The APA turned to Geomedia to help them create the :30 TV spot that would drive this national ad campaign. The project turned out to be an interesting one for us and more importantly a successful one for the APA so we wanted to share some of the backstory with you here.

The concept for the commercial originated with the internal APA creative team and featured a husband and wife driving down a road where the road signs seem to predict inevitable misfortune. Everything seems to be going wrong before their eyes as they encounter bad weather, a rough road with endless hairpin turns, and no end in sight. But wait! A text message chimes in alerting them that their paycheck had been safely deposited and the world around them suddenly transforms to a place of beauty and color.

Nice idea APA. Like the way they use the images to tell the story.

Let’s get started.

Our first challenge was to find a place close to home that was wide open because just about every shot in our storyboard involved either the interior or exterior of an always in motion picture car.  

Our Picture Car

The location also had to be one that would down the road help us create the images of falling rocks, lightning strikes, and dangerous terrain that were vital to telling the story. More on that later.

To find a practical filming location, we began scouring the Hill Country West of I-35 from Austin to San Antonio. Though safety for our crew and talent is always a primary concern, we were once again quickly reminded during our location scouting of that infamous production axiom:

…that where ever man built roads, he also built power and telephone lines….

So after striking out on our first few candidates, we continued the search from just south of Austin to Wimberly, to San Marcos, to New Braunfels, before checking out Canyon Lake where we stumbled upon the perfect shooting location inside Canyon Park which is owned and managed by the US Army Corp of Engineers.

The park embraced a beautiful scene with awesome vantage point perspectives that would allow us to shoot from one hilltop to the next. Traffic control and lock-up would also give us the advantage of being able to shoot all day without too much interruption.

Finding the perfect location.

With our location approved, we filled out our own Geomedia RED ONE Camera Cinema Package by renting trusty old #11 from GEAR Rental in Austin, who also supplied specialty lenses, Grip and Lighting Package and Jib Arm. Keene Television and Film Services supplied the car rigging support for the car and lighting mounts.

With our toolbox now complete, we scheduled the shoot for one very ambitious day — filming from 8 car-mount positions, 2 crane set-ups, and several vantage-point shots captured with a Canon 400mm T4 telephoto lens. Assistant Director Tony Griffin and 2nd AD Donald Banks choreographed the ever-moving production convoy as Scott Hayes of the San Antonio Police Department lead the unit safely through the over 26 mile course.

(left) Our Crew getting "The Rig" ready (middle) "Hanging On"- Ron Meneses, Tony Griffin, and Jason Keene, (right) "Checking the Vantagepoint" – Zach Nasits- DP, with the APA Agency (Mark Coindreau-PR Manager, Erika Hurst-PR Coordinator, Ismael "Smiley" Garcia- Multimedia Manager, Bob Stover-Congress Production Manager

Jason Keene came on board as Key Grip and Gaffer Ron Meneses master-minded the rigging and lighting for the moving car shots. They were assisted by Roger Eickenroht and Eric Untersee. 1st AC Wes Turner managed the camera, and Donna Horner kept our people looking great with her talents in make-up.

Film Crew from (left to right) Wes Turner-1st AC, Tony Griffin-1st AD, (on ladder)Zach Nasits- Director/DP, Ron Meneses-Gaffer, Eric Untersee-Grip, (hidden) Jason Keene- Key Grip, Roger Eickenroht)

Creating practical effects on location or using tow rigs and process trailers with lighting effects and wind machines were outside the scope of this project.  And a poorman’s process trailer wouldn’t provide belief to the viewer without an even more elaborate greenscreen and photo-real compositing solution. Plus with over 90 straight days of 100+ degree weather and with no rain or clouds on the Doppler for months we were going to rely once again on the Geomedia 3D Animation/Visual Effect department to save our bacon.

