Physics and design duality for realistic rendering
Realistic 3D render of a material is often characterized by the way the image or object tells its story and how it echoes in the cultural memories of the audience that will see it.
There is a large part of subjectivity. In this way realism could be seen as sketchy and limited, but not only, because our eyes and brain help us identifying how an object should be.
As thinking beings, we use our sight in conjunction with our cerebral capacities which take care of the analysis, the sensations, as well as the storage of this information. Repeated over time, these procedures become automatic and forge our experiences which serve as a foundation for the constitution of a cultural and visual base. This background culture allows the analysis, the comparison and the identification of objects, forms, and materials in order to constitute our perception of realism. Thus, it seems obvious to us, even without need of artistic culture, to rate the degree of realism of an object produced by a 3D renderer.
Realistic rendering is not only a matter of context and subjectivity, but also a concern of physical properties; many of which can be handled by Ocean™ using a large panel of tabulated material data, either theoretical or measured. (see an introduction to material measurements)
When samples cannot be measured, the solution to have a decent visual and a good physical representation lies in the collaboration of skills: R&D engineers and graphic designers. Work collaboration is the key to achieve this type of project.
To illustrate how the collaboration between R&D and graphic design works we will cover the creation workflow for the nacre material:
Figure 1 – Rendered nacre material in Ocean™
Collaborative workflow, where physics meets design
At Eclat digital, we sometimes work on where optical material properties cannot be measured because of the complexity of the considered materials which sometimes need the use of several instrument devices. (sample size, surface condition, short lead times are other constraints). (see BRDF measurement section in this article)
It is at this point that it is possible to see things in a more “artistic” and theoretical way. Graphic design comes then in complement of R&D engineering material knowledge; theses skills are present in Eclat Digital and provide essential expertise for building realistic materials and shaders.
The need to have a real sample for R&D engineer and graphic designer is the key to understand the light interactions to set up the Ocean™ material shader, and then, choose the best theoretical BSDF models.
The need to have a real sample for R&D engineer and graphic designer is the key to understand the light interactions to set up the Ocean™ material shader, and then, choose the best theoretical BSDF models.
From observations to creation
The first thing to do is to become « familiar » with the material to be rendered:
Pictures
Taking pictures of the sample from several angles with a neutral light source (white if possible) is a good start :
Figure 2 – 15° to 90° picture of the real sample, observing light interactions and surface topology
References
Finding references of the material on the Internets is helpful to understand context and material in other lighting conditions:
Figure 3 – Examples of some reference images where light interactions and the iridescence effect are clearly visible, similar to the figure 2 real sample observation.
Figure 4 – FYI: Electron microscopy image of fractured nacre surface. It is composed of mineral and organic layers. (Public domain image)
Explorations
Deduce the visible interactions with light (depending on the angle of view, the light source type) with a simple diagram of what is visible (reflections, iridescence, surface irregularities…)
Figure 5 – This image shows the different parts and details of the zones that can be visually interesting to create a good material « simulation ».
The purpose of all these steps is to split the observations into distinct parts to have a precise idea of the material’s optical properties.
After sample observation and collecting images references, it appears that the surface gives the feeling of waves : an alternation of concave and convex areas separated by sharp borders. This alternation can be reproduced by using various images generated with the help of image creation software. (see Fig.6)
This material also produce iridescence, characterized by white/Green/Pink colors on the surface.
The use of a graphic creation software allows the step-by-step reconstruction of the surface. This consists in the arrangement of several noise textures adding/multiplying up to each other. These textures will then be used in Ocean™ to simulate light behaviour on the surface according to the intensity of the pixels of the textures. It will give the illusion of a material with a rough surface simulating nacre layers structure.
Figure 6 – The texture result of the nacre surface simulation. In the 3 noise patterns : the lighter the pixels are the more they simulate high areas and the darker they are the more they represent deep or hollow areas. Transformed into a normal map where RGB (Red, Green,Blue) colors correspond to the respective X, Y and Z coordinates of surface normals. This normal map is used in the Ocean™ bump node.
Figure 7 – Using a CAD software to preview the result of the normal map applied on a plane with an environment image for the lighting simulation.
Figure 5 – Creation of the iridescence blend map that is used for the color distribution observed on the sample.
A mix of theoretical model and graphic design skill
R&D engineers at Eclat Digital oversee the shader conception and BSDF customizing operation, including the choice of a Lobe BSDF model, that is, after several renders tests the best option to mimic the material surface. By conception we talk about the way to organize the different nodes and functions inside the Ocean™ shader (Bump, BSDF and what is contained in each of these parts).
In this shader we see the normal map (seen in Fig.6) created by the graphic designer used in the Ocean™ bump node. The function of this normal map is to fake the bumps to artificially simulate the surface asperities by mathematically modifying the geometry surface normals orientation. This “trick” is useful to avoid complex modelling of a surface and can be applied on a flat geometry to pretend the surface is irregular.
There is also a blending map (seen in Fig.8) used to mix the 3 colored lobe BSDF, each of these colors are distributed according to the greyscale of the image. This texture acts as a mask: the darker parts of the blend map displays the first Lobe BSDF, the middle grey for the second and so on. These Lobe BSDF were created by a R&D engineer, each have different spectral contributions according to the observed sample.
Figure 9 – Ocean™ Shader for the nacre material. White/green/pink Lobe bsdf distribution within the greyscale gradient blending map image.
Conclusion
This project shows us the importance of collaboration between various skills. Accurate and methodical observation are essential for the realization of a material without measurement data in the context of image production. Here, the graphic design allows the creation of complex surfaces without the need to scan the sample to reproduce its surface while being procedural, and thus modifiable, reusable at will.
Responses