SL Coins

[MagWeb]

Hi

scanning coins is a nice thing to show microscanning qualities. Until now, scanning coins was a laserscanning thing.
Maybe some guys will join this thread to add their cents.

Here´s my first attempt scanning a coin with DAVID 2.x

NL meets SL:
Singlescan, Interpolated at 3, no smoothing;

Gunter

[hal]

... incredible precision and accuracy.
as ever at your feet, Master Jedy
I'm impatient to try the SL technique, but I want to wait the full version implemented into David r.3.
Hope to try it first of the next summer

Mattia

[DBS]

Incredible !!!!

I have no words. Guis your are the best.

[WalterMo]

Hello Gunter,
Very nice! Simply perfect.

But I am waiting like Mattia for DAVID V3 to then start SL scanning with full load output.

Greetings to Bavaria,

Walter

[Khalid Khattak]

exellent..will wait for V3...

[MagWeb]

Hi Guys,

thanks a lot...

Think. there´s no need to wait until 3.0 is out.
This was made with 2.x: no tricks but simply using a "negative corner" to avoid internal reflections doing the reference scan. The either scan was done planeless...

If I learned one thing today: Focusing and setting the cam is the key to get it.
V.3 can never compensate cam or projector issues ... Essential stuff is here allready to do real good scans on "patient" objects. V.3 will speed it up, but will not be a miracle.
If you own a projector or plan to go for SL; Do it......now.

Think in this pre-3.0-phase we could give some valueable hints for further developement (and: ease the preasure to do a release) -

Simon and Sven gave us the chance to gather experiences with SL for free! So use it, do it... if you do it, you´ll feel 3.0 in a preview.
....
Here some further results of today...maybe not as perfect as the NL-coin above

Gunter

[hal]

Think. there´s no need to wait until 3.0 is out. [...] V.3 will speed it up, but will not be a miracle.

That's exactly my problem: the time
I know that the v.3 SL system will be the same of the v.2, but the speed of the entire precess will be the real new feature.
I've tried, long time ago, to do some scans with SL system, but the workflow is too much long for me, now. I'll be waiting, and in the meantime I try to learn from your experience. Thanks for sharing.

Mattia

p.s.: May I ask you to know the total time that the actual SL process require to obtain such beautifull coin's scans? How many minutes takes to you, the entire SL process? Thanks.

[MagWeb]

Hi Mattia,

just tried it out:
3.30 min from powdering the coin to saving the mesh ( on a ready calibrated setup ) to get this.

With slight adjustments of the cam settings, maybe a check if the coin was actually in the depth of field this could be improved...

Gunter

Edit: Added another one. Both below are special German 5.-DM coins 32.5 diameter

[hal]

ny thanks for the info. My first attempts were more time wasted... so I need to train myself better and more seriously.
Hope to find some time to try and post the results.

Again my compliments for the excellent results.

Regards,
Mattia

[Khalid Khattak]

Incredible scan results.. what camera you are using?.. can you post the hardware setup pictures?

[MagWeb]

Hi Khalid,

I used the same setup as posted here:
download/file.php?id=4140&mode=view

but without the water-basin.

Gunter

[moog]

Your results are flawless.
Please can you tell me how you have avoided ripples in the scan, in another post, someone corrected this by altering the gamma in photoshop.
You mentioned that you've found that camera set-up is very important, can you elaborate.
i've done a scan at low resolution 640 x 480 with the camera zoomed in, which i would think (with optical zoom) would work the same as moving the camera closer, but my scan looks like this....
With interpolate and smooth average set to 1.
I plan to try moving the camera closer, and increasing the resolution, to see if this improves the scan.
Any tips would be appreciated....

[MagWeb]

Hi moog,

a generated simulated SL moving line owns a perfect motion.

That´s why I allways use planeless scanning.
AND
UseRememberedPlanePoses set to FALSE.

This way the reference scan is used only to get the basic range information (were the projection is in 3D). Inside this range all is thought to be ideal motion.

Now, your repetition images have to be of good quality:
1. Pay attention that contrast/exposure is set to keep the shadows dark (means not brighter than the first image shows darkness) . I test this on projection 7.
2. The sin wave projections have to be sin waves in the images. Not oversaturated nor to dark - This is a tricky thing, one has to get a feeling of... I found it much simpler to start with some basic cam setting and to dim down the projector ( by using RGB channel mixing - I made my own projection maker - gonna post it as soon as its messy code is a bit cleaned up ).
3. The depth of field should be equal for cam and projector and your object has to be in that field. This depth of field is a rather small one on such small dimensions. So I use a 120° corner to do calibration and referencescan.

Gunter

[moog]

Would you mind looking at these to see what you think needs improving..
They are images 1 to 7 (it was too big a file to include them all)
I moved the camera closer ( EX-FH20), and set the remembered poses to false.
I don't know if i can set contrast and exposure i'll have to look at the manual.
I increased the contrast of my projector, but it didn't have much of an adjustment
Thanks..

[moog]

I've just been looking at the manual for my camera, i can set a manual exposure, and also "white balance" can i assume they mean the contrast?