Slashdot Mirror
Fuel3D Start-Up Promises Affordable Point-and-Shoot 3D Scanner
MojoKid writes "Fuel3D Inc. is a new start-up that recently arrived to Kickstarter, promising an 'affordable point-and-shoot 3D scanner' that will allow anyone to easily take 3D images for rendering and ultimately production on standard 3D printing platforms. The Fuel3D is a fully 3D surface scanner that samples a large number of physical and color measurements including geometric stereo and photometric stereo data, which it then combines to create the image. The kicker is that the device—which kind of resembles a Roomba--costs under $1,000, and it works just like a point-and-shoot camera. You simply attach a tag called a target to the person or object you want to scan and snap the picture. Then, you can work with the image and export it in a variety of formats."
I want a program where I can take a video camera(aka smart phone), and walk around inside a building, aiming the camera around the walls and all over. And the end result would be a 3d model of the building I was in. This would be great for making video games. And attach it to the Google Car: And we could play Need for Speed: Cannon Ball run.
God spoke to me
3D? That's terrorist talk!
I love all this fancy new 3D technology thats coming out.
If only the real world was in 3D, that would just be awesome!
How is it better than a Kinect and open-source software you can download for free?
The multiple-flash light source thing is a good idea. Someone did that about seven years ago with a modified Canon PowerShot. They had four flash units, one at each corner of the case, and it would take four pictures, one with each flash. The paper shows the test setup, but later they built a camera with the four flash units built in.
This Kickstarter project doesn't really produce a 3D model by itself. You need an external stitching program to put multiple images together into a model, they say.
Programs for building a model from multiple 3D images have been around for 15 years or so. Much of the early work was done at U.C. Berkeley. A Stanford group did a lot of high-res work, including imaging parts of the Vatican. Autodesk 123Catch, can build a 3D model from a collection of images without a special camera.
The 3D resolution is a lot better, just see the kickstarter page.
http://www.kickstarter.com/projects/45699157/fuel3d-a-handheld-3d-scanner-for-less-than-1000?ref=live
Also it uses a proper depth sensor that doesn't need to be manually calibrated. (Last I read Kinect has to be manually calibrated for depth each time it is used)
Yes you can create colour 3D models with a Kinect, but this is designed to that a snapshot that is usable for high quality 3D prints, use for commercial 3D, or scaling down to game polygon count.
From what I'm reading the Kinect is kind of like your camera phone. Not bad, but not particularly good either. The Fuel3D is more like a DSLR.
Microsoft, Apple, Google, Amazon what's the difference? All steal money from devs and control with walled gardens.
You simply attach a tag called a target to the person or object you want to scan and snap the picture.
So then I have a 3D model of my object - with a tag on it? I can't see anything about tagging at the official site or the Kickstarter page.
Here's an actual link to the company. I'd have thought TFS would have one, let alone TFA, but no.
systemd is Roko's Basilisk.
How are people to scan their cocks with this thing? After all that's what most people would want a 3D scanner for. It needs to come some kind of stand (with removable gloryhole) that the scanner rotates around.
Now we can have more realistic dildos than ever!
> When the folks behind Fuel3D say they have a 3D camera scanner, they’re not talking about a simple stereoscopic device; rather the Fuel3D is a “fully 3D surface scanner consisting of a large number of physical and color measurements” including geometric stereo and photometric stereo data, which it then combines to create the image.
Plain Wrong. As far as the kickstarter page tells, it is only stereoscopic cameras with stereo algorithms to calculate 3D object data.
> In contrast, the core technology behind Fuel3D fuses geometric and photometric stereo 3D recovery techniques and is finely tuned to capture high resolution 3D color images. This raw capability provides the opportunity for an object to be captured in true 3D geometry and full color.
Almost like video trace (http://www.youtube.com/watch?v=z4obTZV2QEM) but without manual postprocessing.
btw. the kickstarter is already funded:
http://www.kickstarter.com/projects/45699157/fuel3d-a-handheld-3d-scanner-for-less-than-1000
Pixel depth info from focus stack:
http://www.stareat.it/sp.aspx?g=3ce7bc36fb334b8d85e6900b0bdf11c3
Combined with high resolution Lytro type camera, that would be quite interesting.
In fact the scanner combines several known technologies in a unique way that has never been commerically offered before - it is not just using photogrammetry. It also has simplified 'scanning' into a point and shoot interface (unlike the arm waving required for other handheld scanners) - then add in the price point and double rez compared to the nearest scanners like the Creaform Go and this is a new product category. Full disclosure - I am one of the folks working with the kickstarter campain.