This paper aims to contribute to the colour management methods employed in the digital post–production of cinema images. It consists of building an accurate characterization model of the process of printing from a negative film to a positive. Several different generic methods were tested and their results analysed in order to assess their performance and gain more knowledge about their intrinsic nature, as well as that of the process itself. The neural network model outperformed the others. Its flexibility enabled it to achieve errors well below the variability of this process. It is the only one that can reach the very high quality required by the digital cinema industry. Colour copies of the figures and the VRML files may be found at the author's web page at alexisgatt.free.fr/publications/negative.htm.