Dear DAVID Community,

As you are probably aware, the entire DAVID company, team and product range has been integrated into HP Inc.
This forum is now read-only. Relevant content has been migrated to HP and merged into HP's Support Forums on November 1, 2016.

To start new discussions for 3D Scanning, please register and post your new topic at the HP Support Forums

How to calculate maximum precision with a setup

Which camera works? Where can you buy an adequate laser? And more...

How to calculate maximum precision with a setup

Postby jantje » Tue Mar 03, 2009 12:23 am

Hi

I've been going through the forum but don't seem to get my head around the impact of the width of the laser and the setup on the final scan resolution.
My case is: I know my laser is bad (2mm wide) but is that the reason for my bad scan results? In other, do I need to buy a new laser to get better results or is something else wrong? And if I need to buy a laser what should be the width of the beam (if at all of any importance) given the precision I want.

In short this is the loop that has been running through my brain.
Logically thinking I would say as follows: If my laser is 2mm wide then David can not position the laser with a precision higher than 2mm. Taking into account that you measure the reflection 2mm is an optimistic measurement.
But David is smart and takes the middle but I don't see how that then relates to the quality of the scan.
On the other hand if a detail is smaller than the width of my laser how can it not be invisible in the laser line of the camera?

As I have understood from other discussions it is no use having a laserwidth way below pixel height ( I'm focusing on the height of the scan here) as pixel height is maximum precision anyway.
So I did some calculation on the maximum precision without taking the laser beam into account.
Reasoning is as follows:
The scale is the size in mm between 2 dots. There are 4 of these dots so the height the camera must capture is 4 times the scale. The theoretical minimum height of a pixel is 4 times the scale divided by the camera resolution height. (This is not taking into account the fact that the setup is not flat)
Using this formula I get the figures as below
scantabe.GIF


Which means that in the advised 640*480 the theoretical vertical resolution is 0.83. With this resolution you should be able to see the wrinkles of older people ;-) on a face scan. Which I do not see (I'm a techy not a beauty consultant).

To summarize: My questions are:
Are the setup versus camera calculations realistic?
Is the width of the laser beam the maximum resolution? If not; is there a relation between the 2.
How can I calculate the maximum/realistic precision given a setup?

Please help me…..
Jan

PS
I'm using
Camera: Logitech QuickCam Pro 9000
Laser: laserline supercross laser 2 (home construction)
Computer: Lenovo T61 with 2 gig
Model: body cast face that can stand still for a while ;-).
User avatar
jantje
 
Posts: 107
Joined: Tue Mar 03, 2009 12:08 am
Location: Belgium

Re: How to calculate maximum precision with a setup

Postby MagWeb » Tue Mar 03, 2009 12:34 pm

Welcome jantje!

jantje wrote:Which means that in the advised 640*480 the theoretical vertical resolution is 0.83. With this resolution you should be able to see the wrinkles of older people ;-) on a face scan. Which I do not see (I'm a techy not a beauty consultant).


Your numbers are right. But why you cann´t get the wrinkles on a small resolution:

I did a rendered simulation of a laserbeam (same width) hitting the same simple object at different intersection angles (45°, 30°, 15°).
The horizontal lines represent some pixelrows.
You see, that the last example with 15° touches only one pixelrow. you would see only a straight line in the cam.>>> means, the less profil an intersectionangle and/or the objects surface create, the more important will be the resolution , for if you double the number of lines, some profile would be detected also in the last case. A wrinkle only gives a small profil compared to the whole face (if you scan it in total). So it might happen, that the 2D picture of the cam doesn`t see it at all. So either you increase the intersection angle, or the resolution, or you wait till the wrinkles get deeper :wink:

Gunter
Attachments
intersect.jpg
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Re: How to calculate maximum precision with a setup

Postby jantje » Tue Mar 03, 2009 9:24 pm

Gunter
Thanks for the feedback. Your explenation feels like sound and correct to me. But as I already stated I don't seem to get my head around it.
What I think to understand from your reply is:
If you want more detail you need a big intersection angle. (And this is more important than the laser width?? :? )
The intersection angle is the angle between the laser and the surface.

But as the surface is not straight (otherwise I would draw it in a 3D tool) there is no "real" way to control the intersection angle, is there?

So I don't get my head around how this helps me finding the maximum precision for my setup.

