Note: a true write up would have included the values used to achive the results.
I implemented the following features:|
Create a composite image of myself and a famous place.
I started with the scene of the city and myself. I created the masks by
hand and them composited the images together.
Point Sampling, Bilinear Sampling, 3x3 Gaussian Sampling.
The goal of this filter is to approximate an etched image where the line width is
inversely proportional to aveage luminance of the area it covers.
The filter has three parameters: theta: this is the angle of the etching lines (in degrees) lw: this is the maximum width of a line in pixels. This also loosely represents
the amount of gray levels available (i.e quantization level) sw: this is the distance between lines in pixels The filter works by first rotating the image by theta degrees. Then the average luminace is computed for two 1x(lw/2) regions of the image this number is used to determine how many pixels to set to
black = (lw/2)*(1-avg_lum/255). Thus the average luminance is kept approximately
The image is then rotated back, and then cropped. Note at this time the crop starting
point is sliightly incorrect. I will fix this in a later version
This algorithm need a high amount of highly correlated data to yield pleasing results.
So it is advisable to use the sampling flag set to 1 or 2 and the scale flag set to
a value proportional to the lw value. It may also be useful to blur the final result
or scale it back down to it's original resolution it the image was to be displayed on screen
For example the Einstein image was created with the folowing commands:
image.exe -sampling 1 -scale 4 4 -NPR 30 16 2 < einstein.bmp > einstein_etch.bmp
image.exe -sampling 1 -scale .5 .5 -blur 9 < einstein_etch.bmp > einstein_etch_small.bmp
The Dobkin image was created with the following commands:
image.exe -sampling 1 -NPR 30 16 2 < dobkin1.bmp > dob_etch2.bmp
image.exe -sampling 1 -scale .5 .5 -blur 9 < dob_etch2.bmp >dob_etch2_small.bmp
Art Contest: Please enter my dobkin and einstein images into the art contest