3d scanner

[Morten]

So in short how would you like the scanner to work knowing that there is no motor and there is 1 or more red line lasers that you can turn on and off?

I have no idea, but would like to say I approve of your approach. This attitude, and the original idea, is why I joined the funding effort.

[MemorianX]

If you want to do the rotation cheap and manual it would be quite easy to rig something up since we would have access to a 3D printer, you could just print a gear like structure as the platform and rotate it manually one step at a time maybe use a nail or something to stop it at a precise spot.

[quertz]

Hey guys,

considering you've got the printer setup and can capture around 30fps per sec, I would suggest to use the mirrors of the peachy and make one of them oscillate really fast (or use some other way to get the laser to draw a full line in each videoframe).

Then I would use the second mirror to slowly tilt the line to the left and right to get a distance value for a full 2D grid.
Then you can rotate the object by a step and repeat the process.

This would work similar to many other 3D scanners (eg. Kinect v1) which project a certain pattern on the object and use a camera at a certain distance to get the 3D information.
https://www.youtube.com/watch?v=fJpNHBfLfdU
This 3D animation shows how they use patterns to do the scan. Our "pattern" would be the time difference of the laser-line.



This would have the advantage that, as we will always get a full 3D-picture, the rotation steps can be a lot larger and easier to do manually.

greetings, quertz

[mike_biddell]

Hey guys,

considering you've got the printer setup and can capture around 30fps per sec, I would suggest to use the mirrors of the peachy and make one of them oscillate really fast (or use some other way to get the laser to draw a full line in each videoframe).

Then I would use the second mirror to slowly tilt the line to the left and right to get a distance value for a full 2D grid.
Then you can rotate the object by a step and repeat the process.

This would work similar to many other 3D scanners (eg. Kinect v1) which project a certain pattern on the object and use a camera at a certain distance to get the 3D information.
https://www.youtube.com/watch?v=fJpNHBfLfdU
This 3D animation shows how they use patterns to do the scan. Our "pattern" would be the time difference of the laser-line.


This would have the advantage that, as we will always get a full 3D-picture, the rotation steps can be a lot larger and easier to do manually.

greetings, quertz

Quertz..... a narrow mesh (generated by both mirrors oscillating), is an excellent idea. As a matter of passing interest, someone invented a bicycle lamp which projected a mesh onto the ground to identify potholes !!!!! A square mesh would probably suffice.

[quertz]

@mike_biddell

I seriously just read "to identify plotholes" and thought: that's not how this works

Actually I rewatched the video of the update and it seems to me that they are planning to do something like that.
I'm just not sure whether they are aware of the similar solutions that already exist.

I would definitely suggest making the angle between laser and camera even smaller. I think the optimum is somewhere around 20 - 30°.
You will get the resolution back by getting the whole 3D image and not just a single line.

Greetings,
quertz

[Bryce]

Not sure how much help it will be, but I've heard other 3d scanners highly recommend trying to coat the item to be scanned in a white powder like talcum powder to get uniform reflection and avoid the light absorbtion problem.

[rylangrayston]

Great Ideas everyone!
Iv tried using a very old version of the peachy to move the laser in a line quick enough to simulate a line laser lens.. its a great idea and it still has potential.
one of the problems with doing that is the that typical web cams dont give you control over the exposure time, and the exposure time adjusts automatically, so its tricky to
get expose a line that dosent overlap uneven brightness. The first scans shown in the main campaign video did use a peachy printer with one mirror vibrating very quickly. and as a temporary solution I used a DSLR camera which allowed me to control the exposure times and minimize this problem.
I like the idea of a scanner hack where a line laser ( with a line laser lense) is placed in the peachy printer head and one of the mirrors causes the plane of light coming from the laser to move across the model.

We also wrote software quite some time a go that takes N video frames and mixes them together to create a long exposure shot... for example filming at 30 frames per second, you can take 60 frames to get a 2 second exposure image. This could be useful when drawing a grid.

I Think the first version of the scanner will be simpler than these ideas but I can see how as scanner hack today could become the standard 2.0 scanner of tomorrow.
oh and something to stop it a precise spot... yes Ill see if I can work that in today, seems useful.

[mike_biddell]

Great Ideas everyone!
Iv tried using a very old version of the peachy to move the laser in a line quick enough to simulate a line laser lens.. its a great idea and it still has potential.
one of the problems with doing that is the that typical web cams dont give you control over the exposure time, and the exposure time adjusts automatically, so its tricky to
get expose a line that dosent overlap uneven brightness. The first scans shown in the main campaign video did use a peachy printer with one mirror vibrating very quickly. and as a temporary solution I used a DSLR camera which allowed me to control the exposure times and minimize this problem.
I like the idea of a scanner hack where a line laser ( with a line laser lense) is placed in the peachy printer head and one of the mirrors causes the plane of light coming from the laser to move across the model.

We also wrote software quite some time a go that takes N video frames and mixes them together to create a long exposure shot... for example filming at 30 frames per second, you can take 60 frames to get a 2 second exposure image. This could be useful when drawing a grid.

I Think the first version of the scanner will be simpler than these ideas but I can see how as scanner hack today could become the standard 2.0 scanner of tomorrow.
oh and something to stop it a precise spot... yes Ill see if I can work that in today, seems useful.

A bit of Googling..... structured light scanning looks to be the way to go, projecting a series of grids onto the object. Whether Peachy could achieve frame rates for persistence of vision (25-30 fps) I dont know. My cheap laser pointers come with a series of lenses, so a cheap 'grid' lens should be achievable, if Peachy cant achieve a persistence of vision grid.

The David scanner uses a projector to create the grid:- http://wiki.david-3d.com/david3_user...ructured_light

[mike_biddell]

errr this laser pen has a lens which produces an interesting pattern of squares:- http://www.ebay.co.uk/itm/New-Red-La...cAAOSwMmBV5qMF

http://www.ghoststop.com/Laser-Grid-...-greengrid.htm

Since the peachy can turn 4 line lasers on and off, this scanner does that and produces good quality scans..... the main scanner rig to hold the camera and line lasers could be printed by the Peachy itself. Using 4 lasers at different angles to the scanned object seems to give good quality scans. http://www.instructables.com/id/Build-a-30-laser/

[quertz]

I think you could get a really good quality using the peachy for structured light / laser scanning
https://en.wikipedia.org/wiki/Struct...ght_3D_scanner