DAVID 2.2.0: Scanning without background panels!

[Sven]

Hello Raya,

As we send the license via e-mail, there are no shipping costs. I have sent you an e-mail with details.

Best regards
Sven

[RAYA]

Hi Sven

So how can I have usb version to load it on any system?

I want your bank account also.

Raya

[WalterMo]

Hi,
The question was: Would it be possible to scan even a coin without background panels? And if yes, what about the difference to a „normal“ scan method with a background corner.

For this comparison I used my stepper driven teeter board with a slide projector on it. It just figures that for a coin scan a high mechanical reproducibility between reference- and object scan (repetition) would be necessary. A perfect programming of Sven was assumed.
The attached screen dump shows left the coin scanned with the „normal“ technique and in blue the scan without a background. All settings were the same and the two scans were made in one go.

The periodical horizontal stripes on the right coin are due to an imperfect drive of the teeter board. The threaded rod together with the cardan joint caused a small but visible (at the drive) deflection. For such a heavy duty the mechanic was a bit too simple. But David works well!

Merry Christmas to all members of this forum who celebrate it


Walter

[RAYA]

Hi

I have attached a scan without background panels but when I change the resolution the mesh reformed also.

Raya

[WalterMo]

Hello,
Now after improvement of the drive the scan of a coin without background panels is much better. I have removed the cardan joint and made a stiff connection between motor and threaded spindle. But now it was necessary to let the motor tilt in one direction.

Further details:
Camera: Logitech QuickCam Pro 9000
Light source: Slide projector with a 150W halogen bulb
Slide: Razor blade one

The stepping motor was connected to an up/down counter to come precisely back to the start position.


Walter

[RAYA]

Hi Walter

You have mentioned important factor to decrease backlash of a tilting system (removing cardon) I suggest that you change your mechanism so that you have a single axis transverse first and then add some two arm to it.

Thus you will have a tilting device which has less backlash.

Raya

[Bongobat]

Hi Walter,

Very nice results. How do you avoid noise spikes from reflections without coating the coins?

Thanks,

Greg

[WalterMo]

Hi,
RAYA, the imprecision of my coin scan from Dec 24 has nothing to do with backlash problems. There was/is always a small force in axially direction on the spindle. Also the threaded spindle itself was ok. It was only due to the imperfectness of the cardan joint because of a small radial deflection during rotation.
I am still convinced that a teeter board is better than a tilting board because a smaller motor is necessary. Especially if a heavy projector is in action.
A photo of the changed drive will be presented shortly on the thread „White light instead of.....“

Greg, I don't know how I avoided spikes. They don't appear. I have played a bit with the camera settings, especially with brightness and contrast. Nothing more. The intersection angle was about 60° from the vertical and the scan was 1x interpolated and 1x smoothed. The quality of the silver coins was perfect. They looked like as just comming out of my own mint press.

Walter

[RAYA]

Hi Walter

My mean of backlash was deleting radial force this act as backlash to having constant angular velocity not nut backlash.

My mean of tilting is acting as door opening/closing with line replacement and I think you mentioned tilting board instead of rotary board.

Raya

[vectral]

Hi to all, i am happy that someone appreciated the scanning method that i illustred a lot of posts ago . I obtained really good results with shadow scanning in that way because the object can fill the cam and tiny details can be captured. I am working on a new setup using a stepper motor and a cd-player mechanics to scan without background..... hope to post some good results early....

[WalterMo]

Hello,
This nutcracker was fused of nine single scans each scanned without background panels.
The scanner was a stepping motor device already presented some time ago:

viewtopic.php?p=260#p260

In the meantime the 3.75° motor was replaced by a 0.9° one. So the resolution could be increased to 0.009° per step in the half step mode. This corresponds to a displacement of the laser line at 1m to 0.157mm per step. Furthermore an up/down counter was connected to the stepper to come easily back to the starting position after the reference scan (and the other object scans of course).

Further details:
Camera: Logitech QuickCam Pro 9000 at a resolution of 960x720 pixels @ 15 fps
Laser: Focusable green LC532-5-3F(16x65) from David Laser Shop
Scan speed: 7.7 steps per second
Single scans: 1 x scanned (other is impossible at the moment), 1 x interpolated and not smoothed
Fusion: Poisson, resolution = 1000, Smoothness = 2, Removal = 3

Walter

P.S.: Do you know whom famous man this nutcracker shall allegorise?

[Bongobat]

Hi Walter,

Nice Scan.

To me it looks like Charles de Gaulle?

Greg

[WalterMo]

Greg,
Yes, it's General Charles de Gaulle. I hope I haven't offended the French people.

In the meantime I have put the nutcracker as a pdf file on the David WIKI, Multimedia Gallery.

For the transformation from *.obj to *.pdf I have used your recipe. Thanks for programming.
But having views from all sights it's obvious that there are some imprecisions. E.g. the space between nose and peak of the cap isn't free. And the peak of the cap itself has a gap. But for me it's OK.

Walter

[hal]

Hello WalterMo,
perfect scan! Seems that the new David without panels is great as the version with panels. The precision seems the same.
Compliments, bravo,
Mattia

[WalterMo]

Hi,
Yes, now it's proven by MagWeb and me that David also will work as perfect without background panels as with panels. And the electronics should also be no problem. But perhaps the mechanics!

During the testing phase for the new feature Sven noticed that the mechanics of my old scanner, already presented in April 07, didn't work perfect. The cause seemed to be little flatness imperfections of the sidewalls of the worm wheel. After a replacement of the cogwheel set the scanner worked well as required.
This means that much more care for the mechanics is necessary when we want to scan without the panels. Especially when the object is „far away“. We must realize that every mechanical inaccuracy will be amplified by the distance.
Also any backlash between the two cogwheels must be avoided. My old scanner had a wounded spring around the shaft where the laser was mounted to prevent this. Now I have improved this mechanism by a stretched rubber ring wounded around a spool. See attached photo.

Further details:
0.9° stepper motor, half step bipolar mode.
Gear transmisson ratio: 50:1
Laser tilt resolution: 0.009° per step