‘Huis aan de werf’ testing days, wed. 14 nov.

Post by: Joris Weijdom Add comments
Bodycount

Thanks to theatre ‘Huis aan de Werf’, we were able to perform the comprehensive 3d scan tests, Joris did earlier, in full body scale. With the current knowledge of the David scan software and the rebuild laser device we will put a full body scan to the test in the next couple of days.

Our main goals:

– are we able to build a full body scan setup (or multiple, and can we make it work).
– can we make a solid full body scan.
– can we do this in a certain amount of time.

We started out in the koepelzaal of HadW and got hold of the following hard / software:

– 4 solid wooden frameworks of 2,4 * 1,2meters. (painted white)
– 1 webcam, (quickcam pro)
– 1 dual core laptop
1 laser device
– 1 beamer
– small staircase
– nuts, bolts & tools
– gaffer tape for marking object locations and measurement
– black cardboard, scissors, Stanley knifes
– chair to tape the webcam on
– double sides tape to stick the black cardboard to the frames
– software: David scan software & shapefusion, .NET framework 2.0, directx9.c, win xp/2000
– custom lightning setup / complete darkness / TL light
img_7928.jpgStep 1.
We took out the beamer and projected the calibration model, that comes with the David scan software, on to the wooden frames. We made sure that we projected as straight as possible so we had the least deformation of the projected image. This way we could measure the exact size and radius of the calibration points.
img_7935.jpgStep 2.
Next, we cut the 25 calibration points out of black cardboard and tape these on to the wooden frames. We still had the beamer projecting so we could easily stick them in to place. We used double sided tape in order to remove and re use the calibration dots again.

p1010021.jpgStep 3.
We took the wooden frames and positioned the corners in a 90 degree angle, just as in the small Buddha setup. We had some metal plates, to ensure the firmness of the setup, and screwed them together.
We had to define the placement of the setup in order to get the best out of the workspace.

p1010029.jpgStep 4.
We already had several computer desks set up, placed the camera in front of one of the laptops and taped it to the desk to maintain a fixed position.
Next, we started the David scan software.

p1010026.jpgStep 5.
Because of the calibration that is up next, it is very important the camera will not be moved before, during and after the scans otherwise you’ll have to calibrate again.
To calibrate, aim the camera to the calibration points on the wooden frames and try to get a full screen image of the points setup.
David supports the painting of black noise that prevents the software from detecting more then 25 points. Also, try to keep the corner of the wooden frame centered in your camera image (If your image resolution is 800 x 600, your center line will be at aprox. 400 px.)

Step 6.
We have calibrated the camera and ready to adjust the visible camera settings. Important settings are saturation and white balance. We will discuss these settings in the next few days. Make sure you see a clear light / laser line in the preview window.

p1010034.jpgStep 7.
Next up, we need to define the placement of the laser device. It is important that our angle between the camera and laser is at least 20 degrees at with a maximum of around 40 degrees. Although David might get scan data while the laser line isn’t aligned on the proper angle, your mesh will become much more solid when you define the right angles.

p1010046.jpgStep 8.
When everything was set up, we wanted to make our first scan but found out we were all dressed in black clothes. (we discussed this and agreed that we all should wear kaki clothes in the next few days).Luckily, Arnaud had a white t-shirt underneath and became the first test object. A persons breath is already visible in a 2 or 3 minute scan so Arnaud had to make sure he would stand as still as possible. Maybe, we have to make some sort of placeholder / exo-skeleton were people can lean on.

Step 9.
Directly after the scan, we refreshed the camera settings to a proper view so we could grab Arnaud’s texture. We wanted to combine this texture with the mesh later on in the process. In the next days, we will try to make multiple scans of 1 person to combine those in the shapefusion software in order to get a more full and rounded up mesh.

Step 10.
Repeat step 6 through 9 over and over again for a couple of days to acquire a solid scanned mesh that can be recognized as your own virtual self.

p1010043.jpgOur first day of testing was one of continuous camera, software and lightning tweaking, but we had some successful tests made on full body scale. A couple of important aspects to keep in mind and some important notes:

Stability of the laser line trajectory. For now, we manually move the laser device up and down, but it is preferable to do this automated because of the slight tremors of your hands.

What medium to choose as line projector, a laser or beamer. Although a beamer can provide you with a bright white line and can easily switch between colors and the speed of the trajectory, a red laser is very bright as well and the line is much thinner at a large distance (less then 1 pixel).

Software tweaks. When we used the beamer to project a bright white line, we actually told David we used a red laser and got much better results in doing so.

Use of colors in clothing. Dark clothes do not reflect enough light to be visible on the final scan. In optimal conditions, you could scan dark brown clothing.

Furthermore, important factors in establishing a high quality solid 3d mesh are the scale, the stability of the laser trajectory, the angle of the laser line and the duration of a scan.

Some questions we will seek answers for this week:

– can we make use of a full scale scan setup in our setting. – how do we dress it up?
– how will we give a meaning to the 3d image?
– can we tutorialize our steps for future use in a live performance?
– can we scan black clothing?

Comments are closed.

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