I then kept working on the modeling and texturing, while also starting work on the shaders (Fig.05b – 05c).
Fig.06 shows the final modeling, sun and camera positioning.
I did most of the modeling work with Maya. This is the software I know the best and am most comfortable with. But I wanted to use V-Ray as the renderer. So in the end I took everything into Max. I also used XSI; the characters are originally XSI's rigged man model – I'll get back to that matter later. I guess my workarounds are not a classical workflow, but that was probably the most efficient for me.
I ran into a couple of challenges along the way. The first one was the composition. I originally wanted only one character in the image. I thought of a girl sitting in one of the baths. To make sure she would stand out, I practiced some openings in the roof to let the sun in and illuminate some specific spots on the floor – where she was sitting (I thought it was brilliant!). But when I showed it to people, they would just not notice the character. The whole image didn't have any focus point to it, which was a problem. Interesting images have interesting focus points. I thought then of having the girl wearing bright colors to stand out. But I wasn't sure about this avenue. I tried yellow tones for the girl clothing, I wanted to create contrast with the greenish coloring of wall and floor tiling (Fig.07a).
Then, after a talk with one of my coworkers, we ended up developing the idea of having a second character somewhere in the foreground that would be observing the girl and would hopefully direct the viewer's eyes to the middle of the image, where she was sitting. Then, maybe they could follow her gaze to travel further in the image. And then they would see… not much, actually. The girl was staring at one big door, and there was not going to be anything exciting outside. And again, people were seeing the second character, but that was it. The girl character was not standing out enough to attract the eye – Fig.07b illustrates this idea.
So, I added a second female character. A total of three characters seemed to work. Also, the two women chatting together would attract more attention. I was now satisfied with my composition. The scene could be read as taking place in the pool of a harem, where two girls are bathing under the watchful eye of a guardian Eunuch. I do not claim costumes or patterns would be historically correct, it is only my personal interpretation.
The second great challenge I had was the characters. I am no character modeler. And since in the end I wanted an illustration more than a modeling masterpiece, I planned to draw them mostly. But yet to get the positioning right, along with proportion, I needed a model to rely on. I decided to use XSI's rigged man model. There are some basic options you can use to customize it such as thinner, larger, with or without muscles and that was perfect. So actually, even the two women were made out of this ‘'guy'' model. Positioning was not perfect, but good enough that I could use that as first layer.
For the foreground character, I kept the pouf and floor to help the positioning. For the two girls, I used a simplify model of the pool (Fig.08).