Table of Contents Show
Archive notice: This article was originally published on September 1, 2018. Links and embedded videos are preserved as part of the historical record.
The protagonist’s fall towards the ground is a metaphor for the fatalism of life: we are born, come to our senses, move through life and inevitably die. The theme of “life and death” has accompanied Thomas Sali since his childhood, after his mother died in a car accident: “After my mother’s death, I came up with my own theory of being and existence, because as a child I couldn’t imagine a person disappearing. Even today, the thought of no longer being there at some point is still too unwieldy for my head. In my opinion, existence manifests itself not only through tangible and visible objects, but also through ideas, thoughts and actions that we leave behind every day and that change and influence people and things. Two years before the completion of the film project, I had a first, still very rudimentary idea for ‘On Your Way’ and was looking for concepts that visualise my concept of life and death.”
In the scenario, the pieces of rubble represent thoughts, ideas, memories and the knowledge that people acquire in the course of their lives. “This is how we create our own reality. That is our personality. That’s what the rocket stands for in the film,” says Thomas.
Team building & planning
When Thomas presented his idea in the weeklies at Animationsinstitut, fellow students Philipp Mekus and Marcel Ruegenberg joined the project. Philipp took on the role of VFX supervisor and Marcel that of technical director. With Eric Sonnenburg, who joined as producer at the end of the year, the core team was complete. As “On Your Way” is a live-action film, the team continued to grow until the shoot in July 2016. By the time the film was completed, around 70 people were involved in the project in all departments, with the core team working continuously on the project for around a year and a half.
In order to keep to the production schedule, the team worked to ensure good organisation with various production plans for pre-production and shooting, a detailed shot breakdown, which was created after the film was edited, and a production plan for post-production based on this. Here, the team noted which assets had to be built and which tasks had to be completed in the individual shots, as well as which artist would take them on – naturally with a schedule for each task.
The production schedule was initially updated weekly and later at shorter intervals. Thanks to the shot breakdown, time-wasting tasks could be quickly identified and realistically planned:
“That’s how we got all the shots through to the end without the threat of burnout. That’s important for me to say, because even though it’s the greatest thing in the world for me to make my own film, it shouldn’t be done at any cost,” says Thomas.
Good preparation is half the battle
In order to keep the initial idea as raw and unfiltered as possible and to waste as little time as possible, Thomas initially only focused on a few images: He created a portfolio of mood images and a mood film consisting of internet footage. The aim was not to achieve good visual quality, but simply to preserve the initial idea. “It was a very helpful tool that allowed me to check whether I was straying too far from the original idea at an advanced stage of production,” recalls Thomas.
The artist then created Photoshop and more precisely elaborated 3D concepts. While Thomas was busy creating the storyboard, Philipp and Kiana Naghshineh continuously created further concepts: “The storyboard images were created in the classic way with pen and paper so that I could work on the filmic resolution in a raw way and had no inhibitions about simply replacing unsuitable images with new ones. It was important to me that the story took centre stage in this phase of the production and that no vanity should arise with regard to the storyboard drawings, which as a designer you fall in love with too quickly and then find it difficult to detach yourself from them.” The storyboard provided the basis for the first animatic with static images.
However, it turned out that the animatic was difficult to understand this way, so Thomas rebuilt and animated all the storyboard images in Cinema 4D. Especially in view of the fact that the team would continue to grow and the animatic would be the basis for the work of everyone involved, it had to be as precise as possible so that everyone could interpret it correctly. “I’m not normally a C4D user. But the Toon Shader is a perfect way to make the animatic look visually raw and cohesive, which was crucial for clarity. It was much easier for me to make the most important things comprehensible in 3D than in a 2D drawing. Being able to work with lights in 3D was also a crucial tool for the animatic. Because there are virtually no fixed spatial reference points in the film – after all, the protagonists are in free fall – the sunlight was the only constant that helped with the spatial orientation,” explains the director.
Thomas spent two months refining the animatic, with cameraman Martin L. Ludwig helping in the final phase. Meanwhile, Philipp used the animatic as the basis for the tech-viz, which was used to plan the shoot in advance in the studio, which proved very successful. “This meant that problems could be solved before they even arose,” laughs Thomas. “Things like wire rig systems, setups for the production design, the use of bluescreen or the rear projection – Philipp planned all of this in advance so that we could concentrate on the tasks on set from the first day of shooting.”
