Conclusion

In my experiment I wanted to make a program to contour images and make a 3D model out of it. In short it was a success, as the programs work enough to contour a specific image, however there are a few shortcomings. Firstly, on a laptop it takes 30 minutes or so to contour one image with one method. Additionally, this is an AI program, so it doesn’t detect the artist’s purpose as well as a 3D modeler from something like Unseen Art. Thirdly, either the filament or 3D printer seemed to print fairly messily, with strings of filament all over. Fourth, the program is difficult to use for an outsider, to use the color contourer, you have to say the name of the file you want to contour, the type, 0 or 1 for having white be high or low. After this, 3 windows pop up, one being the normal image, image with contours highlighted, and the contoured image. Both programs run, being able to contour an art piece from Picasso and then be able to 3D print said model. The color program was very sensitive to different colors and brush strokes, and could mess up a whole print since it thought it should be contoured differently.

Second drawback, the printing itself. The printed models look somewhat sloppy for some reason, either being the printer or print head if not the filament itself. As the photos show there are strings everywhere and it looks chaotic. I haven’t been able to do more tests with the printer itself to see if it is the printer, my modeling, or both or even neither. That shortcoming can probably be tested in other experiments, just seeing if 3D models change in quality in different printers or modeling softwares.

Additionally my program struggles with solid color things, like hillsides for example. I used Picasso's art piece since it has many varying colors in zones from each other, making it easy for the algorithm to detect colors/zones.

Overall the program works, as it accomplishes both color and zone contouring. As stated before it completes this but does leave some quality/refinement to be desired. To improve on this some further experiments could be testing separate 3D printers to see quality, as many of the models were messy. Another experiment could be comparing a contouring program to someone manually creating a 3D model. A third experiment could be expanding on mine by making more methods of contouring to be able to do more average pictures, like landscapes.