Jan

PS With no "real" way I mean that the scanning object is the way it is. You may find better directions to scan it but that is practice not theory.
User avatar
jantje
 
Posts: 107
Joined: Tue Mar 03, 2009 12:08 am
Location: Belgium

Re: How to calculate maximum precision with a setup

Postby MagWeb » Tue Mar 03, 2009 11:46 pm

Hi

intersection angle = angle between: laser--surface and surface -- cam...>>> different for each point on the surface

importance? depends on the situation -- only spent some time to give a hint -- doesn´t matter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Re: How to calculate maximum precision with a setup

Postby jantje » Wed Mar 04, 2009 6:36 pm

Gunter
Please don't get me wrong, I may sound aggressive but it is pure despair not anger. I do appreciate your feedback. I'm pleased with the information you provided. I'm pleased with your responsiveness. I'm pleased that you have confirmed that my figures on the screen resolution are ok. But as I stated before I don't seem to get my head around the problem I currently have at hand. Reading through the forum I notice people having the same problem so I'm asking to bear with me while I fully explain where I'm coming from.

Where I come from.
Being a newbie -in 3D scanning, cameras, lasers, step motors, controllers and PC connections- it is hard to make simple decisions that experience users make without realizing they made a decision. This is my story:
One fine day I find the DAVID scanner web page. I think cool great I wanted to do such things already a long time. I look around in the house; find a laser and a web cam and quickly create a setup with the free version. I'm happy with the first results (taking into account these are the first scans so precision should increase over time with skill) so I buy the soft.
In the end I'm not pleased by the increase of the scan precision . I do a better setup, a bigger setup, a higher camera resolution, buy a good camera but for none of the setups I'm happy with the scan outcome. It is the precision that is not as good as I want it.
So I read and read and change the setting of the camera. Results improve but not enough to satisfy me. So I look at buying a better camera (+100 euro) or a new laser (+100 euro) or a laser mover (+100 euro as I want it ready to run without calibration panels)
This becomes lots of money so I would feel better not having to buy them all in one go.
I would even feel better if I'd know for sure that I did reach the end of the possibilities of my current hardware. Unfortunately I haven't found a way to do so right now.
Therefore I decided to create a thread so I could get some ballpark figures. I'm still hoping to get those because I know I need to buy a laser but is it really necessary for me to by the LC532-5-3-F(16x65) 178,5 euro plus a wall plug 16,9 euro?
Or will the LC650-5-5-F(14x45)-NT (99,9 euro) do?
My worse David nightmare would be to buy the 99 euro laser not being happy with the result then buy the 178 euro laser to find out I could have done with the laser I had in the beginning. :cry:
Being a newbie I am allowed to not having a clue. :wink:
I feel that if I would know "How to calculate maximum scan precision with a given hardware" I would be better positioned to answer my own questions and so would other newbie's. Maybe this is a wrong feeling. If so please tell me how I can know?

What are the remaining parameters I have to find out?
The parts of the scanning hardware are:
1) The camera
2) The laser
3) The calibration board
4) The laser mover
I'll go into each of what I've understood till now:
The camera:
I know the numbers of the limits of the camera (see above). But are these figures real?
I mean: did anyone ever reach 0.25 mm scan precision in 600X480? If not what is the limit?

The laser:
I know green is better than red but by how far? I also read somewhere that the width of the beam is important. But that is about all I know from the laser.
Is the width of the laser beam the maximum precision? If not; is there a relation between the 2?
I notice that the shop doesn't mention the beam width. Does that mean it is not important after al?

The calibration Board:
This one is nicely documented. Thanks for that. If I understand correctly the board has no impact on the precision.

The laser mover:
As far as I understand the laser mover should not increase the precision.

Hope it is clear what I want to achieve.
Jan

PS:
Rereading my post I notice that maybe it is not clear what I mean with precision. With precision I mean the smallest object that can be scanned. So assume you scan a flat surface with rectangular dots with increasing size. What is the smallest dot (x,y,z) that will be correctly scanned.
User avatar
jantje
 
Posts: 107
Joined: Tue Mar 03, 2009 12:08 am
Location: Belgium

Re: How to calculate maximum precision with a setup

Postby MagWeb » Wed Mar 04, 2009 11:14 pm

jantje wrote:
The camera:
I know the numbers of the limits of the camera (see above). But are these figures real?
I mean: did anyone ever reach 0.25 mm scan precision in 600X480? If not what is the limit?

