» Bugfix

from: dach posted on: September-21 2011

Good news for everybody who had problems using the scan application. There was a small but extensive bug on initalization of camera matrices if no calibration file for the given kinect device is present. Thank goes to Ling Huang informing us about the problem.

If you tried the scan application and were disappointed in seeing nothing on the screen just try it again with the updated code. It should work properly now!

Feel free to inform us about any experienced issue using our framework.

» some example meshes

from: schroedt posted on: September-15 2011

» Algorithms for the recognition of footprints

from: dach posted on: September-14 2011

Tobias implemented some experimental algorithms to extract the footprints of out models.
The long term ambition is to fill the bottom side of our models because there is no possibility for scanning the contact surface.
So the area 2mm over the bottom of the point cloud is taken and by using a alpha-shape algorithm the edges of the model are extracted and clustered. After that a convex hull is generated from the clusters and from this hull a concave hull is derived. This approach should provide an encircling line. Having this line it should be possible to fill the area with ponits.

» Improved code online

from: dach posted on: September-13 2011

As mentioned in the post before our new code is now available in the download section.
To indemnify you for the long waiting Christian added a new algorithm for the color mapping which provides much better results then the previous algorithm.
To illustrate the difference take a look to the both frogs below (left: old algorithm, the right one is generated using the new one)

» New algorithm added

from: dach posted on: September-12 2011

To improve the result of the poisson surface reconstruction we added another algorithm for the computation of the normals.
Therefor we took a look at some libraries and decided to use the vcg library. This C++ library for processing and manipulation of triangle (and tetrahedral) meshes provides a very practicable approach for the computation of normals.
Using their algorithm enables us to compute smoothed normals with respect to their neighbors which results in a very pretty to watch set of normals for our point set !
The updated code and the instructions for the installation will follow tomorrow.

» Gallery added

from: dach posted on: August-25 2011

To give you the possibility to view our scanned point clouds we added a WebGL based 3D gallery.
This gallery enables you to browse our models and move them around.
If your browser does not support WebGL you are welcome to use the picture based gallery with snapshots from our scans.

» Validation of our scanning results

from: dach posted on: August-19 2011

Since we are now able to scan our objects in a satisfying way we did a validation of the scanning results.
The size, the reflectivity and the structure seem to be the most important characteristics of the object we would like to scan.

» Small Video Demonstration

from: schroedt posted on: August-15 2011

» Surface Reconstruction

from: dach posted on: July-18 2011

In order to reconstruct a mesh from our pointcloud we examined some approches like Poisson Surface Reconstruction, Ball Pivoting or Power Crust.
After some experiments using Meshlab (a very good tool for processing meshes!) we decided to use the poisson surface reconstruction.
Especially the implemantation from Michael Kazhdan seem to be a good base and we are ahead to implement his algorithm.

» New scan tablet

from: dach posted on: July-7 2011

After we scanned a few object using our small scan tablet Tobias constructed a larger one.
This new tablet enables us to scan larger object like the Nao robot or the bust of Alexander von Humbuld. It is also possible to place this three-sided utility around fixed structures.

» 1. Science Slam in Freiberg

from: backasch posted on: June-16 2011

Diesen Samstag, den 18. Juni, findet der “1. Freiberger Science Slam” statt und wir sind auch mit von der Party.
Martin R. wird dort unser Projekt vorstellen und freut sich ├╝ber zahlreiche Interessenten. Also wenn Ihr Zeit und Lust habt dann kommt doch einfach vorbei.

Wann: 18. Juni 2011, 21 Uhr
Wo: Karl-Kegel-Bau in Freiberg, Raum: KKB-2030

weiter Informationen erhaltet ihr unter 1. Freiberger Science Slam

» First Scan !

from: dach posted on: June-15 2011

Today we scanned our first object:

Yes it is a telephone – the good old and reliable standard phone.

» A Masterpiece Of Architecture

from: backasch posted on: June-14 2011

For our way of 3d scanning with the Kinect, we need some tools.
We did not have to rent Bob the Builder, because we have Tobias – “the hatchet man”.

He build a pair of 2 smaller Scan-Tablets. We can use the Scan-Tablet for calibrating the Kinect too, because on the back of each tablet is a checker-board.

For more information follow “THE LINK“.

» Portable Kinect

from: backasch posted on: June-13 2011

To use the Kinect in the wild wild nature, where no man had gone before, or in the urban jungle where we can’t find a power outlet, we had to make the Kinect portable.

For more information please read: Battery Pack