'Dream of Anhui' is a Special Venue ride project, a flyover tour of several historic & important locations in the Chinese province of Anhui.
As the Environments Supervisor and Environments Art Director my roles ranged from early tests and concepts to pipeline design and shot planning. Once into production my duties included asset selection and builds, production assembly, reviews & guidance of team member efforts, as well as leading a team in the production of two locations.
The initial plan as imagined before anything was begun was for the project to be almost entirely filmed with helicopter footage. It became very clear once the environment team began to explore what was required and what was possible, and synced in with what the client was looking for, that an entirely virtual 3d project would be required, and because of the extent of the camera moves that there would very little allowance for tricks and gags.
Early work was done using terrains cobbled together from Google Earth information and photographic references combined with super rough models. Working with Layout Supervisor Chris Piazis I prevised many locations and the client selected the final spots. Once the final locations were selected we sent a crew to China to hike and scan large areas around each of the locations using drones. When this data was returned I worked with the photgrammetry to orient and incorporate it into the master layouts and these became our starting point and reference for each location build.
In practical terms, the 3d data for each of the locations was reduced to the minimum amount of area we felt we could get away with considering the travel paths of the cameras and what people would see. These areas were then divided into 'sub-sections' in order to divide the work among multiple artists.
At this point its worth briefly discussing the software we selected to use as our layout tool, and the team structure we created. Once we understood the parameters of the project we conducted an exhaustive review of all tech options we felt were available to us, looking into Maya, Lightwave, Houdini and other packages, trying to assess how we would or could use each of these appropriately. While we were initially focusing toward a Houdini solution for the layout and assembly we ended settling on a new (at the time) program called Clarisse, from the French company Isotropix. What we able to do with this program and how it allowed us to work combined with the speed of its renderer and the amount of geometry it could handle - easily - made it the hands down winner. I'm happy to discuss our use of Clarisse if anyone wants to send me a note.
Regarding the team production structure, we ended up deciding on a group structure related to the the asset based 3d "world building" process we built the development around that required a combination of TD and Layout skills as the primary function. The assets and previs setup would feed each team, but the 'assemblly TDs' would be free to develop each section as their lead saw fit, and we tried to balance each team with a combination of skillsets so they'd have options.
As this was a massive 3d fly over - some shots travelled kilometers over the terrain - there really was no way to fake anything, so it was entirely a 3d asset based project - we built these locations. To facilitate this Ben and I bult upon the pipeline we had developed for The Crossing and working with our development team created a full asset management system that allowed us to place and manage hundreds of assets for each shot.
My responsibilities at this point were focused on identifying and designing the assets required for each section and in managing and balancing the asset team's desire for detail over keeping the geometry light and simple. Keyart was created and reworked, and various look direction cards and sheets were created and adjusted to keep the asset production in line. We wanted to keep things as tight as possible within the pipeline but also allow the freedom to be somewhat flexible in creating assets that were needed.
A large part of the asset build was also the Hero location items: the temples, buildings, dams, bridges and architecture we identified as specific and key to each location. These were built entirely by the asset team. The process was to first build them roughly, to barely better than blocking level detail and get them into the asset management system so that the layout could begin, and then as the layout was developing to continue to refine them into proper production acceptable assets.
Once the base assets were ready and slotted into the asset system for a shot the production team began by working with the previs terrain and combining them with the information and textures gathered from the photogrammetry and reference shoots. Some locations were all natural while others were almost entirely cityscapes so each location had to be treated differently within the basic scope of the shot design, and each team was given leeway to develop the terrain and layout as best they saw fit.
A breakdown of how the project was brought together.
The final renders were 6k 180 degree spherical renders...