Our first visual effects task was to create our overcast skies and lightning strikes.  We laid the groundwork for this stormy and “ominous look” with some in-camera setup on the Red One in conjunction with a Tiffen Cool Day for Night Filter. Sky replacement and lightning FX within the foreboding storm clouds were done in Adobe After Effects by Geomedia FX artist Martin Jaeger. Color grading to enhance the day for night effect were achieved in Apple Color.

Shot with Tiffen Cool Day for Night Filter with 3D Enhanced Skies

Next on the CGI agenda was building the fallen rock that blocks the road for our picture car. Tracking markers for the placement of the rock slide were set up on the road during principle photography. Boulders, rocks and debris were photographed and digitally composited into the final scene utilizing the tracking markers to match camera moves.

The Geomedia sound design team then scored the spot, created foley effects and tweaked the final mix down.

Here’s how the final spot turned out.

Thanks to the crew, cast and the great people at the APA for making this project such a pleasure to work on.

Kiko wanted a graphic look with almost painterly strokes as a metaphor for how this would turn out. We took inspiration from a variety of spots and ended up coming up with our own unique look and style for the client.

We decided to go with an “ink drop” effect, where color and detail would be added to a parchment like paper background through ink saturating it over and over again with a variety of colors and values. The first process in this was to create a realistic ink spreading effect.

We used Softimage particles along with a variety of After Effects turbulence and distortion filters, but decided it would be best to shoot real ink interacting with real paper to give the effect the authenticity required to sell it. We set up a rig to shoot the effect on camera. We ended up using a mirror at 45 degrees under a glass table to get the necessary angle without having hands and fingers interfering with the shot. We shot a variety of fluids, food colorings, papers, inks, and oils until we had enough versions of the type of effect we were looking for.

After digitizing the footage into the computer, we had to move on to creating the iconic Mexican urban landscapes that would be used in the spot. The client had sent us hundreds of images of Mexico City landmarks and places of interest. Rudy Martinez painstakingly rotoscoped out all of the buildings so we could use them in our 3D world. Troy then took all of the cutout buildings and created the 3D sets complete with roads, cars, highways, and various other urban elements to complete the look. Once the sets were finished, he choreographed the brilliant camera choreography and started rendering the shots inside Softimage.

For rendering style we chose to go nearly photorealistic. We would then take the photorealistic images from Softimage and edit them in After Effects to create our effects. For each render we would create several looks for it inside our compositor by running a variety of wireframe, cel shading, and other filters over it. By mixing our different variations of the same image, we were able to create a look that we could tweak over and over again without having to completely rerender the image should a change occur. It also made it possible for us to tweak certain aspects of the image over time, while leaving others the same to further the “painted on” effect.

I handled most of the After Effects duty with several cameos by Joe Schaertl, a Geomedia editor / compositor artist. By layering the ink blots and using them as luminance mats for the color passes of our rendered images, we were able to create a very fluid ink wiping effect. With control over opacity and subtle color changes, in addition to A LOT of finely tuned layering, the ink smoke effect became very believable and ultimately very beautiful to watch. By mixing our render variation layers back on top we were able to control where we wanted detail and color on every part of the screen.

The project garnered a bronze ADDY at the recent 2009 ADDY Awards Banquet here in San Antonio.

We are pleased to bring home the coveted Gold Addy statues for our work in two categories, Video Sales Presentation with “Presentation” for >Dell and Special Effects, Video or Film with “Motociclista” for ActII.

We also received a bronze Addy for the visual effects work on “Obras”, an uniquely stylized animation spot created for the government of Mexcio City.

There was an amazing array of creative work in the competition, and we’re certainly proud to be amongst this year’s crop of winners.  Here’s a link to a complete list of all the 2009 Addy honorees.

We were handed creative storyboards that included two :30 commercials and four :15 commercials. The :30 commercials featured life-style vignettes with the actual bank officers. The :15 commercials were more “abstract” concepts. They were boarded as single lock-down shots that portrayed HCSB’s relationships with their customers using local iconic imagery. These shots needed to be personal to the bank yet designed to keep the television viewers “dialed-in.”

