• CGSociety :: Tutorial
    17 December 2009, by Jose Alves da Silva 'Zeoyn'

    When I decided to enter the 'Secret Agent' CGChallenge I had no idea that I was buying a ticket for the greatest roller coaster ride of my life. My main goal was to test my skills and hope that some of the 'gurus' could give me advice on how to improve my technique.

    Even though two months sounds like a lot of time, I quickly found I'd have to find a steady rhythm if I wanted to finish on time. My full time job and two kids made sure I couldn't even begin work in this until after 10pm each night. I've slept a lot less than I would have liked. By the deadline, I was exhausted. However, the support of the fellow challengers boosts the adrenaline levels and kept me going. The first thing I did was to read the challenge rules carefully. The most important thing for the judges was 'storytelling'. I knew the image had to accomplish these two goals: scream 'Secret Agent' and tell a great story!

    I explored the character look in Photoshop in the hope that tweeking his physical image would suggest to me a better story. I have made a few sketches and came up with a middle aged secret agent based on clich?s: stylish suit, dark glasses and straight pose, but with a contrasting ridiculous element: the mask on his head that looks like a sock and reveals him as a 'not so' Secret Agent.

    From this moment on, the story line, the defined character, color palette, composition and setting were all developed simultaneously and it is very difficult to separate them, as all of these are inter-connected. I found it easier to create the whole puzzle myself, in fact it was the funniest part of the project. The story was tweaked during the whole process, so there were very few moments where I felt bored (except for the UV's unwraps of course).

    ZBrush has achieved a wonderful user interface that once you get its grasp you have the feeling that the computer is no longer a barrier between you and your creation. You definitely feel like you just have to worry about sculpting, so it is possible to actually sketch and explore ideas interactively in 3D.

    Cinema has filled our heads with references, especially the James Bond series. When we think of a 'Secret Agent', some very clear images pop up. I have collected a bunch of images of some of the most famous agents to use as reference.

    A highly clich?d situation in a Secret Agent film would be facing the threat of a world domination menace. I knew it would be great if I could take these stereotyped elements together and give them a humorous twist.

    While my first ideas were quite different from the final one, I wanted a character which looked quite sharp and elegant, a serious but yet ridiculous agent, definitely the Anti-Hero type.

    I started preparing a low polygon anatomically-correct face in 3ds Max and imported it into ZBrush in order to start experimenting with the proportions of the facial features. Cos this challenge was all about having fun, I got carried away and made a few work flow mistakes. I started the face way too soon, sculpting the mask on the same mesh as the head. At the time I was too worried about solving the tight mask with all its wrinkles and loose tissue on top (that would give that funny sock feeling) and rushed into solving that problem. However, the head topology was not adjusted to the mask and it brought me some problems.

    With a highly detailed head in my hands and everything else to do, I exported the high polygon head back to 3ds Max and started to work on the body.

    I was still looking for the character's proportions, so I created only the parts of his body/suit I needed for his silhouette.

    I modeled base meshes for the coat, shirt collar, trousers, hands and shoes with some extra edge loops around the joints, so that they deformed nicely while posing the mesh. When I was satisfied with his overall proportions I exported the body to ZBrush.

    I still prefer to pose models inside ZBrush, even if it for a still image. When I divide the mesh into different subtools and polygroups, it becomes easy to select/isolate different body parts. So with the transpose tool, the posing is faster than having to rig the whole mesh.

    The life of a Secret Agent, must have its ups and downs. I portray a bad moment in which the agent is surprised while on a mission to unveil a world domination plot. I thought it would be funny if the antagonist of our highly trained agent was an apparently harmless small creature: a mouse.

    However, the story could get better if things were not so 'black and white' or 'good vs evil'. It is also true that mice have been victims of human exploitation through animal research. What if those mice were freeing themselves from captivity and just trying to get their destiny back in their hands?

    It was time for the character to start acting. His pose had to transmit that he was surprised by a mouse while in the act of spying. From this moment, I know the result would have been a lot poorer without the feedback from CGSociety's community. I have received a lot of great advice and most of the suggestions were really helpful. I am very thankful for that!

