GMC Big Dig is big - the carmakers vision, as well as the task laid at the feet of the filmmakers.
The spot was helmed by Space Programs Justin Klarenbeck, and completed by rhinofx in NYC. We spoke to Arman Matin, Creative Director/CG Director of rhinofx and not only do we review before and after shots, but we reprint the original pitch document for the visual effects - so you can judge for yourself how the team performed.
As the ads VO attests, you may not have the contract to build the next great suspension bridge, or the work force to dig the next Panama Canal; and you may not be working on the next big sports complex but your truck could.
movielink(GMC/BigDig.mov, View the final spot)
movielink(GMC/BigDigAfter.mov,Download a Quicktime before and after comparison of shots)
To communicate the scope of these enormous undertakings, exhaustive research was done on locations and shot in HD. These were combined with extensive CGI to create the proper look and atmosphere. The spot remained the same from agency board to completion, with the exception of the stadium, which was supposed to be the West Side Stadium in NYC but was changed to be a stadium of the future. All in all, the effects team was required to build 3 huge locations, a suspension bridge, a canal, and a sports stadium.
The matte paintings were done in phases. First, they team gathered relevant references of bridge construction, Panama Canal, and stadium constructions. Based upon these photographs and some creative conversations with the director, the team chose the shot angles and decided on an approach for each shot. After creating a 3d rough layout or previs, the team agreed on the camera movements. During this time, they analyzed the amount of parallax each shot would have and whether a 2d or 3d matte painting might be required. Composition was created based upon location and reference photographs. And the director decided on color palates for each vignette before the shoot so that concept frame work could begin.
These rough previs frames were used, along with the photographs, to create concept frames. "We tried to incorporate some of the architectural style of the original Panama Canal into modern day construction sites to create a balance of nostalgia and modern day engineering marvels" explains Arman Matin. These concept frames that were based upon 3d previs, served as crucial elements in the whole production process, including client approval, on set supervision, and final compositing.
Based upon these concept frames, 3d models were created and textured and light direction defined. The resolution of the matte paintings was kept above 2K and the team actually created all the comps at 2k but down converted to NTSC for delivery. From a 3d tracking, match modeling or compositing standpoint, working on 2K provided rhinofx with the necessary details and clarity.
A major part of the production was modeling and laying out the sites. The team relied heavily on references to pick every single model that they built: "specific cranes that would be used for specific types of the construction, the type of scaffolding that goes around a bridge versus a stadium, just to mention a few" he adds.
Despite being 2d as the end result, all the matte paintings were based upon textured and lit 3d models. For shots that had moving cameras, they rendered textured occlusion, highlight, and other passes that were composited using Digital Fusion. "For compositing we used After Effects, Digital Fusion and Inferno based upon the preferences of the compositors" explains Matin.
A 2 day shoot provided the elements required. The crew shot around LA under the Vincent Thomas Bridge and in a mining quarry. The parking lot of the Vincent Thomas Bridge served as the location of the stadium element shoot. Due to the limited time, the team had to end up shooting passes in the middle of the day with hard shadows for the stadium shots. "All these shots required extensive lighting and shadow passes to change the noon light to an afternoon light" says Matin. Spark, dust and background action plates were also shot. Plus the team ended up adding these elements in CG as well.
Once shot, plates were transferred at HD resolution. Every shot that required a background change was tracked in 3d using Boujou. Shots that utilized a 2d background were also tracked in 3d to confirm the amount of parallax that might happen. This ensured that the 2d plates could be positioned and tracked correctly.
Once tracked, CGI models were aligned to the provided matte painting, allowing the team to generate a multitude of passes which included dust, glows, fog, close up rain, sheets of rain, rain on lens, rain splatter on ground, workers, sparks, and more. Rhinofx used Maya and Softimage for all the 3d work and Digital Fusion, After Effects and Inferno for compositing. The Matte painters also provided the creatives with rough comps of the live action and matte painting "which were very helpful for the direction of compositing". Rendering was all done in HD resolution and 16bit depth.
"Due to the extremely tight deadline and the scope of the spot, the project took careful planning and orchestration of all facets of production. Complete cooperation between the Director and VFX Supervisor/Director was necessary to create a project like this" explains Matin.
What follows is the visual effects original treatment presented to the director and agency. Read this treatment and judge for yourself how the team at Rhinofx performed.
