3d scanner

[ivin]

I've daydreamed about glasses free 3D tv's and wonder if my day dreaming might relate to this somehow.


Since 3d TV with glasses often requires flickering filters in the wearers lenses to show/hide the left or right frame. I thought what if your TV was able to shake left and right and only displayed the left or right frame (or any in between). You might achieve a good sense of depth.


so maybe there is a way to either shake the camera or model on an axis as it rotated to "see" better into the nooks and crannies. and that way you're not just scanning the same section over and over again trying to get higher resolution, you can attack it from more sides. the system would have to know the exact location of everything to account for the shake, thankfully thats not my problem .

[mike_biddell]

I've daydreamed about glasses free 3D tv's and wonder if my day dreaming might relate to this somehow.


Since 3d TV with glasses often requires flickering filters in the wearers lenses to show/hide the left or right frame. I thought what if your TV was able to shake left and right and only displayed the left or right frame (or any in between). You might achieve a good sense of depth.


so maybe there is a way to either shake the camera or model on an axis as it rotated to "see" better into the nooks and crannies. and that way you're not just scanning the same section over and over again trying to get higher resolution, you can attack it from more sides. the system would have to know the exact location of everything to account for the shake, thankfully thats not my problem .

Ivin

Your dream makes sense, the difference between left eye and right eye view or two separate and distinct view is called parallax, which gives depth perception. This is used as a 3d scanning technique.

[quertz]

Hey guys,

I've actually already tried that.
There is free software out there which you can use to turn a lot of pictures into a 3D-model, but it doesn't work that good.

Peachy would hopefully be great for small objects like those you probably want to replicate using peachy

[Feign]

On an entirely different note, I had an idea while playing with the 3D scanner I already have. Rather than positioning multiple cameras or moving one camera to multiple positions, have a set of large mirrors placed parallel to each other at opposite ends of the scanning area. Orient the laser line to project exactly perpendicular to the mirrors, so that the laser and all of its reflections form one line across the object to be scanned. Then orient the camera to view the line at on a 45 degree angle from the plane of the laser line. Place a visible black line along the bottom edge of the mirrors so that the software can calculate where each line segment ends and have the software overlap and flip the segments as needed to get possibly dozens of similar but different angles of the exact same laser line to fill in gaps that would normally happen from overlapping segments.

Then go completely crazy, have four mirrors oriented in a square box, with two differently colored lasers and the camera at an isometric angle to the box, to get hundreds of differing angles on the same part with each line of each laser.

[harpo99999]

feign, it sounds good except for one fact, the object to be scanned BLOCKS the laser beam path, so the laser does not hit the mirrors on the other side of the object, but yes offseting the camera in both the x & z directions while still pointing at the centre of the scan space should help scanning the objects, but it would probably need two cameras(one on the plane of the laser(s), and one above the laser plane

[Feign]

I guess I didn't clarify, the lasers should be oriented above the scanning, not even with it...
But then, I'm picturing a very wide fanning laser line. I don't know if a basic lens could really get the kind of spread I'm picturing here.

[rylangrayston]

Hey Feign
Have you built some of this?.... it seems to be a new way to scanning that I havent seen documented anywhere.
Ive read you post quite a few times and I think Iv got it modled right in my head now.... hold up this would take just a few min in blender, im just going to go do it...
Screenshot.jpg

ok so something like that ... where the yellow box is the peachy printer and so using the mirrors you can hit the object from ALOT of angles.
and obliviously the laser wouldn't keep going threw the object being scanned like it dose in this screen shot.

It seems to me that its possible that you don't need to rotate the object smoothly, instead you could just re position it 3 or more times.

[harpo99999]

rylan, that is an interesting concept, along with feign's ideas, perhaps a mirror inside cube with the peachy generating several scan lines (and moving them) perhaps with the cameras in the upper corners of the cube, but the issues with this varient are significantly higher processing needs from the several cameras, but the benefits are reduced time to scan, with the need for few(less than 5) bursts of scan and then stitching the scans together along with the resolution of the camera ie higher res is approx higher detail/accuracy of the model

[Feign]

That diagram has a lot of the concept right, though in this case, the Peachy can theoretically get in its own way... However, looking at it diagrammed out, I see that the only way it can actually hit itself with the laser is if it misses the object many, many times. Also, in that configuration, the sweep of the laser has to be precisely aligned in both axes to form a coherent line. If the laser is above the box, then it can scan across the part, only worrying about being perpendicular to one pair of mirrors at a time.

I haven't built anything out yet, though I've done a little bit of work with crazy "infinite mirror" boxes before.

[mike_biddell]

That diagram has a lot of the concept right, though in this case, the Peachy can theoretically get in its own way... However, looking at it diagrammed out, I see that the only way it can actually hit itself with the laser is if it misses the object many, many times. Also, in that configuration, the sweep of the laser has to be precisely aligned in both axes to form a coherent line. If the laser is above the box, then it can scan across the part, only worrying about being perpendicular to one pair of mirrors at a time.

I haven't built anything out yet, though I've done a little bit of work with crazy "infinite mirror" boxes before.

I've just ordered the bits for the sardauscan. All the .stl files are there, so printing these using Peachy should be relatively easy. It's a ready made 4 laser solution for about £12, as I already have an HD webcam. Not too shabby. That's not to say I'm not looking forward to the Peachy solution.... because I am !!!!

https://www.youtube.com/watch?v=TPb-gEVkBYo