Just shy of a year ago, the nine of us individually submitted applications in hopes that we might be chosen for the 2020 Pixar Undergraduate Program (PUP). Little did we know our love for feature animation and years of hard work would bring our unique group together. After getting through multiple rounds of interviews and a nationwide applicant pool, we were offered the opportunity to work alongside and learn from the brilliant minds at Pixar!
Sadly, what felt like only a moment after meeting, we got the unfortunate news that the program was canceled in order to safely navigate the global pandemic. Not wanting to lose out on such an incredible opportunity to work together and learn from each other, we decided to make the best of the unfortunate situation and continue to collaborate. This sparked our ‘Lost Pups’ short film, Upside.
Upside is all about silver linings, and finding light in unexpected places.
After the official cancellation, David Luoh, the PUP’s head coordinator, reached out and asked if our group would still be up for anything during the summer. He was delighted to hear about our ambitions for a group project. David and his team of Pixar veterans and PUP’s alumni set up an online mentorship for our group, where we were able to learn virtually from experts in almost every department at Pixar, along with offering feedback on personal projects and our film.
WHERE THE IDEA CAME FROM
The project kicked off the pre-production stage with a group brainstorming session. Everyone pitched their own ideas — the range of stories varied drastically but each was equally ambitious. After a consultation with our mentors, or the “vets,” we realized our biggest challenge would be sharing our resources and work. We chose to narrow our focus and craft a project that would be able to hone our individual skills as technical directors (basically, not character animation). We ultimately decided on creating a short film focused on environment vignettes and moments that were driven by our collaboratively written poem:
‘It often feels like shadows consume us
And we focus on the dark,
But what we fail to see is that
every shadow has its start.
Light lives in us and around us,
blanketed under warm rays,
it peeks through quiet pockets,
dances through the rainy days,
and roars to remind us,
of Hope, and of Life,
Because light lives
in unexpected places.’
Anne Chen, who came up with the original idea and early poem, was able to provide some early visuals for the world where this poem could take place.
Initially, we chose to have a character that represented light, but after idea development and project constraints, we had to focus on how to creatively portray our ideas purely through environments and small moments. Once our “script” was finalized, our stage was set and we were able to begin production! Kind of…
Although our team possessed the passion and work ethic to adequately complete our project, we lacked stylistic direction. We struggled at first to capture a look and feeling for what this whimsical place was. After discussion with the group, we wanted to create a mood that would be calming but at the same time uplifting. That is when the phrase “Lo-fi with glitter” was born.
Not long into production, we realized that we had to unify our production in order to ensure the cohesion of our short. To help anchor down this project’s theme into a visual language, Do Park and Anne Chen created a style guide that gave some general rules to “Lo-fi with glitter.”
With a group approved style guide, Anne and Do were able to go in and start flushing out all of the scenes. The key to the project was constant feedback and collaboration. They would pitch environments or ideas to the group, get feedback and then go back and forth until we all felt comfortable with our final composition.
Once final compositions were set, the team could move on into the 3D production pipeline. As we started to rapidly iterate in 3D, we were able to find problems quickly and figure out what new things needed to be added or taken away. This made the art team continue to design throughout the production pipeline as draw overs or other production art was needed.
Once we had some concept art laid out for our possible environments, we decided to begin box modeling each scene in order to get a sense of how we were going to fill the space. We assigned each scene to a person and we modeled out the scene with shapes that were simple but still represented what they were supposed to be. Once these layouts were done, we decided to open up camera layout to the whole team in order to get different ideas out there. We then had a meeting where we looked at each shot and decided on a general camera direction for each scene. From here on out, we decided it would be easiest if one person took on the camera layouts to keep the pipeline simple and cohesive. Christian said he was open to doing the camera layouts because he had never done it before and was open to learning.
Creating an animatic with these white box layouts was something that needed to be done early on so that we could begin to go back and forth on how the film was feeling in terms of mood and story. After learning about layout from Adam Habib and Andrea Goh, we learned that we should put purpose into our camera moves and how we can use the camera to direct the viewer’s eyes and add to the story. The next step then was to write down everything our story was about and what we wanted to convey as we moved throughout these different scenes. We used these notes and ideas to give our camera movements more intention. Throughout our whole production timeline, we made sure to keep revising the animatic with updates of scene models, textures, and test lighting. All of these were needed in order to make sure our film was on the right track. The lighting tests proved to be very important and helpful to our lighters in order to begin portraying the importance of light in our short, even though we were nowhere near ready for lighting final shots.
