Recreating Barbossa in 3D: Modeling and Texturing Workflows

Recreating Barbossa in 3D: Modeling and Texturing Workflows

Lukas Kutschera did a breakdown of Barbossa made in ZBrush, Maya, Marvelous Designer, and Substance Painter during the course Character Creation for Film/Cinematics at CGMA.


I’m Lukas, a self-taught 3D Artist from Germany, working in a small VFX studio as an intern. I got into 3D randomly, 1.5 years ago playing around with Maya. That was kinda fun, but once I tried out ZBrush for the first time, it really blew my mind and that led me to my interest in character art. For now, my goal is to work as a (character) modeler on big movies/TV shows.

1 of 3

Barbossa: Idea

When I skimmed through a lot of Pirates of the Caribbean images, I really liked the composition of the one with Barbossa and thought I could reproduce that in 3D. I also noticed that a lot of character artwork where the main character interacts with another character or a special object/environment tends to be really popular, so having Jack the monkey on the shoulder seemed great.

Sculpting the Face


For likeness, I like to collect as much reference as possible from different angles with different lighting. I know some people believe it is good to limit the amount of reference, but I think the more, the better. Every picture will give a new insight. If I think I see a mistake in my current sculpt, I will look at a couple more pictures to verify it, because different focal length and lighting can change the look quite drastically. Likeness sculpting can be a real time sink so I try to make some changes quickly and then move on. When I get back to it, it is easier to see mistakes right away. If there is a 'secret' to getting a true likeness, I have yet to find it and I don’t think I achieved likeness with this one at all. But the more likeness attempts I do, the more I come to believe that just getting more familiar with different faces and facial anatomy slowly leads to better results.

Skin Detailing

In terms of brushes, I use Clay Buildup, Move (topological), and Dam Standard for the most part. For the skin detailing, I did quite a bit of manual sculpting.

For the micro-detail, I imported the red, green, and blue channels of the Texturing XYZ displacement maps that I projected in Mudbox (you can find a similar workflow here). They give different scales of detail and it is hard to tell how intense these Layers need to be until you do test renders.

I also used some pore alphas to enhance certain areas.

All the detail was stored in Layers in ZBrush, so I was able to adjust the intensity of individual passes to get the desired result in Arnold. The detail can look too intense in ZBrush, but since the SSS takes the edge off, it might be what is needed.

Clothes and Accessories

After blocking out the body in pose – because I knew the result I was going for, – I brought it into Marvelous Designer and blocked out all the clothing there. There were no sophisticated garments there, really basic stuff as you can see. I'll go over a couple of important settings that I feel are most relevant:

The pistol is a mix between polymodeling with ZModeler, a bit of sculpting on the wood, and using the 'make 3d' option to convert alphas of ornaments into geometry that I then retopologized. The necklace is pretty much sculpted and the ring has a sculpted bear head and a polymodeled ring. I also used, as I did with other ornaments, the 'frame mesh' option + an insert curve brush to insert, for example, a braided pattern around the ring or a seam along the jacket.

For the stitching pattern on Barbossa's shirt, I used a pattern blueprint, built curves in Maya, and extruded a braid along these curves.


I used Dynamesh to block Jack out and then used ZRemesher and polygroups to retopologize. The clothing is done in Marvelous Designer, the same as Barbossa's clothing.

Hair and Fur

Barbossa's hair, eyebrows, and beard are done by exporting a small percentage of fibermesh as curves and then assigning a Maya nHair system to it, which allows you to basically treat every curve as a guide and gives you a bunch of settings such as the amount of hair, profile, thickness, noise, curl, etc. In that way, it is like XGen, but it has very reliable, easily modifiable results and it is a cool mix between 100% manual work and just letting some guides do the work. I think It is great for certain hairstyles and it gives the artist more control over the hair.

The monkey is done with XGen interactive groom. I painted a color map for the fur in Substance Painter. There are also a bunch of interactive groom descriptions for fuzz on clothing which helped the realism by breaking up the edges.


I used the Texturing XYZ albedo map and applied it in Mudbox, then went over to Substance Painter to use that as a base to start my texturing process. For the color palette, I just made sure to give it some interesting variation that made sense story-wise. Hue, saturation, and value were adjusted in Maya to make it work with the lighting setup. For the clothing, I did add some bright microfibers in Substance Painter but I also used XGen interactive groom to create more fuzz, especially along the edges.

For basically all my texturing, I have the same approach which gives pretty reliable results:

1. Have a good high poly sculpt with lots of detail sculpted in, it gives Substance Painter the chance to bake useful curvature maps that can be used with smart masks and materials.

2. Obviously be aware of the storytelling aspect, so the textures should make sense and support the composition you're trying to create.

3. Establish the main color and metalness and roughness information to determine the material. You can also use smart material and imported textures as a base to get a lot of information quickly. You can then remove and dial down what you don’t need.

4. Get some larger-scale color variation that makes the texture look more interesting from a distance. That includes hue, saturation, and value, although I don’t think it's good to push all three of those, because it gets too intense and irritating. Color gradients can also look pleasing (dirt gradients that make sense, too). In general, I think color is not terribly important if the character looks good, although it can be a powerful tool.