The HCSB creative called for a “film-look.” so we bid the project for 35mm or 16mm film production, as well as an HD package incorporating 35mm film lenses to capture a more filmic, shallow depth of focus. The large format “cine-style” HD camera is more expensive to use than most comparable film packages but offers a net savings in post-production with no film transfer, printing, or shipping costs.

The HCSB client chose the HD “cine-style” turnkey production.

Our DP/production director Zach Nasits phoned our buddies at Gear in Austin, TX to order the Varicam camera package outfitted with 35mm film lenses. As an option, Kirk Miles, the camera technician at Gear, offered us the RED ONE camera for the production and mentioned that it would cost about the same as the HD package would with an add-on lens adapter and 35mm lenses.

In the summer of 2007, Zach had worked as a 1st assistant cameraman with the RED camera on a test shoot for Robert Rodruigez at Troublemaker films. Jim Jannard, the founder of Oakley and inventor of the RED, had brought down two prototype RED cameras: “Boris” and “Natasha” for Rodruigez to investigate. Zach mentioned that prior to that the only other people who’d tested the camera before Rodruiguez were Steven Soderbergh and Peter Jackson. Both would go on to shoot features with the new camera, with Soderbergh receiving accolades at Cannes for his films “The Argentine” and “Guerrilla”

With a RED on order for months now, we had been waiting for the perfect opportunity for a first production with this revolutionary format. Many early adopters were posting positive reviews on the RED’s performance in the field but working with the 4K format in post would present new challenges. We agreed that shooting 4K would be an attractive option for the HCSB production. We proceeded to work out the post workflows and integrate new hardware and software into the pipeline.

Zach and I headed up to our Austin offices and performed tests at Gear offices, investigating the RED camera’s latitude, formats, frames per second, ASA’s, depth-of-field, and pure “user-friendliness” before sending the RAW footage back to Geomedia’s main facility in San Antonio, where we would push the 4K files through post-production and test finished output.

Testing proved extremely successful… It was time for Geomedia to paint HCSB’s television campaign RED!

“It was handled a lot like a film job” said production coordinator Murray Breit. “We synced sound by clapping slates. Zach metered light just like we were using film… We sent the footage through a digital in-house 4K tele-cine process and everything looked amazing! We were able to do everything, start to finish, right here. Breit added “The process we developed for editing, color grading, conform and finishing worked perfectly… Most importantly the project came in on time and on budget and the client was very impressed with the quality of their new spots.”

Geomedia utilized the RED camera on location the day after the HCSB shoot for a commercial for Fuse Gym based in San Diego, CA. The RED camera package consisted of Zeiss Super Speeds, an Angenieux 10:1 zoom lens with full film camera support. Zach captured 4K at 500 (ASA) to have more latitude into the high-lights and rated the camera at 160 ASA as a result of their camera tests.

He explains the differences between each concept and clients needs, “It’s great that we had consecutive shoot days and got to capture two completely different looks with this camera… HCSB was more conventional beauty shots while Fuse was way more gritty.. shooting a 6 stops ratio.”

With a set of Super-Speeds on the way and a full slate of RED shoots on the books, we are more anxious than ever for “our” RED ONE #2357 to arrive at our door! We have not decided on any names for “her” yet!!

Brand new Computers and monitors were sent from Dell HQ and delivered to Geomedia cloaked in secrecy. These machines were to debut at CES and were not yet on the market. Using a variety of tools including a set of electronic digital calipers, the geomedia team set about measuring and mapping the dimensions the machines in order to accurately re-create them in 3D space. Overall there were 10 computers and 3 accessories that had to be modeled. We budgeted essentially 1 day per machine, so we were modeling, texturing, lighting, and rendering 1 computer per day (very looong days). All modeling was accomplished in Softimage/XSI by Jeremy Kenisky with a cameo appearance by yours truly on some minor modeling chores. Here’s some of Jeremy’s copious measurment notes and a laptop model from the early stages of the process.

