Digital selfportrait
The image has been produced by using a computer program I wrote. Input data for the program was obtained by using a 3D scanner (it uses a laser scan to adquire 3d data).
This scanner needed various passes (from different points of view) to cover different parts of the surface. Then a comercial program was used which joined polygons into a single 3D model, containing 260000 triangles. I was not absolutlelly quiet between each scan, and perhaps this caused the stitching error you can see (whole process took a few minutes).
From this data, the program used a realistic rendering algorithm, (based on Monte-Carlo techniques), wich simulates light propagation and reflection on the 3D model. The program was run on a 8 CPU linux computer cluster (with different processes comunicating via the MPI library), and was written in C++ programming language. The program completed the image in less than 5 minutes (4m 45 sec.)
The program used an algorithm called path-tracing. For each pixel, a total number of 400 light paths were simulated, and average pixel radiance was computed from this set of samples.
Finally an image was obtained. I used GIMP to post-process this image. Background was removed (set to black), and contrast was enhanced on the head. Finally, noise was removed by using an edge preserving low-pass filter (a little bit of noise was produced by Monte-Carlo algorithm, and then it was made more visible because of contrast enhancing algorithm).
The image I've uploaded is a JPG version fo the original tiff image, thus some JPG compression errors are visible.
All equipment used belongs to the Computer Graphics Group at the University of Granada.
Digital selfportrait
The image has been produced by using a computer program I wrote. Input data for the program was obtained by using a 3D scanner (it uses a laser scan to adquire 3d data).
This scanner needed various passes (from different points of view) to cover different parts of the surface. Then a comercial program was used which joined polygons into a single 3D model, containing 260000 triangles. I was not absolutlelly quiet between each scan, and perhaps this caused the stitching error you can see (whole process took a few minutes).
From this data, the program used a realistic rendering algorithm, (based on Monte-Carlo techniques), wich simulates light propagation and reflection on the 3D model. The program was run on a 8 CPU linux computer cluster (with different processes comunicating via the MPI library), and was written in C++ programming language. The program completed the image in less than 5 minutes (4m 45 sec.)
The program used an algorithm called path-tracing. For each pixel, a total number of 400 light paths were simulated, and average pixel radiance was computed from this set of samples.
Finally an image was obtained. I used GIMP to post-process this image. Background was removed (set to black), and contrast was enhanced on the head. Finally, noise was removed by using an edge preserving low-pass filter (a little bit of noise was produced by Monte-Carlo algorithm, and then it was made more visible because of contrast enhancing algorithm).
The image I've uploaded is a JPG version fo the original tiff image, thus some JPG compression errors are visible.
All equipment used belongs to the Computer Graphics Group at the University of Granada.