Our investigation of digital archiving research projects has shown that this is an active and exciting field today. Through the groundwork laid out by the VASARI and MARC projects, there are now a number of other digital archiving projects. Many of these are web-based, such as the online presence of the famous Hermitage museum in Russia. Now that the tools for digital archiving have been developed, individual groups are using them to create digital image libraries around the world.

            The implementation of the digital archiving process that we performed was especially enlightening. While following the footsteps of the VASARI project, we found novel ways to simulate complicated technologies. In doing so, we gained a better understanding both of the engineering solutions that the VASARI researchers used and of the difficulties inherent to the problem of digital archiving. The camera calibration procedure showed us the importance of understanding the specifications of the hardware we used. Once we had taken the sub-images, we learned that the image capture is only the beginning- stitching images together is not an easy task. It would have been nice to print out a final full-size replica of our painting using calculated XYZ values, but unfortunately that was not to be the case.

            Were we to do this project again, the area we would most improve would be the mosaicing of the sub-images. Perhaps more customizable stitching software would allow us to create a better stitched image. Of course, it is also clear that if we had more exact registration of sub-images we could even use simpler mosaicng software. Better registration might include a computer controlled camera positioning system, the precision of which would tell us exactly what parts of the sub-images matched with other sub-images.
 
            Another way to improve the project would be to use a customized digital camera that did not perform the processor adjustments in the first place. If this were the case, we would not have to do things like correct for the gamma function. Thus we would retain the data lost by our conversion approximations.