A Tree Grows in Old Mexico

The metaphor of a single seed from which grows the mighty oak, symbolizes the humble beginnings and 100 year rise to prominence of the international business conglomerate Grupo Salinas in a recently completed :30 for Filmmates Mexico City.

Director Manuel Bierjerman envisioned a time-lapse progression from a vintage, undeveloped landscape, to a fully formed modern Mexican city park spanning 100 years of history. With an ethereal feel and a palette inspired by the famous Mexican artist Doc Atl, the spot is set in motion by a young girl nurturing a seedling. The sapling grows and a city develops… Driven by the economic opportunity created by the young company, then known as Salinas y Rocha. As the evolution brings us to the present day, the girl again appears at the site of the now towering oak, this time as the elderly matriarch. Along with her, generations of offspring symbolize the growing family of companies under the Grupo Salinas brand.

Principle photography of the little girl, the family and actors portraying company employees, was filmed in Mexico City in green screen under the supervision of our Animation Director Troy Davis. Aside from these few live-action elements, the entire spot was realized by Geomedia artists completely in CG. As if the prospect of animating a growing tree weren’t daunting enough, our animators also had to create and choreograph thousands of CG characters, animate a city evolving over 100 years, create a CG landscape with mountains and sky… and ultimately have all the CG elements come together to match the live-action footage, in one continuous shot… All in 30 seconds!

During pre-pro meetings I suggested we investigate a specialized software package for the creation of our tree. I’d heard of a tool with the curious name “Xfrog” from German software developer Greenworks in Berlin. The software was originally developed for the SGI platform and it’s name is an acronym which stands for X-windows based Finite Recursive Object Generator. We tested the software and consulted with the original developer in Duisburg, Timm Dapper. The software is available as a stand-alone but was also available as a plugin to one of our workhorse 3D packages Cinema 4D. Early tests were encouraging. The tool had extensive control, including such esoteric parameters as “branch deviation, tropism and phyllotaxis“… What fun! We decided to proceed with Xfrog for the design and animation of our tree. As I had gone and opened my big mouth to suggest it, I was given the task!

It’s projects like this that have made me pre-maturely gray! Animation parameters that seemed perfectly capable during early testing, quickly fell apart when pushed to the extent required to grow a tree from sapling to full maturity. The interrelatedness of control parameters was particularly frustrating. Just when we were having success with one parameter, another would be wrecked or cause anomalies in the motion or the generation of subsequent branching. It all went around in a seemingly infinite loop of madness for nearly a month! Finally after lots of brute force keyframing, I wrangled the thing into submission and we had our hero tree. In spite of the difficulties with Xfrog, it is brilliant software and I shudder to think of having to pull off such a complex task in a conventional 3D package. Camera data, exported from XSI in .fbx format, was imported into C4D to match final camera animation. Beauty, shadow and ground passes were rendered for later compositing into the final scene underway in XSI.

All the while I was trying to avert insanity over the tree animation, Troy had his hands full trying to fit over 2500 3D characters through the pipeline! The director called for crowds of “employees” to encircle the tree while the (virtual) camera cranes overhead. In the final shot the employees turn over cards forming the logos of the various companies under the Grupo Salinas corporate umbrella. In a final card flip the Grupo Salinas logo is formed.

To accomplish the effect, a library of digital characters were modeled and rigged in XSI. Walk cycles and a card lift sequence were animated for each character. Troy then wrote a custom script which would arbitrarily select from this library and place the characters in a radiating position around the tree. Random seed values were incorporated into the script in order to position characters in natural distances and angles relative to one another. Scripting also controlled the timing of the characters as they walked and lifted their cards overhead. The animation of the cards and the parsing of the shared texture map of logos was also choreographed via custom scripts. The complexity of executing the scripts (not to mention writing them!) as well as the shear volume of geometry involved, made for extremely slow going during this phase. Extensive optimization of the scene as well as delicate settings within XSI to optimize the Mental Ray renderer, were required before the scene would render without choking!

The time-lapse evolution of the city skyline was also modeled, animated and rendered within XSI. The client supplied photo references of period architecture and descriptions of the early Salinas y Rocha building and it’s various incarnations to the present day. Geomedia 3D whiz kid extraordinaire, Jeremy Kenisky, painstakingly modeled and animated each building as it is “built” from the ground up. Jeremy also created the sequence of the original SyR store as it undergoes a dramatic transformation between its intermediate forms and it’s familiar current day modern architecture and signage.