Bongobat scans with the Philipps SPC900 . Its max res is 640x480. His coin scan should be the best result with the smallest details at this resolution:
http://www.david-laserscanner.com/wiki/ ... f/dime.pdf

The laser:
I know green is better than red but by how far? I also read somewhere that the width of the beam is important. But that is about all I know from the laser.

There are 50% green 25% red and 25% blue sensors on a color camchip. So you´ll get the double resolution using a green laser. The less width the laser owns the nearer is the calculated centerline to real values (edges).

Is the width of the laser beam the maximum precision? If not; is there a relation between the 2?

In perpendicular direction to the laserline >> yes. In direction along the line, the resolution isn´t limited by the linewidth. The speed you move the laser is important in that point too. DAVID calculates the centerline between the upper and the lower edge of the laserline. If you move the line slower than 1 pixel row per frame, one pixel is calculated more than one time while the line moves. So you can get a "oversampling" effect making things visible smaller than the linewidth.

I notice that the shop doesn't mention the beam width. Does that mean it is not important after al?

No, not at all. To get linewidth values, look at WalterMo´s http://www.david-laserscanner.com/forum/viewtopic.php?f=6&t=300

The calibration Board:
This one is nicely documented. Thanks for that. If I understand correctly the board has no impact on the precision.

The precission of the boards and fixing the patterns precisely is most important!

The laser mover:
As far as I understand the laser mover should not increase the precision.

A laser moved too fast smeares ( like a racecar, shot with a cam with long exposure) . A laser freely moved by hand, tends to be too quick some times. You´ll get best results with a well timed smooth movement.

Again. The intersectionangle affects the scan too. As showed above, you might get a good result and a very very bad one with the same setup. If the line would be parallel to the viewing line of the cam you would get no quality at all .

Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Re: How to calculate maximum precision with a setup

Postby jantje » Fri Mar 06, 2009 1:24 am

Gunter
Many many thanks.
I think I got it now.
I've extended the table above with the information you provided and I get the following result.
precision.GIF
maximum precision calculation

Why this is so important for me? There are 2 reasons:

The first reason why it is important is that I can now feel confortable with my scans :D . I know now that given my hardware there is little room for improvenment of the resolution of my scans. If I want to do a better job on my scan I need new scanning material.

The second reason is that I'm happy to know what to buy.
I'm hoping to be able to scan full people one day. The intend is to print them (http://www.shapeways.com ) with a 3D printer.
The printer currently has a maximum detail of 0.2 mm. As printing is expencive I will not print above 200 mm. I'm taking 2000 mm to scan a person (scale 500). So I want to have a resolution of 2mm when scanning 2000mm and scale down 10 times. So that means I'll need to scan 1600 1200 with a green laser. (assuming I do not buy a higher resolution camera).
The width of the laserbeam can be 1.67 which wil be any green quality laser. But red lasers are out of scope for my final goal
So I'm thinking to buy the LFL532-5-3(16x90)-NT.

I'm hoping that other newbies can use the table to feel confortable about their scans as uploading a scan is not that obvious for an unexperianced newbie.

Thanks again for the help and please tell me if the table is wrong.

Jan
User avatar
jantje
 
Posts: 107
Joined: Tue Mar 03, 2009 12:08 am
Location: Belgium

Re: How to calculate maximum precision with a setup

Postby hesaeg » Wed Aug 04, 2010 2:07 pm

hello

I'm new on using the DAVID scanner, and I'm sorry if my english is not very good. :oops:
I'm very interisting in the precision that I can have with the DAVID, but I don't understand how you make the last table.

can you help me please?
I would like to do the same with the logiteck quickcam 9000...

thank you if you can help me !!!

Hesa
hesaeg
 
Posts: 1
Joined: Wed May 05, 2010 3:30 pm
Location: Paris

Re: How to calculate maximum precision with a setup

Postby Archicgi » Sat Apr 02, 2016 4:07 pm

"Hello!

My name is Mike! I represent Archicgi 3d rendering company.
Thanks for yours post, I really like it and want to know can I publish it at our site [url]archicgi.com[/url]?"
Archicgi
 
Posts: 1
Joined: Thu Jan 14, 2016 10:01 am


Return to Laser and Camera Hardware

Who is online

Users browsing this forum: No registered users and 2 guests