View Full Version : Home-build 3d scanner

11 November 2004, 11:09 AM
Hi all.
I just wonna show you my friend's and mine project we are working on.
The PR-SCAN lol
It's 3d scanner based on linear laser, but because mine laser will arrive only in 10 days,
we use max to set-up scene and get simulated pictures.
You can find more on
Any comments are welcome


Rabid pitbull
11 November 2004, 05:14 PM
Pretty impressive! What will the new laser add to the system?? So far looks great for a home built system, keep on tweaking it.:thumbsup:

11 November 2004, 07:04 PM
Now if you could only build a 3d printer to compliment it.... :D

11 November 2004, 03:33 AM
I'm not gonna ask where you aquired a brain for scanning :twisted:

11 November 2004, 05:31 PM
is there more info on how we could do this, that looks GREAT

11 November 2004, 09:01 PM
Basicly all you need are this four thing:

1. web cam or better yet digital camera
2. linear laser to project the red line on your object
3. rotating table with some kind of degree mark
and finaly
4. software, which will calculate all data from pictures and create a 3d point cloud

very good explanation is here:

try googling for "diy 3d scanner" for more


11 November 2004, 07:36 AM

We upgraded vertex optimization.
Little analyse on webpage.
Next on list are edges.


11 November 2004, 09:03 AM
That's really impressive! Those shots are quite a bit better than the last ones! Keep us updated with your progress! GJ!:)

11 November 2004, 09:20 AM
Quite a welcome alternative to the $20,000 commercial hardware. :)
Curious though... What happens with your script if there's a break in the model's geometry?
Like say you had a can with a 1inch hole in the side. Have you coded it to take into account that break in the laser line?

11 November 2004, 12:31 PM
:thumbsup: RC1 DEMO released :thumbsup:
Try it

restricted on 20 samples

11 November 2004, 08:19 PM
Quite a welcome alternative to the $20,000 commercial hardware. :)
Curious though... What happens with your script if there's a break in the model's geometry?
Like say you had a can with a 1inch hole in the side. Have you coded it to take into account that break in the laser line?
Some of their earlier images featured a bottle of fabric softener( or something like it) which had a handle on it. The hole was shown just fine, although their more recent render of the fire hydrant looks quite a bit better.

11 November 2004, 12:02 AM
I'm not so sure that they would work the same way. With a handle, if you put all of your samples together (provided you've taken enough), you can still make out the geometry within the hole. In my can example, I'm just talking about a hole in one side of the can, so you're not going to be getting -any- geometry within the hole. So I'm just curious to know if the software would interpret that properly to create a hole in the mesh as well.
If you look through some of the laser scans for the fire hydrant, you can clearly make out breaks in the geometry - where the laser line went flat against an extrusion. I'm guessing that their software is coded to ignore these breaks, and stitch the lines together at the closest points to these breaks in the lines. In the example I'm suggesting, I'm just wondering if it wouldn't do the same thing there, since it will not be reading any actual geometry within the hole in any of the samples. Would those gaps in the laser be stitched, or not?
Not really a matter of the program's functionality (honestly, how often is someone going to cut a hole in a can and try to scan it? :p ), just curiosity.

11 November 2004, 08:10 PM
I just wanna add that this application requires Microsoft .NET 1.1 framework


11 November 2004, 09:56 PM
I got my laser.
The revenge of hedgehog (


11 November 2004, 12:32 PM
Hi, me again.
With a pumpkin scan

11 November 2004, 07:05 PM
Don't Linear Laser use lights to reflect and judge all of what the laser reflects off of? If so will the laser only be useful in low light situations, so the lasers detemination isn't altered by the light?

11 November 2004, 01:58 PM
Upgraded to FLAT scan and poly output

@NinjaSquirel: Wrong, laser do emit light, so the darker the better.
PS: all mine scans were done in extremly bad condition :thumbsup:


12 December 2004, 07:02 PM
Very curious.
Over 1000 views and no replys.


PS: Ver.1 will be available before 2005

12 December 2004, 07:00 PM
On your site you said you have a pc controlled rotating turntable. Do you have any pictures, how did you build it? Have you written the software to control it? Have you thought about how to control the cam from the pc too? Sorry about all the questions. I'm building a laser scanner for a class here at my university, and you seem to have alot of great ideas. Get back to me and lets see if we can get some more buzz started about this subject.

12 December 2004, 01:28 AM
This is awesome.
Scan some more things. Looks great.

12 December 2004, 04:26 PM
i control the digital camera (Nikon coolpix 4300) with Snappix ( via serial cable.
my linear laser is powered by USB connector, it only need 3 volts.
my rottating platform will be (hopefully) done this year and will use another com port to control.
I plan to use a simple stepper motor (like in floppy drives 5.25) and control it with uIC.

Why I use so many ports?
So that you can use what you have and not to be locked on mine gadgets.

raca (

12 December 2004, 07:32 AM
now available


12 December 2004, 11:31 PM
This is sweet. even if you just use it for refrence. Great job, and good luck to you.

08 August 2005, 05:58 PM
Hi all
I have question to RACA. I have created my own 3d scanner, and Iam intrested what is the output of your application. Is it just single points or does you create mesh?
If yes can you tell me what kind of allgorith do you use to create such a mesh? In my application I simply create curve for every frame and after whole scan, rebuild curves and loft them. This method is prety good for simple objects, but later on I would like to use more then one laser to create particular object and after that this method i completly unusefull.
Whole aplication is written in MEL, it is quite slow but it works :). For example 36 spams sphere is created in 6000 seconds...

08 August 2005, 08:00 PM
Hi Bartolomiejk,

Good luck with your scan-project.
About my workflow, I sample every horizontal line for most red pixel, then i write the position and the index for each and every point. So at the end i get a poit cloud. But from cloud to a mesh is realy a small step so i conect the closest points together. My raw file is here ( for you to check.


08 August 2005, 08:17 PM
Holy damn.. That's nice... Can I have the recipe for this? :p

08 August 2005, 08:56 PM

I think this step is not as small as you said, I even think that it can be more complex then scanning. Connecting closest points is not the best way to create mesh, for some objects your scan can be a disater using this method. I thinking about an algorithm which will choose the 'most important' points of scan, delete unusefull points, then try somehow to connect them (maybe using some index array which could be created during scannig) and create cleanest mesh as posible.

CGTalk Moderation
08 August 2005, 08:56 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.