After 2 weeks of intense modeling, the entire lineup was finally complete and animation was ready to begin. At the same time as the computers were being created, Troy was developing lighting setups and fine tuning material settings. Troy also created HDRI environment maps to “light” the models by setting up an actual turntable set in our studio just as one would do for a high end product shoot. Bracketed exposures were captured on a Canon 20D digital SLR camera, of a chrome sphere from which the HDRI image was derived. This image was used to create the environment in 3D to accurately reflect lighting and bounce cards from the studio setup. Since final animation would be rendered in 1080p HD, it was important that render parameters be highly optimized in order to have all rendering complete and ready in time for CES (January 7th-10th). Separate Render passes were set up for ambient occlusion, mattes and reflections… We opted to create motion blur as a post process in the composite, which worked great.

By the the 3rd week, during Christmas weekend, we started doing animation tests. At this point we had essentially 5 days (including the weekend and christmas) to deliver animation sequences to the client, with an extra 4 days on the backend to do final renders, compositing and finishing. We completed 4 animations in 4 days (very looong days), making it one of the fastest projects we’ve done in a while.

The animation goal was simple – show off a set of computers for a specified time as creatively as possible. The spots were divided into 4 categories – GAMING, HOME, MUSIC, and PHOTO.

The gaming spot consisted of 3 computers – The all new Dell XPS 630i, the Dell M1730, and the Alienware ALX-51. For this spot we decided to go with a “transformer” effect to get from the XPS 630 to the M1730. We would then use the screen of the M1730 (a 17″ 1080p laptop) to transition to the Alienware. One of the biggest challenges to this transformer like effect was designing the movements and deciding what pieces went where. The computer models had been modeled as solid objects and had to be redesigned and broken apart to achieve the movements we envisioned.

HOME

The home spot featured the XPS 420, XPS ONE, and all new Crystal 22″ Monitor. We designed his one to be pretty straightforward – just a big turntable camera move revealing all 3 images.

MUSIC

The music spot featured the XPS 420, XPS M1330, Inspiron 1525, and Inspiron 531s. The spot begins with the XPS 420, which suddenly gets a “sound wave” sent through it as it breaks apart and then back together again. It then transforms into the XPS M1330 laptop. As the laptop revolves around, we see it open and the screen is showing colorful audio EQ-level meters in sync with the current music track. As we get closer, we realize that the levels bars are actually Inspiron 1525 laptops. We go into the screen and show the laptops moving with the music. We orbit them and as we pass the 270 degree mark we transition into a white set with the Inspiron 531s.

PHOTO

The photo spot features the XPS M1330, Inspiron 1525, and Inspiron 531s. For the beginning of this spot, we wanted to use the photo theme as a gag in featuring the 3 colors of the M1330 laptop. As the camera comes down we see a laptop turn and open as a photoflash freezes the action. The image then unfreezes while the camera move continues and the next computer opens up with another flash. This happens one final time with the last laptop and then we fly into the screen of the last laptop to reveal the Inspiron 1525 laptop. In what looks like a flower from a top view, we see all the colors comprising the Inspiron 1525 line as they open up in sequence. Finally, we zoom into the Dell logo on one of the laptops and quickly pull out to transition to the final Inspiron 531s.

Additional 2D animation was created to simulate Windows Vista screen interaction and Dell branding. These were then mapped onto the 3D computer monitors. Much of the compositing was done via the XSI compositor with final tweaks such optical effects and color correction performed in After Effects. We utilized motion vector data from XSI to create the motion blur. This greatly sped up the rendering process helping make the deadline without sacrificing realism. Final animation was rendered at full 1080p Hd resolution.

Finally… The completed animations were delivered in Austin 3 weeks and 3 days after the project started. There they were incorporated into the rest of the presentation video. While we couldn’t attend CES to see the finished work, we heard that all was well received. Requests for follow up projects are already in the works!

