Making Of 'The Virtual Mime'
Modeling the Head
I started the modeling process with the head. To get the correct proportions I took photographs of a friend and mapped the results onto two planes. Special thanks to Rob Yorke.
A spline cage was made from the pictures by first tracing the profile and front view, then simply adjusting the splines until they looked right. The advantage of this type of patch surfacing to NURBS was that I could create more detail in a place like the mouth and eyes, but keep the detail low at the top of the head. The output of this spline cage is polygons.
After a surface modifier is connected the surface appears. Two vertices must be locked together to create one edge of a polygon.
When the spline cage was finished it was mirrored to the other side. Later on I fixed problems with the model and welded the two parts together to create one seamless mesh.
The ear was difficult at first to create. I started with a circular spline and duplicated it a few times. I changed
the sizes of the splines to match an ear then connected them together. After that it was just a matter of moving points in order to acheive a realistic look.
When the ear was finished I imported it into the scene with the mime's head. The two spline cages were connected then the individual points were connected. When connected correctly the surface would smoothly appear between the ear and head.
The eye was created by making five different parts as seen below. When a light is placed above the character the iris will respond correctly to the light by being brighter in the lower half. The reason for this is because I want to have a few very close shots of the mime's face in the animation. Also whenever people watch a person on screen they always watch the eyes for emotion and intelligence.
Modeling the Body
I was the model for the body of the mime. The arms and legs were created in a similar fashion to the head. As you can see below, the spline cage for the arm floats right above the surface of the arm. This can be tweaked in real time to see what the result is on the surface.
The vest was more complicated and was created in three parts-two front pieces and a single back.
I tried to create the hand with spline cages but I had to much trouble. It was very hard to get the fingers to seamlessly integrate with the palm. So I tried another way- box modeling. This proved to be slow but it worked very well. I began with a single box for the palm then cut and extruded the fingers. The hand to the right is the one I created. The one on the left is the same model smoothed.
The low-poly hand works similar to the spline cages- it can be tweaked and the smoothed version will update.
I wanted the clothing to have detailed creases and wrinkles, something that I couldn't do easily with splines. After I was done modeling with the spline cages I converted them to low-poly meshes as seen on the right/below. Then I made a reference object and smoothed it. Now I could use box modeling to create more detail. The polygons were cut into smaller pieces and the points were moved around to make wrinkles.
I only made wrinkles in the areas where his joints were because these are supposed to be tight fitting clothes. The nice thing about this is that it put more detail in these areas as well- so that the bending of joints will be smoother.
I used the same idea to create the wrinkles on the sleeves. I spent more time on the sleeves because they will be much more noticeable in the animation.
Here is a finished render of the sleeves.
When all of the body parts were created it took a lot of control and reference objects. It was really nice working this way because I could still go back and change a lot of details.
Texturing the Head
The first step was attach both sides of the head and make them seamless. I welded together each point in the middle of the head and smoothed the points together. Then I moved each point around to get rid of the bump that occured between the sides.
Then cylindrical mapping was used in order to attach mapping coordinates to the face.
The mapping coordinates made it possible to unwrap the mesh into a flat texture image file. Now that I had this I could simply paint on top of it and the texture would fit into the correct place.
I took the side and front photographs and merged them together with Photoshop. I tried to erase as much of the shadows as I could from the image. This was because the 3d program creates its own shadows so the two would interfere.
I edited the original texture map and made the bump map from it. I tried to make a continuous rough surface over the face painted area to get a nice texture.
Again I took the original texture map and edited it. This time I made the specular highlight map. The only area I really wanted it to be shiny was right under the eyes so I made this region brighter. I didn't want too much shine because I think that leaves the model looking like plastic. Also I left a small amount of white on the neck area for the skin.
The eyeball texture was taken from a photograph and mapped onto the pupil, iris and white section. The cornea was left as transparent under another transparent layer that covers the entire eye.
This is the finished result of all the maps working together.
Texturing the Body
I took photographs of various clothes and scanned them in at high resoultion. Then I took the texture and edited it into a bump map to show the threads of the fabric. Using cylindrical mapping, I placed the textures on.
For the shoes I took a photograph of an old leather shoe. I changed the color to black and used the same texture as as a specular highlight map. The result was that the shoe seemed scuffed in some areas and shiny in others.
The vest and pants material was taken from the shirt. I changed it to black and upped diffuse level fifty percent so the highlights would show up more easily. I used the same bump map to show the detail in the fabric.
Here is a final render of the textured body.