Another complex piece of the puzzle involved the natural environment in which the entire spot takes place. The director wanted to see an empty vista stretching to a horizon defined by foothills and distant mountains. As the city evolves the natural landscape transforms into a manicured city park. He imagined clouds passing rapidly overhead to further enhance the stylized passage of time. Early in the project we considered stock photography and time-lapse cloud video but we could not find the perfect combination of perspective or point of view. The single camera pull out and overhead move also presented problems with this approach. We would have to somehow do the environment in 3D.

Like a dumb ass I once again stuck my foot in my mouth and suggested we consider yet another unfamiliar software package to get the job done. Vue Infinite from e-on software turned out to be just the ticket. The software has amazing tools for creating natural landscapes and photo-realistic atmospheres but it’s unorthodox user interface and approach took some getting used to. The biggest problem (besides the horrendous render times) was getting camera data from XSI into our Vue scene. The beta version of Vue we were using had no native XSI support. We finally discovered a circuitous route through various 3D packages to convert the data into something Vue would recognize. The final rendered landscape matched seamlessly.

The live action plates and final rendered CG output from the disparate software packages, were composited in After Effects. Keying, color correction and extensive FX work were also performed to blend all the elements and impart the soft atmospheric feel and painterly palette the director called for.

The spot was an enormous technical and artistic challenge for the Geomedia team and I’m humbled by their show of skill and tenacity over the long hours required to see the project to completion… Having the director declare that the finished spot surpassed the vision even he had in his own minds eye, makes it all worth it.

High Fashion

Pinball Wizards

Be the ball… In this final installment of a trifecta of visual FX spots for Mexico theater chain MM Cinemas, we take the viewer on a pinball’s journey through the machine. Completing a huge re-branding campaign, this exciting, fast-paced theatrical policy trailer reinforces the company’s brand partners and touts the theater’s numerous amenities while encouraging courteous behavior from it’s patrons. This project is the culmination of our efforts through agency SOS Publicidad for client MM Cinemas, in what has been a very successful transition for the company.

Given only the general concept, it was up to us to design the look, direct the action and move the story. In addition to reiterating the typical policy message i.e. turn off cell phones, no smoking… We also had to highlight features such as digital surround sound, the food court, wide aisles and comfortable seating… among others.

Conveying these messages with only a few pop bumpers, flags and drop targets proved to be quite a challenge. For some of the trickier concepts we departed from traditional pinball mechanisms and scale. A good example of solutions for some of the more difficult to convey messages are the scenes describing stadium seating and digital audio.

With most of the story-line and visuals figured out we began constructing the pinball machine. 3D modeling was carried out in Softimage/XSI practically single-handedly by our own Jeremy Kenisky. It was like modeling a small city! There were hundreds of individual elements to construct and numerous iterations of the model were presented before reaching it’s final form. The model had to literally function yet be consistent and visually interesting. Design modifications and additions occurred right up to final delivery as we made sure were getting the best use of available real estate to tell our story.

As the model neared completion, references of it’s geometry were sent to the graphics department to begin the arduous task of texturing. Geomedia artist Rudy Martinez handled the bulk of the 2D design of the table and prepared most the texture maps. A blackbelt in both Adobe Illustrator and Photoshop, He crafted a beautiful theme and palette based on the re-branded motif of the theaters. Additionally, hundreds of functional design elements i.e. arrows, pathways, logos, lights… would need to be designed and consistently integrated. All textures received careful attention to insure sufficient resolution in close-up shots. It was an enormous undertaking given the huge number of geometric elements in the model, all requiring individual textures. Rudy also designed and animated 2D motion textures which appeared on the backglass and pathways. The finished spot would not have turned out as special without such terrific design and artwork… Kudo’s Rudy!