How to make a Devo Energy Dome

To construct the hats, I needed to start with a single pattern that could be used for multiple castings. This was made with 2 thicknesses of multiple density fiberboard (MDF) to create the varied heights of the different layers. The bottom two layers are 1-3/4 inches tall, the third level is 1-1/4 inches and the top layer is ¾ inches thick. The pieces for each layer were cut out and the same size pieces glued together to make a solid block for each layer. The layers will remain separate from each other until they’re milled into the proper shape.

The discs that make up each layer have a slight taper of about 15 degrees leaning toward the center of the hat. To cut this shape I decided to use a rotary table and my Harbor Freight Mini-Mill. Holes were drilled so the wood could be securely attached to the rotary table. The column of the mill is rotated 15 degrees off its vertical center to make the tapered cut.

The wood is bolted to the turntable and the head of the mill is slowly lowered as the table is rotated. I’ve only used this machine to cut metal, so when the MDF began to cut it was hardly noticeable (except for the unbelievable amount of sawdust that went everywhere). The wood is soft and easy to cut, but I still made a lot of rotations and lowered the mill head a little bit each time around.

The rotary table has a handle that spins 72 times for each revolution of the table, so by the time the bit finally made it thru the bottom of the wood, I’d had a pretty good workout. I had to constantly use the ShopVac to try to contain the sawdust.

Finally the layer is complete, with a fairly smooth and angled surface. It can now be removed from the machine and flipped over to form the base. The next layer goes thru the same process…after three more the mold is beginning to take shape. Finally the separate layers are glued together and sanded.

The vacuum process that will form the plastic hats will transfer any imperfection from the mold to the casting so the mold has to be as smooth as possible. I used multiple coats of Bondo to smooth the wooden mold and sanded between coats. The Bondo turns extremely hard as it dries and I eventually just used plaster to make the smoothing process a little easier. The photo here shows the mold right after the 1^st layer of Bondo was added.

I built a simple vacuum forming table based on the online tutorial found here.

It’s basically a piece of MDF with a hole in the center sitting on a small riser. An insulated border around the hole helps to crate a seal when the hot plastic entirely covers the mold.

Underneath an ordinary vacuum hose is securely attached to the table. Silicon sealant was used between the metal and the MDF to prevent air from leaking into the system. I found an extra sprinkler head part in the garage and it made an air-tight link between the vacuum hose and the threaded pipe connector. The vacuum I used was a 1400 watt household vacuum with a hose attachment.

Aluminum framing designed for window screens is used to make the frames for holding the hot molding plastic. The base diameter of the DEVO hat is just under 10 inches so I made 14 inch square frames to allow for a little extra room all the way around the mold. Giant paper clamps are used to keep the two frame pieces held together. One of the most difficult aspects of this project was trying to find metal corner brackets to keep the aluminum frames connected. The stores in my area sell plastic corner pieces that aren’t designed for the high oven temperatures these will be undergoing, so I ended up milling these from a piece of ½” 6061 aluminum bar. They had to be thinned to slide into the openings, but they fit snug and work well to keep the frames together.

I found a local plastic supplier and purchased two 8’x4’ sheets of .080” high impact polystyrene (HIPS) to use as the material for hats. The material is easily cut by scoring one side with a utility knife and bending the other until it pops cleanly from the sheet. I probably could have used a thinner plastic, but I like the sturdy thicker weight of the .080”.

The 14 inch squares of plastic are placed between the two frames and these large paper binder clamps are used to keep them firmly pressed together. I’ll put one or two on each edge of the frame and remove the wire handles from the clamps before they’re heated.

I taped the foam border sheet to the table and checked to make sure the seal of the insulation perfectly matches with the aluminum frame. It’s important that the vacuum be as tight as possible to closely capture the shape of the mold.

