This model was trained from several hundred source images depicting the character from many different angles, positions, and environments with the objective of making the model as versatile as possible. I'm sure there will be coming version improvements as I learn more about this process to dial in the results better. Enjoy!