While modeling and texturing continued, scene blocking, animation and lighting were being developed. Animation Director Troy Davis was responsible for lighting and the overall look of the spot. Troy also oversaw animation and camera work as well as render wrangling and technical direction. The scene had well over 100 lights. High dynamic range (HDRI) arcade environment maps were used to great effect on the highly reflective pinball and metallic surfaces of the machine. Ambient occlusion, shadow and motion vector passes were rendered separately to be incorporated later in compositing. A particularly challenging aspect to the lighting was the sequencing of lights illuminating the playfield during specific moments of game play. The pinball table had around 75 embedded playfield lights. Manually keying believable and interesting sequences would be a bit tedious… Scripting to the rescue! Geomedia artists developed a small animated image map which corresponded with the placement of the 3D playfield lights in XSI. Troy wrote a script in which a light is triggered to be on or off based on the animated image map. Utilizing this method quick changes could be made to the lighting sequences simply by editing the animation of the 2D image map in After Effects. We tweaked the image map animation until we had something we liked and then plugged it into our XSI script to trigger all the lights automatically… Brilliant!

Rendered image passes of the film-resolution frames were composited in After Effects. Ambient occlusion and shadow levels were dialed in. Motion vector data files output from XSI were utilized in the composite to generate motion blur as a post effect. This greatly reduced the already substantial rendering burden of processing the 2K files. Color correction tweaks and subtle glow effects were added on a shot-by-shot basis as needed. Fireworks effects were created via 2D particle system and tracked to match the camera move in the climactic final crescendo as the player scores the ultimate point. Audio was scored and posted in Mexico. Final frames were delivered as digital files for film-out and duplication in Mexico City.

Once again, I’m thrilled by the creativity, talent and teamwork exhibited by the entire Geomedia crew in pulling off such a large and complex project. We’re told that the spot looks awesome on the big screen. We’ve received effusive compliments from the MM Cinemas client and the spot has become very popular with the audiences.

ENJOY THE SHOW!!

CLIENT: MM Cinemas

AGENCY: SOS Publicidad

ANIMATION/FX: Geomedia, Inc.
Troy Davis, Animation Director
Martin Jaeger, Compositing/FX
Jeremy Kenisky, 3D Modeling/Animation
Rudy Martinez, Graphics/2D Animation

Snow Patrol

Who’d of thought an innocent little powdered donut could wreak so much hilarious havoc! In this updated spinoff of our popular 2004 “Avalanche” spot for client Bimbo S.A., a runaway donut careens down the snowy slopes of a ski resort, amassing an interesting menagerie of “wildlife” as it goes. Culminating in a powdery collision with a tree, the characters are eager to go again on this unlikely thrill ride with Bimbo “Donitas”!

Geomedia reunited with production company Sevilla Cineproducciones and Director Walter Luis Brunialti on this :30 for agency Foote, Cone & Belding, S.A. de C.V. Principle photography was conducted in Vancouver BC through Lemonade films, Inc. Animatiion Director Troy Davis flew to Canada to supervise FX shots as well as green-screen setups. Of course the weather was beautiful for the pre-pro days and steadily deteriorated for the scheduled shoot days, resulting in several missed shots which would need to be entirely recreated in CG. Geomedia to the rescue!

The spot follows a powder donut, inadvertently dropped by a young snow-boarder. As it rolls down a ski run it picks up snow and grows to enormous proportion. The increasingly massive donut also picks up an imaginative array of winter characters on its course, carrying them along for the ride. Beginning with a hapless trio of fellow snow-boarders, the cast includes a family of “yetis” and a bewildered sea lion.

Actors in costumes portrayed the yeti (abominable snowman) family. In action scenes with the donut, they were filmed reacting to the approach and impending collision with the giant pastry. The actors were then removed from the scenes and film was rolled to create the clean plates for the CG effects. The yeti shots all required extensive rotoscoping, paint and color correction in order to hide seams in their costumes and soften telltale fabric folds.

Traveling camera shots of the approaching donut or following POV shots were accomplished via Steadicam from a moving snowmobile. These shots would later be ramped in post production and 3D tracked in order to match the animated donut’s movement to the scene as we follow it along it’s course. 3D animation of the donut was accomplished in Softimage/XSI as was all particle animation of snow being kicked up by the rolling donut. Troy devised a clever trick to create the trough left in the snow by the passing donut. A separate particle system would create a trail of particles which would accumulate on a 3D ground plane as the donut rolled through the scene. This trail of particles was rendered in a separate pass as a matte. During compositing a deformation of the background plate could be modulated by this matte to create the look of a depression in the snow. Erosion and fractal distortions of the edge of the matte enhance realism and lend an organic quality of the effect.

