Sadly I can't find any images or videos of the product of the device. The site has a couple of videos showing it throwing sparks, but that's about it.
This is a nice first step and all, but there are a number of challenges here. The irregular sparking and popping noises imply that the metal laid down will also be irregular. This uses a regular MIG welder as the print head, and MIG welding only allows you to lay so fine of a bead. Large welds tend to be brittle due to internal stresses, voids, and the like.
The more successful commercial metal 3D printers build parts a thin layer at a time. A fine metal powder is sintered on each layer using a laser. This technique is called Direct Metal Laser Sintering (DMLS).
[1] http://hackaday.com/2013/12/07/a-rostock-welding-3d-printer/
Prometal, for example, has a process that's very similar to SLA or SLS, but with a stainless steel powder instead of resin. They then infuse it with bronze, using capillary action to do the work. You end up with pretty solid parts.
Otherwise seems like a very interesting project.
Thanks to jeffmcjunkin above for linking to an image of a printed part. It certainly looks better than I anticipated.
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6678...
There's only a vague resemblance between model and part. No way could you use that sprocket. Not only are the edges rough, it has voids. You could grind or machine down down rough edges, but voids make the process worthless even for making blanks. That's about what you'd expect with MIG welding. Welding just isn't a good way to make smooth surfaces.
The low-end 3D printer people tend to obsess on the problems of building a 3-axis motion system (a completely solved problem in CNC machines) and gloss over the problems of what's happening at the weld point. The common extruder-type printers are trying to weld a hot thing to a cold thing, which never works very well. (In soldering, that creates cold solder joints. In welding, it creates bad welds.) Heating the bed plate helps a little, but once the thing being built is more than a few cm high, the build plate isn't accomplishing much. That's why most taller objects made with low-end 3D printers fail.
There's a plastic welding technique that's been in use since the 1950s that works reliably. (http://www.professionalplastics.com/WELDER). Not only is the welding rod heated, the target area is heated with hot air. This produces a solid, reliable weld. I've suggested doing something similar for extruder-type 3D printers, using small lasers (2-5 watts) to preheat the target area just ahead of the extruder. So far nobody has done that.