At our second meeting, Daniel Sirota began creating our pipeline. With 9 people on our team, eventually working across 5 different countries, we needed to make sure we had a digital asset pipeline that was easily accessible and simple to navigate. To start, Daniel created a pipeline document for the project that detailed the directory map for the project folder and file-naming conventions.
He also took a week and programmed a desktop application that would act as a visual UI for team members to create, upload, and download assets from the “storage server”. Storage server is in quotes because we used cloud-based storage sites to hold our assets. Using a cloud service means that at any moment, two copies of the project existed — one master copy on the server and one local copy on each member’s machine with their work. The pipeline application allowed for the downloading of either the entire project folder from the server or individual assets with all of their dependencies into the users’ existing local copy. The application also could be used to create new assets through the click of a few buttons that would automatically have the correct file name and be stored in the correct directory in the local version of the project. Finally, assets could be uploaded to the master project through the desktop app as well.
For the most part, the pipeline worked pretty smoothly. Our biggest issue came about 3/4’s of the way through production when we ran out Google Drive storage space. The entire pipeline had to be transferred over to Onedrive where Daniel had a lot more storage space thanks to his school account. The transfer did take two days to complete; however, because everybody had local copies of the project, we were able to keep working on our individual parts. When the transfer was done, we just merged our progress into the master project.
Aside from the main pipeline, many scripts were created inside of Maya to help with the production. Some were as simple as auto-material set-ups to as complex as fixing file paths inside of the Maya scene. Whenever an issue did occur, we would get on a voice call with Daniel and he would talk us through how to debug the problem and help find the cause. A lot of the scripts came out of these help sessions, so that when the same issues occurred to others, we had scripts that would fix the problem quickly.
As we were nearing the end of production and cleaning up our scene files for rendering, we noticed that each scene contained a material called “Wrought Iron”. We don’t know where the material came from, as none of our assets use it and whenever we deleted it, it would reappear the next time the scene was loaded. Eventually, we gave up trying and left it in. It’s been a few weeks since the last time we touched our scene files but its presence still haunts us.
After getting the initial sketches made by the visdev team, we made an asset list where we broke down each environment to divide the modeling tasks of the sets’ structures and props based on their importance to the story.
Halfway through the modeling stage, we looked at the sets as a whole and realized that changes needed to be made to the stylization and level of details of the models.
For the diner shot, our intention was to make it look old and abandoned, but some of the models looked too detailed or clean, distracting the audience’s eyes from the silver lining that we want them to see.
Hence, draw-overs were done by our directors and modelers made the changes. For instance, replacing the first doily pattern with a simpler pattern, and running it through n-cloth to make it look tattered. At the same time, we feel that the last shot in the attic needed a more detailed lamp as it is the focus of the shot, so we replaced the table lamp with a stained glass lamp. Along with making changes based on the draw-overs, we also went through all of the models to add deformations and avoid straight lines or 90 degree angles because we wanted the sets to feel as if they were lived in.
CHALLENGES OF WORKING REMOTE & STYLE REVAMP
The challenges of working remotely were very prevalent, but we soon discovered that working live with each other on calls and constant communication would be what would hold our film together. We decided a few weeks in that our equal collaboration of ideas was very productive for the beginning brainstorming, but we now needed a director to keep our vision clear and centered. After the initial modeling stages, we all agreed Do and Anne would be perfect to step up and help guide our style and message to be what we all envisioned. They reassessed our sets and created style guides, and we worked to make everything feel more cohesive, meaningful, and intentional.
Stepping back and prioritizing quality gave us the push we needed to finish this film strong. In addition to deciding on directors, we also realized we needed a leader to organize and rally us. We picked Kate Euting to be our producer, so she kept track of everything from assets, meetings agendas, task timelines, and more.
When creating our style guide early on in production, we decided to have our textures be realistic. Inspired by Pixar’s own films, we loved the combination of realistic surfacing on stylistic models and jumped at the chance to practice a similar style in our short.
Despite most of us never having used Renderman before, we decided that it would be the perfect renderer for our short. We had to learn how to use Renderman’s uber shader quickly. PxrStandardSurface is based on a specular -roughness workflow which is different than the Disney pbr metal-roughness shaders many of us were used to. Most of the texturing was done inside of the Substance suite of software, which thankfully had a Renderman export preset which automatically saved out the maps we would use the PxrStandardSurface inside of maya. The map’s exported out of substance allowed us to gain an understanding of how the PxrStandardSurface parameters drive the final look. While we were able to do about 90% of the textures inside of substance, some objects, like the gas station ground, needed to use Maya’s shader network nodes for the surfacing so we could iterate on the final look faster.
