This is the final production week and we tried our best to complete our production process. Though we made some changes about story and sequence, we achieved the goal we set up at the beginning of the semester and we successfully delivered our final animamtion cut.
We also reviewed our work this semester again and listed some lessons we learned.
Art: Details bring the environment alive
The environment art of Woodevil is inspired by the idea from the environment concept that is included in the pre-package, the forests in nature, as well as forests in horror films.
The team is aiming to create an old and dead forest. The forest environment is foggy, gloomy, creepy and mysterious. The trees all have extremely weird shapes to emphasize on the horror feeling. The ecosystem reflects the liveless of the forest, where withered grasses, broken tree trunks and rocks dominate the ground.
During the modeling process, the major challenge is to make sure the assets and scenes are rendered correctly and beautifully in Arnold. The environment of Woodevil is set to be dark and colorless. The team made lots of changes to the texture details to ensure the final scenes deliver the feeling to the audiences.
For tree creation, the team used Speedtree to procedurally model the rough shape of trees, then sculpted the detailed look of the trees in Zbrush and optimize the meshes in Maya. The look dev of the tree is completed in Substance Painter and Arnold.
Another challenge is to make high resolution details for objects showing in close up scenes, two examples are tree body and ground texture. The team’s first attempt is to use UDIM to increase the texture resolution in a specific region. To further improve the result, they ended up with adding additional geometries to the existing models, such as individual bark pieces, to the tree body, as well as using tiling nodes in Arnold hypershade to significantly increase the texture quality of the ground.
Pipeline: Pipeline leads to success
We began with developing storyboards according to the given pre-package, and created animatics to lock down the time for each shot. Then, we started asset production, which includes modeling the virtual environment and shooting the live-action footage. Then, we composite the green screen footage and the environment render together. Next, we feed the composite into the post processing process, when we could manage our layered renders and create various visual effects. This pipeline has benefited us a lot in the semester.
One project control tool recommended for an animation pipeline is Shotgun. It is a cool system that labels each shot with the pipeline layers the team will use and easy for team members to track the process of each shot. Meanwhile, it has one big disadvantage that files on it cannot be directly linked from Google Drive or Perforce which means it will take a lot of time uploading materials for each shot and replacing an existing file is a little painful.
Our management solution was uploading animatics of sequence on Shotgun to serve as a preview of each shot and have a clear look at what a shot should be like, so we use it as a tracking process of pipeline layers. Besides, we used perforce for real version control of real animation sources. In that case, we just have a clear track process of the pipeline and a reasonable version control system of source files.
Green screen footage: Lighting makes the difference
Lighting is a big piece in the practice of shooting footage in the green screen room. So, the first challenge we encountered is the green screen lighting. Our story happens in a dim environment, so we want to match the light settings while shooting the footage. What we want is a dim light look on the character and a bright background, so we can have a good contrast on character pattern and key out the green screen in After Effects. Our initial attempt was using a single dim light source to light up the whole environment, both characters and green screen. However, this resulted in the lack of light projected on the green screen. The dark background became difficult to key out, even with a garbage mask applied. To solve this problem, we finally used two wash out lights to light up the green screen without influencing the character.
The other lesson we learned is about the costume of the character in the green screen footage. Initially, we let the costume of the character have a high contrast itself which means it has clear split lines of white and black itself without adjustment of light. After we tried to composite it into postprocess, we found it became hard for us to match the light of footage with the virtual environment because it was really hard to make the contrast of light clear in such a costume.
Online rendering: Free local resources and push to the cloud
We are using Rendercore as our cloud render service. Rendercore supports a number of DCC softwares and renderers, including Arnold for Maya, which is the renderer used by the team. This documentation from Rendercore’s official website may be helpful: https://www.rendercore.com/en/boards/JobSubmissionManuals/www-lists/e61875e7-7be4-4d50-43e7-08d61a86afad
Prepackage: Visual concept serves as the art bible
We began to prepare the prepackage for the future animation team. Due to the limitation of time, we have decided to focus on the art direction and visual presentation guidelines for the next team. During the amendment of the original designs, we had to address the level of difficulties and challenges that this package might provide for the next team. To ensure future teams get the best out of the prepackage without getting frustrated, we tried to provide a story with visual options that the next team can adjust without deviating from the main content we provide. The process of abbreviating the excessive content in the prepackage once again demonstrated the importance of scoping.