Troy also improvised an ingenious solution to create the climactic snow explosion when the donut finally hits a tree ending it’s downhill juggernaut. The technique involved a dynamic “rigid bodies” simulation to create the chunks of snow and subsequent smaller pieces of snow and particles ejected from the disintegrating donut upon impact. It’s a little tricky to describe and even trickier to set up. First, objects resembling fragmented chunks of snow were modeled. These chunks were tagged as rigid bodies meaning they would obey physical properties like gravity, friction, mass… They would also react in a physically accurate manner when colliding with other objects in the scene as well as with each other. Next a polygonal torus the size of the donut was created and designated as a containing volume for the rigid body objects. A single polygon was removed from this torus through which it would be “filled” with the rigid body snow chunk objects. A preliminary dynamic simulation was run whereby a long tall column of the rigid body objects was “poured” into the torus object filling the volume. The previously removed polygon was then replaced and voilá, a perfect “pre-fractured” donut was ready for the tree impact.

The primary dynamic simulation was now run with our “snow-filled donut” rolling down an inclined plane in 3D space toward a cylindrical “collision object”. At the moment of impact the influence of the bounding torus was turned off. No longer contained by the torus object and with the velocity inherited from rolling down the hill, the snow objects collide with the “tree” object and explode, scatter and bounce as the simulation is computed. The primary snow chunk objects are also programmed to emit a second complex of particles to simulate the disintegration process. This second group of particles, in turn, spawns a third complex of light fluffy snow flake particles. Each particle system is endowed with physical properties which govern it’s interaction with the atmosphere such as turbulence, air friction and wind. The final composited effect with shadows, camera shake and ambient occlusion makes for a spectacular finale.

Now we know that fur seals are not indigenous to Canadian ski slopes but hey, if you’ve already got a family of yeti’s, a fur seal is not such a stretch of the imagination! This little piece of character animation was no trivial undertaking. Seals are not known for their emotive facial expression so striking a balance between realism and the ability to convey surprise without looking too cartoony was a tough get. Modeling, texturing and rigging of the seal character was handled by Geomedia animator Jeremy Kenisky. Early iterations of the seal had a more furry look, similar the endearing little baby seals seen in PETA propaganda. Final versions incorporated a more wet, circus-type seal hybrid type of look.

In addition to some of the animation chores, I handled the bulk of the compositing and FX work. 3D animation sequences were rendered out to separate pass channels. Ambient occlusion, shadow, lighting and beauty passes were combined in After Effects. Color correction, rotoscoping and final sweetening were also handled in AE. As is becoming more and more common, final frames were delivered to the client on disc (in lieu of tape) for ingest, finishing and dubbing at the agency’s facilities in Mexico City.

As always, it was a pleasure to work with our friends at Sevilla and we look forward to the next adventure with Bimbo Donitas!

CLIENT: Bimbo, S.A.
Raul Vargas, Director of Marketing

AGENCY: Foote cone & Belding, S.A. de C.V.
Samuel Beltran, Creative

PRODUCTION HOUSE: Sevilla Cineproducciones, S.A.
Walter Luis brunialti, Director
Carlos Sevilla Ortiz, Executive Producer
Barbara Chavarria, G., Producer

ANIMATION/FX: Geomedia, Inc.
Troy Davis, Animation Director
Martin Jaeger, Compositing
Jeremy Kenisky, Animator

PRODUCTION HOUSE: Lemonade Films, Inc.
Ted Herman, Producer
Al Cooper, Production Manager
Carmen Ruiz y Laza, Coordinator/Translation
Gerry Higgins, Coordinator

Olé Solé

With nary a breather after wrapping a previous all-CG production, the boffins in the Geomedia 3D department have kicked out yet another fully animated spot. This time with The Delta Group through Cougar Films for Sigma Alimentos,

The spot entails a group of four super-hero kids, each of whom represent the flavors and embody the beneficial powers, of a delicious new soy-based protein drink. We find the “Solé Kids” as they are known, in their futuristic ready-room being debriefed on an amazing new source of health and energy. Excitement builds in the Solé Kids, as their robot professor extolls the benefits of the product. Fielding the impatient questioning from our superheroes, the professor instructs them to share the healthful powers of Solé Kids with the children of Earth.