5. Get a bit of harsher, higher contrast detail (it could be bright chipped wood, some stronger discoloration on metal, or simply a harsh dirt spot – just something that stands out and gives more interest, while still being relatively subtle).

6. I have a bit of roughness variation in most layers, but a lot of it will be added with dust, dirt, and moisture, using curvature and AO maps to drive smart masks. These always require some breakup with additional procedural maps and hand-painting, else it will look very uniform. I usually don’t use much height information because I prefer to do it in ZBrush with the high poly sculpt and control micro-level stuff in Maya.

8. I like to add a sharpen filter on lower intensity because it helps more information come through. When rendering, a lot will get lost anyway, so this can really help. There are more filters like HSV and contrast that can be interesting but I prefer to have control in Maya, so I don’t have to go back and forth between programs.

Of course, this is not a perfect formula but I find that I can go through these steps quickly and get at least a good base.

Stitching is kept as low poly geometry for extra realism, although it wouldn’t look terrible if done in Substance Painter. All the fabric height information is done with tileable maps in Maya for higher quality results.


I hope the layer names make sense, but I'll give you a quick written breakdown, too.

I start by importing the XYZ albedo, add a layer in Passthrough mode and HSL filter to make the skin paler, and a contrast filter to bring out some detail. Similar to the approach I described earlier, I use larger color variation next, giving the beard area a blueish tone, nose and cheeks – red, where more blood flow occurs, and the forehead – a bit yellowish because the skin is closer to the bone if I am not mistaken.

Next, I add red, white, and brown spots that are characteristic for skin, especially if exposed to the sun, the way the pirates' skin should be. I also use some alphas for red veins and duplicate that layer with a blur to have it blend in better. The lips are basically some color variation. For the eye area, there is a mix of colors, typical for older people, which you can see in reference. It's a mix of blue, red and even yellow. There are more similar passes next, again some white spots that break up the skin pretty well and look realistic (there are a bunch of procedural maps that can work, from dirt spots to bnw spots; like with everything, experimentation is key I believe).

Finally, I have a scar. Scars can have different colors, but I decided to go with a more pinkish one that got white in the center. Like I said in the beginning, the final touches take place in Maya with HSL adjustment and level nodes because the lighting dramatically impacts the render.


I rendered with Arnold in Maya. For the lighting, I usually use a simple setup with one Key light + either a Fill or HDRI and a Rim light to help separate the character from the background and get a more dramatic look. Towards the end of the project, I was watching Pirates of the Caribbean, and I made screenshots of lighting setups that I liked and tried them out. My favorite render is actually the side view which was inspired by one of the screenshots I took.

My initial light setup for the first render got more complex because I wanted to have a burning fuse and I had to use lightblockers to limit the illumination to certain areas. I also used lightblockers to prevent the strong rim light from blowing out the feathers and hair which happens quickly and looks awful.

Finally, I did some compositing in Photoshop. I like to render my lights individually, so I can play around with the intensity of Key, Fill, and Rim lights. The Z depth pass was useful to blur the cannon in the foreground. I didn't find much use for the AO pass lately but the specular pass can be nice to give specular areas like jewelry or even skin more ‘pop’.

I also added in some smoke, vignette, and fuse flare for one render. And I think adding the ship parts in the background in my main render worked well to give it a more cinematic feel.

Last but not least, I made some level and other small adjustments and experimented with the Camera Raw filter in Photoshop to see how different color schemes look. I took Pete Zoppi’s approach (Pete Zoppi is an instructor at CGMA) with a more desaturated color palette for my main render to make it more realistic and even added a black and white render, which I thought looked interesting.

For the side view, I picked a slightly golden light that gives that epic feel. I was inspired by the film 300 where it is heavily used with great effect. 


It was my first time using Texturing XYZ data to create realistic skin and it was my first time using Mudbox as well, so that was certainly one of the most interesting aspects of the project. Overall, the course gave tons of great tricks to help achieve more realism, from modeling to texturing and even lighting.

I have to say that by far, the biggest lesson was to see Pete Zoppi's attention to detail and his flawless execution of every stage which shows in the super high-quality look of his characters, even in extreme close-ups.  It will really make me reconsider ever rushing any phase in my project again.

The 10 weeks were definitely intense, but before taking the course I was really frustrated with my lack of progress, so I dedicated as much time as possible to this project, hoping it would make a good portfolio piece. I got very good feedback on it so far and I really have to thank Pete Zoppi and CGMA for that!


Working with XGen was the main challenge for me as there were bugs or mistakes on my end probably that made me lose progress. Grooming was also the most time-consuming part. But I tried again and again because there was a lot of new information and I believe even if something works, it can always get better. The chances of getting the best result on the first try are nonexistent. In my opinion, experimentation might be one of the most important things to consider in any project, although in reality, time constraints usually prevent it, unfortunately.

Lukas Kutschera, 3D Artist

Interview conducted by Arti Sergeev

Keep reading

You may find this article interesting

Join discussion

Comments 0

    You might also like

    We need your consent

    We use cookies on this website to make your browsing experience better. By using the site you agree to our use of cookies.Learn more