Now comes the part where I really get to focus on all the small details and nuances of the human face. Because I took the time to focus on the forms and foundation, the details will lay down nicely on the surface and look more accurate. To start, I create a new layer called “eyebrows”, then take an eyebrow stencil I created and softly sculpt it onto the surface.
Then I add some smaller, shorter manually sculpted hairs using the Sculpt brush with the Tubes vdm.
I create a new layer called “skin2” which will be the finer pores and skin texture. Using my selection of high resolution face photographs from www.fineart.sk as well as internet searches of Katharine McPhee herself, I convert those images to black and white and then do a high pass filter on them to eliminate shadow/light information and only leave the actual detail. Then as a stencil, I lay these faces over the head sculpt at it's highest subdivision, and then sculpt in only the flat areas like the cheeks, nose, forehead., neck, and chest.
I also use a general porous skin stencil to “connect” the surfaces and make sure there is texture in all the right places.
To sculpt in the eyelids, I begin with the eyelid stencil I created and rough it in.
This will give me a start to doing more manual sculpting.With the Sculpt brush and Wrinkle vdm, I hone in on the undereye area and the “crows feet” at the outer corners of the eyes. Then I puff them up a little with the Foamy brush.
Then it's just a matter of going around the rest of the face, neck, and chest to put in the wrinkles that bring character to our sculpt. On individual sculpt layers, I put fine wrinkles and imperfections where the face normally moves a lot like the cheeks (smile), the forehead (raising and furrowing eyebrows), and neck (lowering and raising the head).
I try not to put too many strong wrinkles or folds in the female face as it usually makes the character look too old or masculine. When detailing the female face, it's all about subtleties and not overdoing it. Moving onto the lip details, I begin sculpting in vertical wrinkles on the lower lip.
I then do the same with the horizontal lines but I use a wrinkle stencil.
Then I paint a selection on the whole lower lip and chin area and Hide Unselected.
This way I can work on the inner portion of the lower lip without being obstructed by the upper lip. This is so that in animation, if she opens her mouth, she'll have the details on the whole lip.
Then I do the same thing to the upper lip.
I take this same mindset with the eyelids. In our sculpt, her eyes are open but we need them closed so we can detail the upper eyelids in a closed state. This prevents pixels from incorrectly smearing across the eyelid surface (I adjust the uvs accordingly later in the process). I fix the eye geometry by going into 3ds Max and pulling that upper eyelid downward to meet the lower, as if the eyes were closed. Then I send it back to Mudbox.
Since I didn't apply this symmetrically in 3ds Max, I can do it quickly in Mudbox by selecting faces on the side I want to mirror from, click on the arrow under Layers on the right panel, select the option to Mirror and then choose X. It applies the change to the other eyelid to bring me back into symmetry.
With the smear visible on the upper eyelid from our previous sculpting work, I smooth it down to give us a new start. Then I lay down some skin texture again on the correct layer and then add some subtle wrinkles along the eyelid surface.
And finally, I need to add another edgeloop across the eyelid as it's a bit too stretched in the topology to make the right curve around the eyeball. So back in 3ds Max, I make a quick selection across the edges I want to connect and then click on Flow Connect in the Graphite Modeling Tools. This keeps the curve that I've already established instead of keeping the polys flat where I connect across, saving me an extra step.
Sending it back to Mudbox, it automatically prompts for a Transfer Details operation and then I click on Import. If there are any yellow highlighted areas, I go and fix the anomalies and move forward.
To get the details out of Mudbox and use them in another 3D program like 3ds Max, I have to extract a displacement map. This map is the difference between the first level of subdivision and the highest and is exported as a 32 bit .exr file. It uses the UV coordinates we've already created and projects the detail onto the low poly mesh in 3ds Max to match the exact look we have in Mudbox. To begin this process, in Mudbox, I click on the Maps menu, Extract Texture Maps, and click on New Operation or My Extraction Operation if there's been a previous extraction I want to use as a starting point.
From there, I click on Displacement Map and it's rollout pops out. Under Target Model, being sure I've got the head mesh selected, I add the low resolution (level 0). I check on Smooth Target Models and Smooth Target Uvs as that's what seems to work best for my 3ds Max setup. Under Source Models, I Add Selected again and choose the highest level. Below that, I choose Subdivision as the Method, 4096 as the map size, and then I tell Mudbox where to put the map by clicking on the folder icon in the Output options at the bottom. I choose a 32 bit .exr file and then click on Extract at the very bottom of the rollout.
Giving it a few minutes to calculate and save out the map, I then go to Photoshop and open the map to make sure it looks correct.
Next, in 3ds Max, I import the same low resolution mesh from Mudbox, add a Turbosmooth, and a VrayDisplacementMod.
This will be the place for the displacement settings that will determine the look of the rendered mesh. I select 2D Mapping, add the calculated .exr file to the Texmap, set the Amount to 0.394 (a conversion from centimeters to inches – depends on how the scene is set up initially, otherwise use 1.0 for a direct 1:1 scale), and finally the Resolution setting is set to 2048 (4096 would be ideal for matching but I've never noticed a difference) and the Precision raised to 12 for a little higher accuracy. Once that's set, I need lights to even see the results. So I place an overhead Vray light and hit render just to see how the displacement looks.
Here's what the difference looks like with and without the displacement mapping.
After that, I add two more lights, one from the back for some rim light, and then another weaker light from the left for fill. This allows for better overall vision of the model and details.
To add even higher frequency detail, I calculate a bump map in Mudbox. Using the previous extraction operation I used for displacement, I change the Target model to level 4, save out as a 16 bit .tif file, which I then save out as a .png file in Photoshop (I changed my mind).
With the bump map on at a low power, it helps to break up the specular areas and give a little more realism to the render.