Reprint of Original Treatment
GMC Sierra notes:
I am very excited about this project and its potential. Due to our tight schedule, I have put together some thoughts that will be helpful in planning the production from day one till delivery. Here are some thoughts in no particular order:
Location: This is a key part of our production process. The locations will determine what we will be creating at the end. As we are representing 3 locations by filming in 2, preplanning the shots will be very helpful for a convincing result. My recommendation would be to try to take digital stills of the location from the angles that you would want to shoot from. These photographs can be used to create previsualizations of complex shots and also provide material for the conceptualization phase.
Concept art: Based upon your direction, we would use the location photographs to create a rough idea of what the final structures and shots might look like. This phase is important, as this is where we lay the foundation for all digital matte and 3d work. As our post shoot schedule is limited, we have to make as much headway as we can. Based upon these concept art, we would create our rough models that can be used for previsualization.
Previs: This is where we work with 3d camera in 3d space to create a blueprint for shots that we would be shooting. Using low res models, concept art, storyboards and location photographs, we can work out complex camera moves, composition, lens etc. Not all shots need to be prevised, however my recommendation would be to plan and frame at least all the VFX heavy shots.
Complexity: Any moving camera shot is complex in terms of 3d matchmoving and tracking. We have technology to address moving camera shots, however a large amount of moving shots can add complexity to the whole post process and become very time consuming. Simple pans, dolly and crane shots are easily trackable, however a combination of zoom and any one of the above is very tricky to track in 3d.
However we can always use 2d tracking which is fairly easy to do and can apply for all shots where our structures are far away and dont have much perspective change as the camera moves. Any time you expect to see a lot of perspective change and parallax, we have to track the shot in 3d. If you have something in the foreground that is practical in camera and we are dropping in a structure behind the foeground element, and there is no perspective change visible in the background calls for 2d tracking which is simplier.
Another limitation of 3d tracking is that not all 35mm lenses can be correctly tracked. We have to be shooting lens and distortion charts for all wide angle lens, as each lens handles distortion compensation differently.
Normally we are very relaxed about these issues and try to allow maximum flexibility, however the time factor is a big one in this case, and we would like to help you create gorgeous imagery without being bogged down by technical challenges.
Look: This is also important as we would like to start working on digital extensions and matte paintings before the shoot. Wanted to know from you how you felt about keeping each location in three different times of the day, as in dawn, daylight, sunset, dusk or night. This will give us direction in terms of the lighting and mood of the day.
I recommend atmospheric elements, like rain or dust.
Nighttime construction site might look great. This kind of specific times of day and atmospherics can provide directionality and increase believability in all the effects work versus a midday flatly lit shot.
The concept arts can be based upon initial direction of look and lighting.
Shoot prep: We will work with you to determine based upon the previs the elements we might need for the effects work.
We will cover our own reference of still photographs which will be used to build the environment, however certain elements need to be shot with 35mm motion camera for consistency of color, sharpness, lens and movement. The previs will help us tremendously in figuring out all the necessary elements and what we need to create in 3d.
Post shoot: If all goes well, we should have already started on the master matte paintings and have already modeled most of the 3d models that we might need. This will all be based on previs and the concept art. After the transfer we start the process of tracking all the shots, whether in 2d or 3d based upon the type of shot.
Using matte paintings, digital stills, transfer footage and elements, 3d elements we start creating the shots. Not all shots will require all these elements, and some can be reused and some shots might not need any digital work beside stabilization and cleanup. We would like to create most elements before the transfer and start compositing the shots right after the shoot. This way we will have the most amount of time to work on compositing all the elements, which is where time needs to be spend for creating believability.
Edit: In most cases VFX shots should be approved in the edit as soon as possible. In this case it is imperative that we solve most camera framing and movements of VFX heavy shots in the previs state and try to capture the same shot as close as possible in film which in turn is used in the edit. This way complex shots have maximum time to be worked on and less complex or effectsless shots can be played around in edit.
We hope to work with you closely on creating this stunning spot. Please feel free to call, email me any time to discuss your vision and ideas that you would want realized.
We plan to help you in any way and at every possible stage of the process to make this vision a reality.
We've been a free service since 1999 and now rely on the generous contributions of readers like you. If you'd like to help support our work, please join the hundreds of others and become an fxinsider member.