Making Of 'Chesney Baker Caricature'
Hey all, in this tutorial I'm going to show you how I created the final render of my Chesney Baker caricature. First thing, I got the inspiration to do this piece after looking through "Stars," a book of caricatures by the very talented Sebastian Kruger; you should definitely check it out (www.krugerstars.com).This is not my original concept. It is based on one of his drawings.
This is going to be more of a demonstration on my general approach and workflow. All of the actual sculpting was done using Skymatter's awesome new modeling program, Mudbox (www.mudbox3d.com). Since this program is still in beta testing, I can only show so much. Also, when I work, I'm not trying to re-invent the wheel, I use what works for me. My workflow is setup to spend as much of my time sculpting the model. When it comes to rendering, I think it is most effective to render out passes separately and composite them in post. Ultimately, there are many ways to approach your work; I'm not saying that mine is the approach that you should take; it's just what works for me. Try a bunch of options and find what you are comfortable with and what works for you.
Okay let's get started. So going into this project I knew exactly what I wanted as an end result which was a front view render, not photoreal, more of a casted maquette look, kind of like a bust. Knowing this allowed me to focus exactly on what was going to be seen and saved a lot of time in the end. I think that it is extremely important to know what you are trying to accomplish ahead of time. It really allows you to focus on the end result and spend the greater part of you time specifically on what is necessary to reach that final look.
I started by creating a low resolution base mesh in Maya based on Kruger's Drawing and just roughed out the exaggerated proportions. I wasn't focusing on anything other then the major structure and silhouette. The exact positions of the eyes, lips, nose and such would be adjusted later in Mudbox.
The clothing and hair were also done first in Maya as separate meshes. I'm like box modeling so it wasn't hard keeping this model all quads for the Mudbox import.
The final sculpt which includes all of the finer details such as the pores, fine wrinkles, and eyebrows was done using pre-made alphas and a lot of hand sculpting. The final model was subdivided to about 3.5 million polygons. I added smaller details like the carunculas and little individual eyebrow hairs lastly in Maya.
At this point I exported out my displacement map from Mudbox to use in my shader network in Maya. I learned how to render displacements from various Gnomon DVD's, friends, and from one excellent tutorial in particular written by Scott Spencer, called "Zbrush to Maya Pipeline Guide"
Definitely check it out. He explains how to render out displacement maps in Mental Ray in great detail, so I'm not going to go over all of that here. These are settings that I used to export my map out of Mudbox. I tweaked the levels of the map so that it is easier to see.
Now since I was only going to be rendering out one view, render time wasn't exactly a concern, though it should always be in the back of your mind when working. In any case, I like to work from a level where the mesh that you apply the map to is of a high enough resolution so that it doesn't have to be displaced too far to reach the final look. For that reason, I decided to work from the highest level that Maya would allow me to import, this turned out to be about to be the level 2 mesh at about 200,000 polys.
Since there is no camera move or change in our light direction, we can reach the same result in a fraction of the time by using a bump map instead of a displacement map. Sometimes I use both, but for this specific single view render, the bump map alone will work fine. This will not work in all cases but since the Level 2 head includes the majority of the larger wrinkles, the only thing we need is the fine skin detail which the bump map will illustrate. Here's an image of an early test I did to realize this. As you can see here, there is very little difference between the two other then render times.
Now that we have the render matching the original sculpt, let's move onto the shader network and lighting setup. I am using a modified shading network that a buddy of mine gave me back when I was in school. I had to alter it a bit to get more of the brushed plaster cast look that I was going for.
The eye texture was created from scratch using the tutorial, "Realistic Eye Texture Painting" by the very talented Krishnamurti Martins Costa that was in fact written for 3D Total, you may know him online as Antropus.
For my lighting setup, I did a few early tests and since we're not going for photo real, I decided to go with good old Maya software and GI_Joe. You can get this lighting script from highend3d.com. Here are the settings I used.
So first I got my basic color pass with only the light dome and no additional lighting. I then rendered out a separate pass with an additional key light and a subtle rim light so I could control that in the comp. I also rendered out an occlusion pass using Maya's preset render layer at 64 samples. Lastly, I rendered the eyes also in a separate pass and used applied the "use background shader" to mask out the head so that I could independently control the eyes in Photoshop. Here are all of the passes.
Now we move onto compositing the image in Photoshop (I'm assuming you have basic knowledge of the program). I think its best to work with as few variables at a time, similar to lighting, less is better at first. Get as few layers looking the way you like, then introduce more and more passes. So the first thing I did was work on the color passes by correcting the levels. I then took those two layers and blended them via Saturation mode. Here we are at this point, starting simple.
I then desaturated the "CLR_Pass" layer a little to get the colors muted down a bit more. Next I turned on the "OCC_Pass," set its blend mode to Multiple and decreased its opacity. After that I added layer masks to all the layers to remove the black outlines. Then added a gradient to the background and tweaked the overall hue/saturation and brightness/contrast of the two color passes. This is where we are now.
After that I merged the two color channels, duplicated it, desaturated that new layer, sharpened it, and decreased the opacity. Here we are now.
Now its time to add the "Eyes" layer which were originally blue, so I tweaked the hue to match the sepia tone I was going for. At this point I separated the eyes into two new layers to allow independent control of each. I also adjusted their saturation and found that I had to put the eye layers underneath the occlusion pass to dirty them up a little.
Lastly I took all of the layers and did more adjusting. I did some more color correction, duplication, level tweaking, sharpened up areas like the eyebrows and added a little dodge and burn around the eyes to bring out the highlights a bit more. I also re-imported the occlusion pass and darkened up some areas slightly more. And that about does it. Here's the final comp again.
I hope you found this tutorial helpful and can take away something useful from it. It was a lot of fun creating and I'm always up for giving back to our very open 3D community. Thanks to everyone for taking the time to read my approach and a big thank you to all of artists out there who have shared and continue to share their ideas and workflow which help make us all better artists. Until next time,