    After some great advice I have settled on a body pose that could work. The agent was leaning his back against a wall, tucking the jaw inward, with an arm in a protective position, trying to keep his feet off the the floor and with both knees turning inside. If this pose was working I had to test it in a composition and from the final point of view.

    The posed mesh, still in low poly, was exported to 3ds Max. 3ds Max 2010 has the ability to preview shadows and screen space occlusion in real time (in case you have a Pixel Shader 3 compatible graphics card). This allowed me to study the whole composition, basic lighting and shadow placement in real time.

    I have modeled very basic meshes representing a mouse, a flashlight and some boxes as placeholders for props. I started experimenting with different camera angles, placement of the characters, props location, light direction and color.

    I ended up with two elements standing out: the agent and the world domination plan. I have tried to follow the 'thirds rule', placing the agent at the distance of one third from the right image border, with his head and arm also at the distance of one third from the top of the image, and the world domination plan and mouse leader were placed at one third from the left side.

    I also tried to create contrast in both these areas. Another composition element that I emphasized was the diagonal starting on the lower left and ending on the upper right. You can notice that the light cone of the flashlight, the pointer of the mouse leader, the agent's front arm and its shadow are all lined up according to this diagonal.

    Regarding the color, I have decided that there should exist two distinct areas: a saturated red office area, where the mice are meeting, and which reinforces the evilness of the mice plans, and a dark blue area outside the office which emphasized the undercover role of the agent, only broken by the revealing warm light of the flashlight.

    Go to page 2

  • The agent had yet to be finished. In 3ds Max I added a few more edge loops to the mesh in order to prepare it to sculpt the folds and wrinkles on the suit. I have also modeled the shirt, the soles of the shoes, the thickness of the jacket and added a tie. You might have noticed that I had initially planned to use a bow tie, however the long tie enhanced the fluidity of the movement, so I kept it. I have also created the UVs for the body at this stage while the polygon count was low and it was still easy to tweak the UV Layout.

    Back in ZBrush, the final detailing of the agent started and I worked on his face expression first. This part took me quite some time, but I needed to get it right. With a lot of help from the community, the agent ended up with a funny, surprised look.

    Then it was time to work on the clothes. I have picked up a Flatten Brush with a little height (change BrushMod to a positive value of 20) and tried to mark radiating lines originating in the parts of the cloth where there was more tension (for example the knees and the elbows) as well as adding volume and wrinkles to the more relaxed areas of the cloth (usually on the opposite side of the tension areas). A few photos of people wearing suits also helped me place the folds.

    Then I painted the hands and head inside ZBrush. It is a true pleasure to paint directly on the model. The head had a lot of geometry so it held in its vertexes without losing a lot of detail. After finishing polypainting I have converted the colors stored in the vertexes into a texture (Tools>Texture>Colors to Texture).
    I also painted the specular map in ZBrush using the same technique.

    I wanted the agent model to have a lot of detail but without displacement maps as they can be slower to render than using the actual detailed geometry. The model already had a few million polygons and a lot of mice were expected to fill the scene, so I have decided to use Pixologic's Decimation Master.

    This free plugin for ZBrush, that you can download in Pixologic's web site, allows to reduce the polygon count attempting to preserve the surface detail and UVs. The agent model was reduced from 3.2million polygons to 150,000 polygons without any apparent loss of detail.

    Back in 3ds Max, I have imported the decimated geometry and began with the first render tests. My background is Architectural Visualization and I am very used to render with V-Ray, so this was the obvious choice for me.

    In this project, my approach to light was totally unrealistic. My main concern was not that the lights behaved in a physically correct way, but to put the focus on the right things.

    To create the mouse crowd, I model a single low polygon mouse in a T-Pose inside 3ds Max and applied the UV coordinates to it. Then I exported to ZBrush where the proportions were corrected and the mesh detailed.

    I sculpt the flow of the fur on the mesh as I was sure that I wouldn't have the time to render 3D fur. Once again I have painted the mouse model inside ZBrush and projected that information into a texture.