Delta Agency Creative Director Ernesto Barba wanted the spot to have a realistic, optical feel to the atmosphere and lighting in the spot, yet still have the Solé Kids read as cartoon characters with stylized realism. He supplied detailed character studies. Each character had it’s own personality profile to compliment their respective super power. Character style sheets of the Solé Kids were also provided as they had already been created for the product packaging and printed assets. The professor character was unique to the spot and the director would rely on Geomedia artists for his design.

3D modeling and texturing of the Solé Kids was done in-house by modeler/animator Jeremy Kenisky in Softimage/XSI. He also single-handedly designed and built the classroom set complete with props. The guy is a wonder and easily the fastest modeler I’ve ever seen! The character’s hero outfits and stylized proportions would require careful attention in modeling to achieve efficient geometry and insure clean deformation during animation. Early model tests were well received by the client… Except for one “colorfully worded” modification requested by the Executive Producer. It seems Jeremy had made the female character a bit too sexy!! She would have to be toned down a bit in key areas in order to receive a “G” rating! Was hillarious! Hey, the guy’s only 20, what do you expect. We all had a good laugh.

The professor was an interesting design challenge. The director wanted a simple character capable of emoting through facial expression yet still be obviously robotic. Initial concept sketches with predictable mechanical-hinged facial features were considered but never presented. More promising designs maintained mechanical body and appendages (except for the hands) but more “human” mouth and eyes… Still metallic but flexible. Worked great. Simple ball-socket treatments for elbows, wrists and an homage to the “Segway” for his means of locomotion… A few Jetson-esque detail flourishes, a flashing lightbulb and we had our “professor”.

After completion and approval of my design, I also had the pleasure of modeling him. Aside from the mouth and face, pretty straightforward stuff… Carried out once again in XSI. Gotta love those “subdees“

Rigging of the characters was accomplished by New York artist, Kris Rivel. Models were rigged with deep control capability including FK/IK blending and emotion controls. Kris also created “synoptic” views of each character. The synoptic view is a feature within XSI where selections to various rig controls can be accessed via hot links on a bitmap image. In this case hot links were created on an image of the character so animators needed only to click on, say the upper arm in order to invoke that element’s transform controls. As characters become increasingly complex, these workflow enhancements greatly speed up the selection and keyframing of myriad animatable parameters.

As our characters would be required to grasp, speak, walk and even play guitar!… The rigs required to enable this degree of control were fairly complex. To me, a fully rigged character and all its control objects is a thing of beauty! Here’s the character “Ran” including his control rig, playing some tunes! Is that a Stratocaster?

After approval of camera blocking and animatics, scenes were divied up among the animators. Audio tracks, from voice-over sessions conducted in Mexico, were imported into XSI in order to time action and, of course, lip synch when the characters speak. Even though we animate with computers, the animation process is still done by hand, frame by frame… The apparent simplicity of the final product understates what’s really going on beneath the surface. XSI offers incredible control over any animatable parameter. Editable function curves modulate the rhythm and velociy of motion. The dope sheet is a hold-over from traditional animation and can control the timing of many parameters at once. Navigating and viewing the resulting flood of motion data can get more than a little hairy!

Lighting was a critical factor to the overall look of the spot. Deep shadows and volumetric lighting FX create a depth and atmosphere in the scenes. In addition to animating key scenes, animation Director Troy Davis handled lighting duties for each individual scene.

As animation progressed, work-in-progress scenes were periodically assembled and run by the director for any critique of motion or timing. Approved scenes would be put in the queue for rendering. Prior to final rendering, scenes were split up into passes to derive seperate files for specific objects or specific data channels. For instance, depth channel and motion vector data was output and later utilized in compositing to create “depth of field” and “motion blur” effects. This approach is often more efficient in terms of render time and yields more flexibility as opposed to including such effects in the render. Final frames were finished in After Effects and sweetened with color correction and additional atmosphere enhancements.

Creative Director, Ernesto Barba and Executive Producer Gerardo “Cougar” Botello came to the studio a few days before final delivery to oversee the final tweaks. We had a blast with these guys! Only a few minor changes and the addition of supers requested by the client, Sigma Alimentos, and we had it in the can.

We look forward for more adventures with the “Solé Kids” as the saga continues!

Solé Kids_enclosure