There is a lack of in-depth studies focusing on an objective determination of 3D software rendering quality. This is probably due to the vast range of available rendering methods, algorithms and principles, combinations of settings and software solutions. However, the issues of visualisation realism and reproduction accuracy of the interaction between the light and the surfaces (objects, materials) generated by various rendering engines are of prime importance to the 3D artists who are working in the field. In the research three rendering engines in Maya software were analysed in detail and compared to one another. The main findings are summarized below.
The number of rendering settings and their combinations was the lowest in the case of Maxwell renderer, where there is only one parameter that a user can adjust in comparison to 2 main parameters and 6 secondary settings in mental ray and 2 main parameters and 4 secondary settings in V-Ray. Consequently, the number of
tested combinations differs from 1 for Maxwell renderer, to 16 for V-Ray and 24 for mental ray.
In spite of the Maxwell renderer’s ease of use, its algorithms are physically the most realistic with a very long rendering time, which is in general 7 to 10 times higher compared to that of mental ray and V-Ray. Unfortunately, due to the small number of parameters that a user of Maxwell renderer can manipulate, this time can hardly be reduced. On the other hand, in mental ray and V-Ray the user has more possibilities to change the settings and their combinations and therefore to costumize the rendering procedure and to affect the result.
To ensure an unbiased comparison, renderings of the three engines were corrected with light setting correction within each engine and using identical gamma value (2.2). Despite a good visual match of final renderings after the correction, the obtained colour differences (∆E*ab) between outputs of different engines were on average higher than 7.0 (measured on 19 different regions of interest).
Figure