When working on the textures, our mentors taught us to always think about the story of our objects. Every object has a history, and even things coming straight out of the factory have scuff and marks on them.
Much like modeling and lighting, when we started texturing we did not have any formal director to help oversee the color script. Many of the objects in the scene had texture that clashed with each other because they were not close in value. Earlier versions showed sharp changes in colors and contrast when transitioning between scenes. Making Do and Anne directors really helped make all the shots cohesive. They had a broad overview of the entire short, so they could help with each object and let us know what color palates should be used for each scene.
Collaborating on set dressing remotely meant a lot of Zoom calls and iterations. It was one of the times where we could feel like we were actually in the studio together, backseat driving each other in Maya and moving props around live. Not having any characters gave us the chance to really dial in on composition, like it was a series of still life paintings.
Every placement, shape, and pattern had to tell our story. For example, the attic needed to look messy but still focused. Through composition and lighting, we made the same set tell two different perspectives of our story. The first shot focuses on the boxes and toys, while the last focuses on the lamp, even though the camera looks at these from similar areas in the set.
Another example is the wall of the diner. At first we wanted it to look messy as well, but after stepping back and doing some more research, we realized that the order that a person would add pieces to the wall would affect where they were placed over time. This meant we needed to decide the general order of when everything was hung, so it would follow a natural, recognizable pattern.
Lighting was a key part of our theme, so we wanted to spend a good amount of time perfecting the looks. The goal was to have the camera feel like it was searching the scene for the light, so the dynamic nature of the lighting was important to nail down. Miguel Zozaya and Eman Abdul-Razzak taught us so much about realism, intention, and subtle, magical techniques.
Problem Solving —
The very first shot was meant to portray our message in its truest form. While the shadow seemed ‘scary’, it’s actually just being cast by a few random toys and you guessed it, light! Dark and light cannot live without each other, and even the scariest of shadows have light to be found.
Achieving this shadow on the box actually created a few problems. Using Renderman with the render farm stopped us from successfully using a light blocker for the shape of the shadow, which is how the original lighting was set up.
After troubleshooting for hours and hours, we decided to do it in compositing instead. The shadow is a card in Nuke with the exported scene camera in Nuke’s 3D space. This workaround provided efficient control of color, opacity and blur.
The second and third shot both had the purpose of driving home the idea of light in “unexpected places.” A dirty, rundown diner with ripped up seats and a cluttered wall was a perfect opportunity for us to turn something unremarkable into something beautiful.
Adding the floating dust was a big part of the look as well, because this diner needed to feel old and neglected with the light transforming it. The beam of light shines directly on an unexpected $100 tip left on the table, an example of somebody spreading hope and joy to this empty restaurant.
The same kind of story is going on in our barn scene. The cool, quiet barn provided the proper contrast for a significant mood shift. The clouds part in the sky and the warm light pours in, illuminating the unexpected but beautiful piano. The most important part of this shot was to emphasize that while the barn was old and had holes in the ceiling, those holes provided the opportunity for the beauty that is these beams of warmth and light. It was incredibly fun to deep dive into storytelling through lighting for this project and also play with volumetrics.
The fourth shot takes place at one of those isolated, beaten down gas stations that you only notice when you’re stopping on a long road trip, feeling like you’re in the middle of nowhere. The scene begins drearily with thick fog, rain, and trash on the ground. However, once the space widens to reveal the convenience store, we see a beautiful light show created by the neon signs sparkling colorful lights into the puddle reflections.
For the final attic shot, the thunderstorm has subsided and we’re left with the moonlight’s lilting glow brightening the night sky. We wanted to achieve an atmosphere of calm serenity, one that would inspire awe in the viewer and leave them feeling hopeful for the future.
To achieve the colorful caustic effect from the glass lamp when the moonlight shines through it, the upper portion of the lamp was duplicated, repositioned and scaled in random directions to experimentally find the desired visual product. After turning off its primary visibility, its rotation and exposure was then animated subtly to ‘cheat’ the feeling of it coming to life.
Challenging ourselves to learn and use Renderman was no small feat, especially when using a remote render farm not intended for it! Hogan stepped up and was able to continuously troubleshoot and painstakingly work with his render farm to get our shots out. Over the course of 3 months through classes, holidays, and every problem you could possibly imagine, he did it. We owe Hogan our lives.
Our friend, Eugene “Vincesmoke” Jeong, custom produced our music and we couldn’t be happier. We gave him an early version of our animatic for timing and explained our message and he quickly pumped out tracks for us. Lofi with glitter never sounded so perfect!
Original Score by Eugene “Vincesmoke” Jeong
End Credits Music: “Lounge” by HoobeZa