Editing Samples

4.3 Editing Samples

From Figure 4.17 to Figure 4.29, we demonstrate some of our results and show all the corre-sponding texture maps created via our material design system. All the materials are created from a single photo or edit from an image.

Figure 4.17 shows all the maps we used to produce the translucent effect that shown in Figure 4.13. Since we add specular color 2 to create such an effect, the index map has a color of yellow (pure red plus pure green).

In Figure 4.18, we use rotation map to produce the specular effect which changes contin-uously over the phi angle. Thus we can observe two-way highlight rotates when lighting or viewing direction changes.

The specular color maps shown in Figure 4.19 have a stronger color distribution near the periphery region. Such distribution creates the glowing effect of the corduroy which can be observed on the chest of the model.

Figure 4.20 shows all the maps corresponding to the comparison result in Figure 4.1. We can observe that specular color 1 has a brighter intensity at the center to produce glossy effect.

Since for a glossy material, we can observe the highlight effect when the viewing direction is almost equal to the reflecting direction. The half vector in such cases would be almost equal to (0, 0, 1) which corresponds to the center of the specular color map. Thus we can draw a bright white dot at the center of the specular color map and use “Blur/Sharpen” paint tool to soften the highlight to make a material glossy. Such effect can also be observed in Figure 4.22 and 4.24.

For the wool sample shown in Figure 4.21, the specular color 2 is the specular color 1 with a 180 degree of rotation. And for the index map, each index region covers half of the neighbor fibers. Such design makes the fibers which looks closer to each other would look far from each other when lighting and viewing direction changes.

Figure 4.23 and Figure 4.25 show two materials that do not have strong highlights. A more uniform color distribution and darker specular mask is used.

Figure 4.26 and Figure 4.27 show two materials that do not exist in reality. For the sample

4.3 Editing Samples 54

shown in Figure 4.26, we blur the height map to smooth the height field in order to let the material have a melted appearance. For the sample shown in Figure 4.27, we specially design the specular color map to let the light blue dots of the material have some blinking effects. Both of the rendering results are quite satisfactory.

The samples shown in Figure 4.28 and Figure 4.29 are created from the same image that is the diffuse color map in Figure 4.28 with a higher intensity. In Figure 4.29, we try to add some lather on the wood. To avoid drawing the lather pattern on different maps at the same location, we add a transparent layer on the diffuse color map and draw the lather pattern on this layer.

Then we copy the lather pattern and paste to all the other maps with some changes; for the specular mask map and the Fresnel map, since the lather does not have strong specular effect, we adjust its intensity to make it darker. For the height map, the lather region has a higher intensity since it is higher than the wood region. For the index map, we give it the blue color which corresponds to the specular color map 3. Since the specular of the lather may have some perturbation, we add HSV noise to the specular color map to produce such an effect.

4.3 Editing Samples 55

Figure 4.17: Minor translucent material.

4.3 Editing Samples 56

Figure 4.18: Disk.

4.3 Editing Samples 57

Figure 4.19: Corduroy.

4.3 Editing Samples 58

Figure 4.20: Wallpaper.

4.3 Editing Samples 59

Figure 4.21: Wool.

4.3 Editing Samples 60

Figure 4.22: Glossy hemp.

4.3 Editing Samples 61

Figure 4.23: Hemp.

4.3 Editing Samples 62

Figure 4.24: Beige glossy leather.

4.3 Editing Samples 63

Figure 4.25: Rusty iron.

4.3 Editing Samples 64

Figure 4.26: Lava liked material.

4.3 Editing Samples 65

Figure 4.27: Artificial porcelain vase.

4.3 Editing Samples 66

Figure 4.28: Wood.

4.3 Editing Samples 67

Figure 4.29: Wood with lather.

C H A P T E R 5

Conclusion

In this chapter, we first give a brief summary of our method, then discuss the limitations of our system. Finally, several future works are proposed.

5.1 Summary

To overcome the difficulties of material design for an artist, we proposed a simple material creation flow and an artist friendly material design system. Artists can create appealing results from a single or a few images intuitively. Learning how to use the system is comparably easier than learning how to use a node based shader synthesis tool to design the appearance of a material. The method first generates the data for a material representation that consists of the material’s meso-geometry, diffuse part, and specular part. Then with our design system, editing is performed to refine the data. Since all the data of the representation are stored into texture maps and our design system is implemented as a plugin of a drawing system, the editing can be done easily. Artists can bring their creativity into free play to change the geometry or lighting phenomena of the material. By carefully managing the resources, the editing effects can be

68

5.2 Limitations 69

observed in real-time which is crucial for an artist friendly design system. Also, the potential of the editing is not restricted to the editing tools what are currently provided. Artists can combine with any other useful plugins or editing tools to help them do the editing. As a result, our system has a great strength in material editing. It can well produce the appearance of general isotropic and anisotropic phenomena for real or unreal materials. Complex specular phenomena which rotate around the surface (e.g., watch and disk) can also be produced easily. Although it would be hard to create a material that has exactly the same appearance as a material in reality, our results captured the main features of the tested materials and thus produced quite similar results.

5.2 Limitations

Although we can create many general materials by using the proposed material design system, it is hard to design translucent materials because the opaque BRDF model that we used does not describe the subsurface-scattering and inter-reflection phenomena. To extend our material de-sign system for more versatile materials, a node based shader synthesis tool can be incorporated.

Another limitation about our material creation flow would be the geometry part. Although the meso-geometry we derived using photometric stereo or grayscale of the image can produce quite good visual effects, it is not accurate compared to the material in reality. For accurate meso-geometry information, a reconstruction approach in [DTPG11] or other depth acquisition hardware can be applied.