moire digitizing?

[penguin]

My father, a chiropractor, used to have a moire camera that could take photos so he could tell if the back was symmetrical and aligned. It looked like a contour map and was read the same way, each contour was a different elevation and the closer the lines the steeper the grade.

Would it be possible to use something like that to digitize a face, a full-sized car or even a building for modeling? Seems something like that would give you portability and a way to get subjects you can can't touch like statues and artwork digitized.

I'd appreciate any ideas or suggestions on it.

[unterhaus]

I've seen people do that in engineering, I forget the actual application. It's very simialar to "structured light" approaches. We actually have a Zygo profilometer that works along these principles, except it generates interferance patterns using moving optics in a microscope.

[penguin]

Originally Posted by unterhaus I've seen people do that in engineering, I forget the actual application. It's very simialar to "structured light" approaches. We actually have a Zygo profilometer that works along these principles, except it generates interferance patterns using moving optics in a microscope.

Any idea on what kind of resolution that has? Would it work on large objects? I did google 'moire' turns out most photographers are dead set against it, tons of info on how to avoid it but precious little on how to do it.

[unterhaus]

The guys that were doing it actually printed something on transparency paper and then projected it. It seems like you see it most often in undersampled pictures. I would think that if they have ways of avoiding it, that might suggest ways of achieving it?

I think it would work fine on large objects, depends on what resolution you want. The resolution will probably be poor relative to machine accuracy.

[penguin]

I finally found the correct terms to google, made a world of difference. If you're interested look in google patents under 'moire contour', 'moire topography' or 'moire effect'.

There's at least one patent for body scans for garment makers. Fuji(?) had one in the early 80's for making 3d models of buildings using moire so there's been some work in the field.

I've also found some sites that used something similar to measure breast volume, one was a medical study to measure milk synthesis, the other was to make better fitting brassieres.

Seems to me the limiting factor to resolution would be the wavelength of the light source and the diffraction grating used. I'd love to be able to go downtown and get scans off some art deco on the buildings. If they can measure milk production in milliliters like this then surely it could be made to work for my purposes?

[braidmeister]

Minolta used to make a 3D camera that used structured light (line patterns) to create a 3D file. It was co-developed with a company called MetaCreations...It seems the technology was bought out and either shelved or incorporated into something else.

InSpeck uses moire or structured light with their 3D cameras. I almost bought one last year, but after testing it for 2 solid days, I found that it was fairly good for doing faces and things like that, but really crappy at doing things like relief carvings and the like...So I didn't buy it.

-Brady

[unterhaus]

I actually have a couple of Minolta devices on a shelf here in the lab. They use a laser scanner and a camera. It worked pretty well, you'd have to be highly trained to do a good job with it though. The main problem we had was it couldn't handle tires, and we wanted to digitize our robots -- the side view is mostly tire. They have a scsi interface, and use a dongle. Never really understood what you were going to do with the software if you didn't have the $40k box attached to your computer...

[braidmeister]

Originally Posted by unterhaus ...Never really understood what you were going to do with the software if you didn't have the $40k box attached to your computer...

Exactly!

-Brady

[penguin]

So much for cheap and easy, any ideas on how you could get architectural details from buildings? Right about now the only methods coming to mind is either get permission to hang out of a 4th story window or pay a window washer. Since I hate heights only one of those would be feasible for me.

[unterhaus]

if you are digitizing a building you can use cameras from multiple views. There used to be a company that sold software to do that and there was a freeware version.


If you need accuracy, there are 3d laser scanners. Most of the architectural scale scanners are expensive enough that you would have to pay someone to do it. Structured light is always an option.

If it is something the size of a single family residence, you can use a Sick LMS200 laser scanner and rotate it. Still talking about at least 5k and a considerable amount of software.

In the U.S., paranoia has gripped our country to the extent that taking pictures of buildings has become a highly suspicious activity. You should keep that in mind if it's not your building.

[penguin]

I subscribe to the belief that just because you're paranoid doesn't mean someone's not out to get you, too. Unfortunately that does complicate what I want but I figured to cross that bridge when I got to it.

The buildings I like are the old buildings from the 20's when the town was flush with oil money. The detailing was incredible, nothing like the crap that passes for art or architecture today.

I just thought it would be nice to be able to make some of those things when I finally get my router built. I reckon I'll have to fish on it and hope something comes to mind.

[braidmeister]

Penguin,
It might be right up your alley to check out Google Sketchup. I saw one of their tutorial videos the other day and it showed them basically overlaying a picture of a vintage building with 3D elements ~ a kind of 3D version of tracing out an image with 2D vectors. It's probably worth a look.

-Brady