Making of Javier Bardem

Dmitrij Leppée shows us how he made this fantastic sculpt of the talented No Country for Old Men actor Javier Bardem

In this tutorial I will explain how this impressionistic portrait of Javier Bardem came into being by using ZBrush. It was created in the course of three to four hours, and I hope that it will present some out of the box approaches and techniques that you can implement in your own work, and that will inspire and make you feel fresh and fulfilled when working on both personal and professional projects. As my workflow changes with each new sculpt, sometimes more sometimes less and not always for the better, I give you a breakdown of one of many ever-changing workflows used for this specific portrait.

Starting point and references

I started in ZBrush from a DynaMesh sphere as a completely made up head. Along the way it started to resemble the actor, so I found a couple of images online and finished him by observing them in an image viewer opened next to ZBrush. Keeping the references to a minimum, because various camera angles and lenses can cause considerable confusion with the proportions of the person. I chose two main references: a front image and a full profile photo, and two to three of a 3/4 view, an angle from above and an angle from below.

Gathering references online, and by taking screenshots from movies

Gathering references online, and by taking screenshots from movies

Sculpting in ZBrush

When sculpting portraits, I tend to start from scratch each time, as I feel constrained if I use a low poly BaseMesh as a starting point. I find that when I start from a DynaMesh sphere my hand flows freely, and there is no topology restriction. On lower DynaMesh resolutions I concentrate on the major facial structures using Clay Buildup and Move brushes, defining underlying structure and gradually increasing DynaMesh resolution as the sculpt progresses. Once I arrive at the stage when the details are in order, I turn off DynaMesh mode and start subdividing the geometry for enough resolution to start sculpting them.

Progress from a DynaMesh sphere to finalized portrait, with brushes used and their setup

Progress from a DynaMesh sphere to finalized portrait, with brushes used and their setup

Wrinkles and pores

For wrinkles I use a combination of Damien Standard brush (for grooves), Inflat brush (to bump up the in-between grooves) and Standard brush (for a variety of expression wrinkles). When I create speed sculpt portraits I avoid careful placement of skin details, instead I try to hit the overall notes. For pores I use Standard brush again with a default alpha 07 set to spray, and curve of the brush tightened for harder dents. Retrospectively, if I find that the underlying structure is off, I go a couple of Subdivisions down and using the Inflate brush to bump it up and smooth it, that way the finer details don't get affected on higher Subdivision levels.

Close-up of face details

Close-up of face details

Sculpting hair

I masked a portion of the head from which the hair would grow and on the lowest Subdivision level, extracted it into a new SubTool. I don't extract from higher levels because the form still needs to be adjusted and it is far more manageable that way, so I subdivide it later. Then I inflate it to give it some volume, and use the Clay Buildup brush with Backfaces locked for sculpting general form, and lastly Damien Standard brush to carve in strand details. As always smooth where there is excess.

Extracting a base for the hair from the head SubTool

Extracting a base for the hair from the head SubTool

Loose hairs

I created a simple oblong Insert Multi Mesh brush with three Polygroups for loose hairs (one for starting point, one for ending point, and one for in-between which can extend seamlessly per need), enabled it's curve mode and weld points, with which I can then drag hairs directly on the hair SubTool and later on pull them out with move brush and smooth them gently from root to tip. Then I group and split them from the main hair SubTool so I can independently subdivide and have more control over them.

Custom Insert Multi Mesh brush (IMM) for loose hair strands

Custom Insert Multi Mesh brush (IMM) for loose hair strands

Making eyebrows

For the eyebrows I used Fibermesh by first masking the portion from which they will be generated, then adjusting the density, length and making root size a bit thicker than the tip. Then I made a couple of render tests until I was satisfied with the result, and once I am, I press "Accept" to generate editable Fibermesh. For all short hairs I only use the GroomHairShort brush with tweaked settings of Front collision tolerance (under Brush>Fibermesh menu), which I lower to about 6, making the strands cling closer to the underlying mesh when grooming.

Generating eyebrows by using Fibermesh

Generating eyebrows by using Fibermesh

Applying materials and rendering different light positions

I kept it simple with only Basic material for everything except the eyes, which I applied Toy Plastic material. I used one light with intensity set up to 1.2 for front light renders, and around 2.0 for back light renders, that way the rim light is harder. I rendered out a couple of images with different light positions so that afterwards there are enough options for testing in Photoshop when deciding about the final look.

Renders with different light positions

Renders with different light positions

Rendering various passes for the final image

When rendering portrait speed sculpts with BPR in ZBrush, I found that a black and white image is far more powerful than the colored, because of the energy in the strokes and the shadows they cast upon rendering, so I don't even bother with painting color texture except for some darker grays around the lips, eyes and generally the areas that would be affected by ambient occlusion. I usually render these passes: Ambient occlusion, Shadow, Specular (with reflected map material over all SubTools), SSS, Depth, Mask and a variety of renders with different light positions (step 7) to experiment with later on in Photoshop.

Render passes for compositing the final image in Photoshop

Render passes for compositing the final image in Photoshop

Final image composition

Once I have enough render passes I can use, I open them in Photoshop and try them out as I go, slowly getting the idea for the final look. Layers of renders with different light positions on top of each other with either opacity tuned down, or blend mode changed to screen, and on top of them specular pass, SSS, AO and Shadow. On top of that I make final touches with sharpen filter to bring out the details even more, background separation and a bit of contrast using a duplicate of the merged layers image with an overlay blend mode. As the last step I add a depth alpha to the final merged image and apply a lens blur to it.

<h5>Layers in Photoshop, settings and blend modes</h5>

Layers in Photoshop, settings and blend modes

<h5>The final image</h5>

The final image

Related links

Check out more of Dmitrij's work at his website
Alessandro Baldasseroni chatted to us about his career as a character modeler
For more character creations take a look at our ZBrush Characters and Creatures

Fetching comments...

Post a comment