    In order to achieve variety in the crowd, I have defined five poses which would then be rotated, scaled and mirrored as well as receiving tails with different positions. The five key poses were: the mouse leader, the flashlight carrier, a pointing mouse, a mouse looking up and a mouse looking down.

    All the mice shared the same materials. Thankfully laboratory white mice are all very similar to each other.
    I have used inclusions and exclusions to keep the lights from affecting objects in a distracting way, and certain lights did not cast shadows or affect GI, which was the case of the rim lights on the agent and mice.

    I have followed the same unrealistic approach regarding the reflections. Huge light boxes with self illuminated materials, but without generating GI, were created outside the visible area of the camera in order to produce nice white, blue and red colored reflections in certain parts of the image.

    Most of the materials applied to the agent used the standard 'V-RayMtl' shader, with the exception of the skin for which the 'V-Ray FastSSS2' shader was used.

    I have used 'V-Ray Fast SSS2' shader for the mice bodies. The subsurface scattering effect is very noticeable in the ears and paws of the mice.

    For the eyes, it was very important that they glowed menacingly in the dark and for that I have created a 'V-Ray Blend Mtl' material which mixed a sort of red glass with a self illuminated falloff and a yellowish specular and it turned out quite well.

    The mice looked different when placed in different parts of the scene. The ones in the red office look white and cute, while the ones in the corridor look dark and with red glowing eyes gaining an aura of evilness.

    It was important the mice could be seen as the aggressors or the victims.

    previousGo to page 2

  • I love filling the images with secondary stories that the viewer will only find if he spends some time looking at the picture.

    The Mickey Mouse ears were added to reinforce the brilliance of the agent's undercover skills. I made sure the ears were clear in the agent's shadow on the wall.

    The love making mice inside the locker was a funny idea which had a strong connection to the world domination plan. It was also a challenge to transmit this idea only through their tails, but I think it worked out really well as it is not something explicit.

    A sort of batman symbol projection. The drawing board was one important element in the story and I have tried not to use any written words in the world domination plan scheme. The chain smoking mouse is just a funny character.

    Now it was time to render. I have only used four render passes: the main render, rim lights, occlusion and masks. Even though V-Ray is a great rendering engine, I still prefer mental ray's ambient occlusion, so this pass was rendered with mental ray. The pass rez was 6000x3600 pixels.

    This project was mainly 3D, but there are things that are much faster in 2D. At the CGSociety forum, a lot of people had already asked me if I was going to apply fur to the mice. As it was impossible to do it in time in 3D I have risked doing it in Photoshop. I have created a furry brush and started painting on top of the 3D always picking the colors from the base image and trying to keep the rim lights effect. It took some time but I guess it paid off.

    The veins seen in the mice ears were painted, which allowed me to create some variety and avoid a repeated texture from mouse to mouse. The smoke, glows, the volume light coming out of the flashlight and the dirt marks on the walls were also painted in Photoshop. I have, as a final gift, given the brushes I created for the image as a download.

    To finalize I have overlaid a bit of texture and film grain to the image to give it a more natural look.


    Jose Alves da Silva 'Zeoyn' has been working in the computer graphics field since 1992. Most of those years were dedicated to Architectural Visualization, but only because I have a degree in Architecture. My real passion is working in illustration and character creation. My CGPortfolio is dedicated to that passion, which allows me use all my creativity.

    Winning the First Prize Master Award on CGSociety's Challenge XXIV Secret Agent is the main highlight in my career. I believe that continuous learning and improvement is the key to personal evolution, as well as the key to survival in today's industry.

    I really feel that this challenge has forced me to go beyond my limits. I can not stress enough the importance of working side by side with so many talented artists. I will keep this challenge as a great experience and recommend everyone to join this roller coaster ride. Thank you for all the support and advice and I hope to see you in future challenges!

    Jose Alves da Silva 'Zeoyn'
    'Secret Agent' CGChallenge Winners page
    Autodesk 3ds Max
    Pixologic ZBrush

    Discuss this on CGTalk


    previousMore Articles

blog comments powered by Disqus