‘Huis aan de Werf’ testing days, thu. 15 nov.

Post by: Joris Weijdom Add comments
Bodycount

After our first day of testing yesterday, we decided to buy a higher quality camera and bought the Logitec quickcam pro 9000. This camera can record in an ultimate 1600 x 1200 (we record in 800 x 600, but still), 30 frames per second and can grab textures in at least 3 mega pixel.

p1010029.jpgToday we wanted to run some tests in changing the height of the camera. We started with the camera fixed on a chair, but in the afternoon, we got hold of a tripod and were able to slightly change the height after each test.
So far, the hardware we use today is not different from yesterday. The important research we have to do today will be the further adjustment of the software. David does not automatically adapt to an environment so we have enter the settings manually. For this, we don’t use the standard settings display that comes with the web camera, but instead only use the settings menu in the David scan software.

Later this week, we will share the settings we got the best results with, but for now, just a couple of important experiences.
In our tests we used a dual core processor against a Pentium 4. They both obtained scan data but the framerate was much higher on the dual core laptop and this resulted in a much smoother 3d image. A framerate (in the actual preview window of David) of 10 FPS will do, but around 15 FPS is optimal.
Another important issue is noise reduction. This is much more important then the recording resolution because of the scattering in your final image. You will have less floating pixels data in your final scan when you adjust your noise level the right way. Also important in reducing the noise in your image preview is to avoid large brightness difference, but this is also clearly mentioned in the online help of the David homepage.

p1010035.jpgWith the new camera, the calibration went smooth again, but a couple of  errors kept bugging us for the next hours. We browsed the fora and faq’s for some time on the three comments David kept coming during a scan:

  • – not detected on a left or right side / not on a plane

Keep the laser line in a level position, check this through the preview window on your pc). We tried to put extra wooden frames on both sides of the scan setup to enlarge the scan cabin, but it seemed the camera had difficulties only when it crossed the black cardboard calibration marks on the wooden frames.

  • – laser line not detected

p1010021.jpg p1010033.jpg p1010038.jpg
We tried to project in different colours. First we tried a thin shadow line (black) but got no data at all. A beamer producing a black line will spread a lot of bright light as well that needs to be adjusted in the camera settings again. Next, we tried to project a green and red. Both lines didn’t reflect as much light as we hoped so we tried a bright white scan line. We changed the thickness of the line between 1 and 2 pixels in height (1 pixel is aproximately 1 centimetre in height on the screen)

  • – intersection angle to low

Hold the laser in a different position from the camera (most of the time higher). We scanned from a top down position most of the time, but projecting the laser line from a fixed point below the camera worked out as well. Gathering extra information by scanning from both low and high positions will deliver a more solid scan, although it requires the scanned person to stand still for quiet some time.

verhoudingen_scancabine_02.jpgWe have been projecting from different positions during these two days.
The first setup had the beamer approximately 2,5 meters behind the camera and only 50 centimetres on top of the camera position. This way, the intersection angle was  too low, and we changed this by fixing the beamer in a much higher position. (see drawing).

With the new beamer position, we started out to do some extensive testing for the rest of the day. We used several different projection speeds, as well as different scan line thicknesses and colours. We also tried to move the camera position up and down (we kept it at the same distance from the scan cabin, around 3 meters).
These are the settings we used today during the last scans:

Settings:

brightness                4980
Contrast                  1254
saturation                1529
Scherpte                 7058
White balance          3241

Advanced settings:

Scherpte stelling       1
Belichting                 1/120
Versterking               2061

Although we had some better results then yesterday, the overall quality of the scans was a bit disappointing. Most of the time, David picked up little data en has difficulties with dark clothes. Also, David seemed bugged during some of the scans but we are not sure whether it is the laser angle, the background, the colour of the light line or even the scan cabin background. Something we will explore in the next two days.
img_7946.jpg p1010083.jpg p1010085.jpg
But still, When we placed a texture on top of the 3d scan, it really comes alive and has a strange and thrilling appearance. It looks as if it is still under construction, but that’s also the beauty of the scan.

p1010093.jpgThis beauty will have to confront the real life owner in such a way, the person has mixed feelings of leaving his virtual self in our hands, and trusts us with his digital clone.

Comments are closed.

WP Theme & Icons by N.Design Studio
Entries RSS Comments RSS Log in