When the hot plastic is ready to be formed, I’ll have to move quickly to transfer the plastic’s frame from the oven to the mold. To help line-up the frame to the insulation border below, I’ve added two support frames to the vacuum table that will easily identify the corner above the mold so I can bring the frame straight down onto the mold. The smooth (however unattractive) mold is placed onto the table and we’re ready to start melting plastic…we’ll almost.

The mold can’t completely block the vacuum hole, but it can’t be too high either otherwise the molding material will to try to form under the mold. I attached these rubber feet to the mold, it seems to create the right height so the molding plastic just makes an edge at the bottom of the mold without tucking under its edge.

I heated the oven to 400 degrees and placed some coffee cups as risers at the corners of the oven rack. These are used to separate the frame from the wire rack below. As the plastic gets hot, it will begin to bulge and I didn’t want the hot plastic to touch anything in the oven. After a couple of minutes the plastic is ready to go. With gloves now, I quickly grab it out of the oven and rush it to the waiting vacuum table just 3 feet away. The vacuum is already running and I rush to align the edge of the frame with the alignment guides I installed earlier.

The frame is brought straight down over the mold and because the plastic is hot, there is almost no resistance. The vacuum takes just a few seconds to remove all the air from under the soft plastic. It quickly begins to take the shape of the model underneath. Both of these photos here show the moment the aluminum frame has just contacted the foam insulation and created the vacuum seal. The plastic is curving into shape but the ridges and steps haven’t been formed. The final shape is formed in just a few seconds, and after 5-10 seconds all the detail is clearly defined. I kept the vacuum going for about a minute to allow the plastic to cool. I also used a heat gun for a few seconds blowing high heat onto the top two rings of the hat to keep the plastic there soft a little longer. This helped to add definition in an area that the vacuum was stressed to empty.

After the plastic has cooled, it is taken off the vacuum table and the mold is removed. I thought the conical shape of the hat would allow the mold to be extracted easily, but it’s captured inside the plastic extremely tightly and took a lot of pressure to get it to release. I ended up pulling the edges of the plastic sheet while pushing down on the top of the hat plastic to free the mold. After that I used a utility knife to trim the hat from the excess plastic, then used a file to even the edge and finally touched up with sandpaper to smooth the sharp rim.

There were a lot of hats to paint, so I constructed a small spray booth from a plastic storage container and an illuminated ventilation fan. The fan is the kind normally used in a bathroom and I attached a hose to vent the fumes outside. Filter material was taped to the fan inlets to reduce the amount of actual paint moving through the fan. A turntable was used to easily rotate the hat during the spray painting producing a smoother and more consistent finish. I picked a red spray paint designed for use on plastic to prevent the paint from chipping or flaking from the semi-flexible hat. Two coats inside and three coats outside and the hats were beautifully painted.

To keep the hat firmly held onto the head I thought a foam ring would be a good way to go. I found a local foam supplier and had them cut 1 ½” thick foam into 9” discs. This is the kind of foam that is used for camera cases with cutouts to hold custom shaped parts. The foam cut OK with the serrated bread knife technique that was recommended by the foam dealer, but since the bread knife left a bit of a ragged edge, I ended up using an electric carving knife to cut the inner ring. I clamped the knife’s handle into the jaws of my bench vise with the blade penetrating an open hole in a strategically placed table saw platform. I was able to guide the foam smoothly around the powered blades to cleanly cut out the shape for the head. The exposed bottom side of the foam was painted black, leaving the original foam unpainted where it will be contacting the skin.

To secure the foam ring to the interior of the hat, contact cement was painted onto both surfaces and when nearly dry they were pressed together. Once the two sides of the contact cement touch each other they’re connected for good so I used wax paper to separate the pieces during alignment and then removed the paper.

Finally, completed hats! Actually, the hats turned out better than expected. The surface is smooth and reflective and the construction is lightweight, sturdy and comfortable. Now we’re ready to suit up and head to the awards. Check it out.

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.

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

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