WOW where did you learn to figure weight. There is absolutely no way your back plus film will 2-3 ounces in weight. A standard 4x5 film holder is about 5 ounces an its 1/2 inch thick. Your talking about easily adding several pounds of weight on to the thing.
Dude, I was talking about ADDED weight from the roll system, not the whole camera or back. As in: [weight of a back with a roll system accommodated] - [weight of a normal back without a roll system] = a few ounces.
Which was too low. I indeed underestimated. However, the right answer is closer to my estimate than "a few pounds" is. Here, I actually went and looked up the densities and such exactly, and it would be:
1) Two basic hollow plastic (for example PVC) reels with an inside diameter of approx. 1.5 inches and a height of 8 inches each =
9 ounces total (liberal estimate to allow for a small arm inside to hold the axle connections). Very thin aluminum tubes might actually be lighter weight.
2) Assorted screws (I just weighed a bag of random screws I have on the table, approx as many as I would need to make a couple hinges and fix reels and a couple inch long turning crank) =
4 ounces
3) To fit the reels on the sides without getting in the way of the image, I would make the camera a trapezoid large enough to fit them in the corners. The difference between a 14/10 x 5 x 8 inch trapezoidal box versus a 10 x 5 x 8 inch box = .38 inches x 8 inches more material on each side wall + 2 x 8 on the back + 20 square inches more material total top and bottom.
If the walls and top are 3/8" plywood and the bottom is 4/4 lumber, then that adds up to 18.14 cubic inches more wood. For softwoods like douglas fir, this is about
5.5 ounces
Total added weight =
1.16 pounds.
And the roll system could conservatively hold a good solid 20 sheets of film, which would require over
3 pounds with standard 2 sheet holders or grafmatic holders (grafmatics are apparently about equally heavy per sheet as standard holders). Making it a significant weight savings.
Diagram:

edit: the blue circles should be hollow. Too lazy to fix.
Note: this design is assuming a radius of curvature approximately equal to what those people using the "taco method" of developing were using. I do not know if their method involves enough bending to actually damage the film or the image. If so, then the rolls would need to be larger, and so would the trapezoid, and the weight addition would go up. Nobody seems to know how much bending is okay though, so I may need to just test it empirically first somehow before spending time building a camera (shoebox pinhole anyone?)