An "algorithms" course in Computer Graphics relies heavily upon programming exercises to teach students the intricacies of graphics techniques. Such exercises should be substantial enough to avoid "trivializing" the problem and, at the same time, be interesting to the student. Unfortunately, substantial exercises also introduce a lot of overhead not directly related to the central graphics concepts being demonstrated. Overhead such as complex data structure definition and manipulation, external file input and output, low-level user interfaces, and graphics device drivers add a significant programming burden to the student. This overhead also detracts from the desired learning experience of the exercise. Various alternatives for solving this problem while maintaining the complexity of the exercise include: 1) designing the exercise to allow previous solution reusability, 2) providing a conceptual solution, 3) supplying actual code, both "white box" and "black box", and 4) making development tools and aids available, This paper will describe these different approaches, examine their advantages and disadvantages, and summarize our experience in the graphics course at the Air Force Academy.