has more on his mind that just cars and 4,000 MPH Hyperloop transportation systems. He also tweeted his intention of developing a hand-manipulated holographic design engine and designing a rocket part with only hand gestures, finally printing the part in titanium." And now Musk has posted the video he promised showing off the design process: "Musk showed a wireframe of the rocket part, and he was able to rotate the 3D object on a screen with one hand, and with a second hand, he zoomed in and out, moved it around the screen, and spun the object around and "caught it"--all in the air. He moved on to manipulating an actual 3D CAD model and interacting with the software; you can see that he used a Leap Motion controller. Next, he shows off a 3D projection, a freestanding glass projection (Ironman style), and interacts with the model using the Oculus Rift. Finally, he prints the part in titanium with a 3D metal printer. Note that we don't actually see him design anything; the models he works with are already made. Still, it's exciting to see new ways of doing things come to life on screen."
This video is some great gee-whiz showmanship, which gets Joe Sixpack excited about technology and maybe sells a few more Tesla cars a couple years down the road.
But man, I'd hate to try and design anything by holding my hands in the air for six or seven hours a day.
Might be cool for next-generation Kinect-style gaming though.
3D Printing Tips and Tricks at Zheng3.com
In a nutshell, he shows a gesture-operated wireframe or shaded model viewer. All that was shown was zoom, rotation and moving the clipping plane. It must be impressive for the non-CAD crowd, but I didn't see anything new that was practical, and there are far better viewers already available.
Elon Musk tweeted "Will post video next week of designing a rocket part with hand gestures & then immediately printing it in titanium". But the video says he designed the parts in Unigraphics, so it was NOT designed using hand gestures - unless you count using a mouse as "hand gestures".
Anyone that has used a solids modeling CAD application with a Space Controller in the last 17 years has been able to do pretty much all of this and more (not counting the use of Ti in the printer). But Space Controller + mouse users keep their arms on the desk, rather than waving their arms in the air (fatiguing). So thanks for thinking of us, be we designers do NOT want to hold our hands in the air. We had enough of that crap with light pens.
In all fairness, the one nice thing the video suggests is the gesture-operated viewer might make sense for a standing presenter. The Space Controller requires a flat surface to rest upon, so gesture-operation might make sense for this application.
Place nail here >+
This kind of thing has been done for years/decades, and is used a lot in industry already. I know of many companies that are taking their CATIA or NX models, and visualizing them in VR or CAVE systems. You can even buy software which has been written to solve this problem for you ... you just buy the VR hardware, load your model in, and you can manipulate it. For example, here is a video of something that you can go and buy right now ... http://www.worldviz.com/industries/cad ... the sample video uses a stereo projector and a wand, but you could easily use Leap Motion, a full motion capture system, or whatever hardware you have.
The gestures he shows in the video are also very primitive, and did not appear very intuitive. He spins his hand in one direction, and the model rotates about a different axis. Just spinning a model around is not really designing anything, it is just a preview technique. And in some ways, using a mouse could probably be less fatiguing and more intuitive. The real holy grail is addressing the problem of creating models from scratch. And making them accurately so the designer gets what they want, and typically designers will probably want to use a keyboard so they can enter in exact dimensions. But this is not addressed in the video, but I'm sure everyone will think that is what they actually can do.
But because Elon Musk is doing this and now you are seeing it in the popular media, he must have invented it, right?
If you want to see something really cool, check out this system, the Lockheed SAIL, which was done years and years ago ...
http://www.motionanalysis.com/html/temp/lockheedHIL.html
I think it's great. Somebody with capital and grand long term visions.
Even when you have money and a long term vision, some ideas are just dumb. Like the Segway.
This is one of them.
There's a reason why you only see interfaces like this in the movies, and it's not because the technology isn't there. It's because waving your hands in the air for a whole workday sucks. Someone in this thread mentioned light pens. When you no longer had to hold your arm up to the screen because of the popularity of the mouse, light pens disappeared from everyone's desks.
This interface sucks even in concept.
--
BMO
It's not and people don't typically use a mouse for CAD. Sure a mouse is used, but most of the actual work is done via keyboard shortcuts because it's both more accurate and faster.
So by "people don't typically use a mouse for CAD" you mean people do use a mouse. Uh.... OK.
Sorry, but the mouse is heavily used for CAD. The hot setup is a Space Pilot Pro and a multi-button mouse. Greatly reduces the use of a keyboard.
Place nail here >+
The hard part is the real science and engineering behind the part. The deep understanding of the material characteristics that are the result of the the metal and the manufacturing process. The lifetime of the part and how it's affected by it's use history. The stress and vibration the part endures. What safety margins you need. And how to get all that in a cost effective way.
