The BR6104K proved to be a surprisingly useful platoform to develop for; the amount of network interfaces available makes it adaptable enough to provide solutions for a wide variety of problems. Specifically for image processing, however, the 175 MHz ADM5120P and the 32 MB of SDRAM onboard were not adequate; errors after creation of too many threads (6+), slowdowns due to heavy file or memory I/O, and a fair number of library oddities all proved to be obstacles to making an image composition program that could operate at a high framerate. For a project requiring a lot of network usage, the BR6104K would probably be awesome, but the large amount of resources necessary to do imaging work demands more.
On a more powerful platform, challenges would be much different; rather than having to deal with space limitations, algorithms would need to be designed for time efficiency. It's certainly feasible to accomplish what we set out to in this project on something just a little bit more powerful. Faster/more RAM and a faster processor (probably with a different architecture) would do a great deal to bring the MJPEG stream closer to 30 fps. On the software side, piecing together a custom JPEG library to do the image processing on a less decompressed version of the image--RGB data is easy to work with but hard to get--would help save a lot of time. While the goal of smooth video was not realized with this hardware, it very well could be in the near future on a better base system.