Many tasks are currently automated in the manufacturing field, but some are still performed by workers, which require technical knowledge and skills. However, it is difficult for a new worker to learn skills that require precise motion and extensive experience. Conventional on-the-job training (OJT) has a problem, which is the difficulty of defining and evaluating a correct task. Therefore, this study aimed to visualize the proper skills and their effects on quality in metallic painting operations, and to design the concept for a skill training system. Product quality depends on a worker's motion and the effects of tools. To train a skill based on these factors, it is necessary to visualize the proper skill and to devise a method of training. Therefore, changes to a worker's motion and the quality of products were visualized using a motion capture (MOCAP) system and three-dimensional computer graphics (3DCG) software. Furthermore, such a system requires a trainee to experience the proper motion and to evaluate his or her skills. The motion of two skilled workers was therefore analyzed to extract their skills as explicit knowledge. The skilled workers' motion was measured using the MOCAP system. The obtained data were then analyzed as positional and rotational data in three axes for each motion of the main parts of the body. Consequently, 11 types of skills were extracted by analyzing the data. Then, an experiment was conducted to visualize the quality of the metallic painting operation using film thickness as an index. Seven factors for film thickness were extracted, and these were quantified in the experiment. Moreover, these factors were applied to 3DCG, and the film thickness was reproduced by simulation. As the result of a statistical test, the reproducibility of the film thickness was guaranteed in 3DCG. Based on the results, a training system was designed for to enable a new worker to learn the skill.