Making a 3D image is not engineering the part. It's just a drawing from the most trivial aspect - the dimensions of the part.
I'm a fairly regular (though not extremely skilled) user of 3D CAD software. I suspect that this would extend the time to perform simple tasks when compared with a 2D & a 3D (space) mouse.
The fact that Elon Musk doesn't design a part in his demo is telling. That part is VERY simple to create in today's UIs: a simple revolve of a cross section with a couple of patterned extrusions around the circumference. I see 5 distinct operations, and a more experienced person could probably make it in less. Just a couple minutes for even an amateur like me.
I think other 3D users will agree with me that this will increase the time for common tasks like selecting edges for radius/chamfer/draft and the critical sketching/dimensioning of cross sections for extrusion. A mouse pointer is so precise. To get the same level of accuracy with this system, you'd have to zoom in several times to make sure the correct feature/surface/edge is selected before you can perform any operation.
Creating the 2D drawings, which are still required for actual production, will also take much longer with a system like this.
There are many analogies for non-3D users, such as art programs or scale model/figure painting. I bet this is very similar to a programmer watching Johnny Mnemonic or Swordfish and saying "yeah, hacking doesn't work like that, that's ridiculous".
I've pre-ordered a Myo Armband and I'm hopeful that I can make it do some cool things with my CAD station. But for now I don't see this as anything more than a way to show your Director or VP the cool work you've been doing or communicate issues/challenges in the mechanical design to non-MechE's.
Mod parent up.
Look at the video around 1:54, where Musk is saying "go in there and do what you need to do". But all the video actually shows is someone spinning the model around and using a visual cutting plane to display cross sections. At no point in that video is new geometry created. What I was expecting to see was a breakthrough in how to do engineering design in 3D. It's not there.
Back in the late 1980s, Autodesk built a virtual reality system as an experiment in CAD. They got about as far as Musk, although at lower resolution - you could look at models and manipulate them with a gloves-and-goggles interface, but trying to draw surfaces in free space was really hard. Some people can do it. They can also do clay models freehand. Today, there's Autodesk Mudbox, a 3D sculpting tool which is used by pros who can visualize clearly and in detail in 3D. Watch this video to see one at work. That's impressive work. Now see something similar done with 3D input devices. It's like trying to sculpt while wearing oven mittens, and the results are awful.
Somebody will eventually make this work, but the computer, not the human, will be doing most of the design.
If you watch the video, it's easy to see why nobody demoed designing an actual part—the latency of a Leap Motion is just horrible. So bad that if you don't move your hands very very slowly, it will completely miss the motion.
We really really need a new generation of input devices. Not just to deal with things like the Oculus Rift, but because we're trying to do more things in less time. The mouse is great, but people forget that the only reason the mouse is so great is because it has what is effectively dedicated circuit paths right into the CPU and GPU, including GPU circuits created for the sole purpose of displaying its cursor. The first mice weren't nearly so responsive. Their designers eventually came to the conclusion that the system needed hardware support for a mouse pointer, with little or no software stack at all. In consequence, a mouse has 4 milliseconds of latency. Leap Motion (and everything else) has 250 times that latency. Nothing will compete with the mouse until it can match mouse latency, and that's not going to happen without more cooperation from the rest of the system.
Why do movies like Minority Report and Ironman look so good to us? Because the actor was filmed waving his hands around in empty space, then the computer displays were added in after the fact by special effect artists. And guess which frame the special effect artists painted the reaction in? The exact same frame the motion happened in. So it looks fantastic. Instantly responsive. Unless and until these interfaces actually are that responsive, they will get nowhere.
Then start marketing light pens. Make a new kind that can be used on LCDs.
See how far that gets you.
Here's a protip: a lot of the hate pointed at Windows 8 involves a touch interface on a desktop operating system.
It's not just me. It's over 40 years of UI research.
--
BMO
What I mean is that people use the mouse to scroll around, and not for the rest of it.
Uh... no. We use the mouse to draw the sketch before we apply the the data-driven parametric dimensions.
And, I don't care how many buttons your mouse has, it does not have more than a keyboard. What's more, only an idiot uses a mouse as a mouse just does not have the level of precision necessary for CAD.
Yes, a keyboard does have more keys, but the 80/20 rule applies. And saying people are stupid for using a mouse tells me you lack the understanding of how a CAD system works.
There probably are novices out there that never bothered to learn the short cuts, but anybody that cares about efficiency or precision is going to be using the keyboard almost exclusively.
I've been doing CAD since 1980. I wrote a CAD application. I've been and AutoCAD and SolidWorks instructor, and run user groups. If you think a mouse if not used, or unsuitable for CAD, the you are either a troll are greatly in need of proper instruction.
Place nail here >+