Project Overview: Dramatic

Eduardo LÃ'bo shares how he created this striking image and perfected the art of modeling realistic-looking draped material.

Introduction

Hello everybody, my name is Eduardo LÃ'bo, but people here refer to me as Ellobo. I am so glad to have had the chance to write this project overview for you guys. I have been studying 3D really hard since January 2012 (I had to do this because before I didn't paid attention to anatomy and all the foundation studies related to 3D, but now I know that this makes all difference and the progress can be seen on each model).

I am from Recife, Pernambuco, which is in Brazil. There are not so many 3D-related courses here as there are in foreign countries and so that makes me a 3D artist generalist. I spend my free time studying anatomy, rendering, and complex and quick modeling techniques aimed at always making the modeling process faster each time, and the results better.

What I am going to show you here is my approach to anatomical sculpture and the synthesis of an analytical study of the relationship between drapery and topology (something I find really awesome and I hope you like. It can be replicated in any software).

The modeling process was all inside the ZBrush, but the drapery was a mix of simulation in Maya and ZBrush refinement. (Again, the simulation process can be done in any software).

The Pose and Anatomical Approach

I started this scene marking out the pose with ZSpheres. Then I generated the skin using Adaptive Skin while keeping the subdivision density as lower as possible. Starting a model with a lower topology makes it easier to control the gesture and pose of your model and all the silhouettes (Fig.01).

Fig. 01

Fig. 01

Fig.02 shows the second level of the model where the landmarks have all been set, such as the visible bones and main muscle groups. I worked from top to bottom.

Fig. 02

Fig. 02

I used the Clay and Clay Tubes brushes, along with the Standard brush with Alpha 43, alternating back and forth with the Smooth brush throughout the modeling process (Fig.03).

Fig. 03

Fig. 03

On the next subdivision level, the process remained the same, but as I started to get more polys I had to refine the model. The secret is to never model just one area. You must pass through the entire model, then go back to the same area on the next subdivision and start the refinement all over again.

Fig.04 shows the fourth level of the figure. Now the polygon count was allowing me to add some more details, like on the feet and hands. The gesture of my model was little off at this stage, but I knew I could work on it later.

Fig. 04

Fig. 04

For me, the secret to a realistic hand is to mimic the interossous muscles of the hand (the palmar and dorsal interossal muscles). These muscles are those that are between your fingers, side by side (on both the palmar and dorsal sides). They create a very peculiar attachment relationship with the skin, which is crucial, and is shaped by the angle of the pose of the hand, because some skin tissue overlaps others finger by finger. You can check that out further by looking at an anatomical book or your own hand. The same works for the feet.

Fig.05 - 06 show the results of the modeling process with the gesture worked on a little bit more at level six subdivision (the modeling process was still the same, even using the same brushes defined at the start of this model). What made the modeling stay so clean was the way I added details and smoothed them with the Smooth brush.

Fig. 05

Fig. 05

Fig. 06

Fig. 06

Fig.07 - 08 show the completed anatomy work on the pose (all done without using symmetry). Modeling without symmetry is a very good way to develop and test your memory knowledge, because after consulting everything while you model on one side, you can try to replicate that on the other side without the reference (but always check at the end to see if something is missing!)

Fig.07

Fig.07

Fig.08

Fig.08

The Drapery

Now it was time to build the drapery. As you know, you must study a lot to accomplish a good draping effect, but when you master this technique you can take something rough (but simulated correctly) and then apply all these artistic aspects to that simulation to get another level of model (Fig.09).

Fig.09

Fig.09

I had been looking at the drapery work of other artists a lot and what I've realized is that it is very noticeable when you just sculpt something (even if it's pretty much very realistic), because whether you want it or not, you end up with an artistic study point of view of the drapery. Otherwise, when you just simulate, it gets too loose and never reaches the objective. So the solution I found is to combine them, and in all my studies here the key point is just the topology. The topology is what changes the look and the capacity of your simulation (it is what I achieved in here by making some tests).

So I will explain the process of the simulation inside Maya. It can be achieved in any other software but, of course, the workflow will vary because each one has its own flow of thinking (Fig.10).

Fig. 10

Fig. 10

First, I imported my scene (very simple). Then I scaled it to real size, so that gravity would work to realistic parameters and the other fields could be changed easily (Fig.11).

Fig. 11

Fig. 11

I then created my drapery as it would be in real life, with a simple plane (because it was a simple kind of thick bed sheet). I added another plane on the bottom that wasn't subdivided, to be my ground (Fig.12).

Fig. 12

Fig. 12

After that I selected all the edges of the drapery and create a Quick Select (because I would need to select them later after triangulating the plane (Fig.13).

Fig. 13

Fig. 13

After triangulating the drapery, I needed to select those edges again (that is why I created that Quick Selection)(Fig.14).

Fig. 14

Fig. 14

When I had reselected those edges, I could now delete them, changing in this way the direction of the topology. As I have said before: drapery behavior will depend on the flow and type of topology, beyond the simulation settings of the drapery, of course (Fig.15).

Fig. 15

Fig. 15

Now I triangulated it again, because this is what gave the drapery that flickering effect. I triangulated because if I had simulated it was just quad topology, the result would have been too smooth and boring, like a regular piece of drapery (Fig.16).

Fig. 16

Fig. 16

I then created the cloth in Maya (Fig.17).

Then I simulated it. You can see how my topology worked very nicely to create the behavior I wanted for this thick sheet. (I simulated with a low poly count, but I will reveal soon how to make a high subdivision simulation just after "posing" drapery. Because you know it gets too heavy to process the simulation and still try to control some constraints in real time)(Fig.18).

Fig. 18

Fig. 18

To control and pose my sheet, I created some constraints. A tip for Maya users is to create them in the initial state of the animation and then play the animation using the Interactive Solver PlayBack to change the position of the drapery in real time (Fig.19 - 20).

Fig. 19

Fig. 19

Fig. 20

Fig. 20

The Last and Most important Tip

You might be asking yourself: how can I subdivide my model after it had been triangulated and I've posed the drapery? The answer is simple. There are two ways:

First Technique

After you have created your drapery (no matter what form it has), you must first be aware that it only has quad topology, so you can simulate it at a low poly count and very quickly.

After posing and getting all the constraints in position, stop simulation and just subdivide it by going to Polygons > Mesh > Smooth (in Maya) and achieving your high subdivision.

Then triangulate and play the simulation again to the new poly count entered in the simulation. (Once the model is converted to Ncloth in Maya, you can still change the topology at any time on the 3D model).

Second Technique

This second way is more complicated, but is for more complex forms and topology. If your topology has a different flow, and can't be achieved by just deleting the edges, what you can do is the same process as the first way, but instead of triangulating, export the mesh to ZBrush and use the Micro Mesh tool to swap the topology of each quad very quickly.

Then import it back to Maya, convert to NCloth again and just play the simulation. But be aware that with this second technique, as you import a high poly count model to Maya, the software will need a lot of memory to convert the high poly model to an NCloth simulated model (Fig.21 - 22).

Fig. 21

Fig. 21

Fig. 22

Fig. 22

Here's my final image (Fig.23).

Fig. 23

Fig. 23

Fetching comments...

Post a comment