Shooting with Ncam and rear projection
As one of the biggest challenges of the project was the high number of shots – around 60 shots with a film length of around 6 minutes – the team wanted to shift as much as possible to pre-production and shooting in order to take the pressure off post-production. “We achieved this by shooting a third of the shots in front of an interactive rear projection. To do this, we produced matte paintings before the shoot, which we imported into Unity and projected onto geometry. During the shoot itself, we had the real-time AR system Ncam at our disposal, which transmitted the position of the camera live to Unity and always shifted the projection content so that it adapted to the camera’s line of sight. With this technology, it was as if the camera was looking through a window rather than at a screen – so the background appeared through the camera as if it were kilometres away. This worked surprisingly well most of the time – except when there were too many moving objects in front of the Ncam and it could no longer track them reliably. In these cases, we had to help manually,” says Thomas. “Nevertheless, the method meant that many typical blue screen shots no longer had to be edited in post-production, or only slightly, which saved an enormous amount of time.”
The idea of shooting with Ncam and rear projection caused the most scepticism among pitches during pre-production. “In retrospect, it was the only solution to be able to realise the length of the film at all,” Thomas is certain.
Digidoubles and rocket parts
Digidoubles were required for some of the full CG shots, which the team created by scanning with Agisoft. The rocket parts were also created using photogrammetry. Production designer Leonie Lieberherr built four large side panels and the interior of the rocket for the shoot. Marcel scanned them and they also served as reference models for the missing rocket parts, which were modelled in Maya and textured in Mari. This task was one of the most complicated because Thomas didn’t have a precise idea of the engine at the beginning: “That’s why I combined the concept phase and the modelling for this asset and tested directly in 3D whether the engine matched the rest,” says Thomas.
Photos from the photogrammetry photo booth were used as textures, as the photoscan textures were not detailed enough. Most of the textures were available in 4K resolution, which was completely sufficient because the characters and objects were subdivided again and full CG shots were only used for long shots. Only the interior of the rocket engine was in 8K resolution because it was used for a close-up shot. Thanks to the precisely planned animatic, which ended up looking more like a previz, the team knew exactly which assets would be seen up close, which saved a lot of time in post-production. Not only because the data remained smaller, but also because many texture details require exponentially more love and time.
Animating a fall to the ground
“To get a real feel for falling, I guess you really have to fall – there’s no way around it. That’s why we did a parachute jump with the actors shortly before filming began. Philipp and Martin were brave enough to jump with us. It was out of the question for me,” says Thomas. “As a result, we had good reference videos of people falling that we could use for the animation.” The team also studied many other skydiving videos and Thomas and Philipp discussed and analysed every detail with the animators until everyone was satisfied with the animation. There were different depictions of the falling movement: Sometimes the camera moved at the same speed as the people, then hardly any movement is visible or only on the simulated clothing. Then there are shots in which the camera remains fixed at one point and only the characters move. This makes the falling look extremely fast, which was used to depict the panic.
As the digidoubles could only be seen in the long shots, they didn’t need fine rigs: they consisted of a simple biped rig focussed on the body. Only the fingers were details that had to be animated because they had to be able to grip.
All-purpose weapon Houdini
The simulation of the digidouble hair and clothes, all the debris and some of the clouds were created with Houdini: “As most of the simulation tasks were applied to finished animations that we imported into Houdini via Alembic cache, it made sense to do the finalisation in Houdini as well. Marcel simplified the shading and rendering processes with specially developed scripts so that even artists who were not yet familiar with Houdini could work with it after a short familiarisation phase,” explains Thomas.
Shading, lighting & compositing
Arnold was the engine of choice for rendering. Another important reason was the use of Anders Langland’s al-Shader, which was particularly interesting because Psyop had recently released Cryptomatte, which greatly supported the compositing process.
The lighting process ran smoothly as there was only one main light, the sun. The team used a classic HDR to brighten up the background, while a directional light provided the sunlight. As the students were able to use the Film Academy ‘s well-equipped render farm, the team didn’t have to worry about render times. Other reasons were the 2K resolution and the clouds, which were mainly created with matte paintings and therefore hardly mattered when rendering. The most complex shot was the final long shot in which the rocket disintegrates.
The compositing was done with Nuke because Thomas and Philipp have a lot of experience with it: “With Nuke Studio, we also had the option of bringing all the comps back into the edit, which is great because all the shots can be viewed in context. You can see the forest again, so to speak, and not just the individual tree, which was important for Philipp as VFX supervisor and for me as director to be able to make decisions.” The biggest challenge in compositing was to raise the quality of the rear projection, bluescreen and full CG shots to the same level so that the overall effect was consistent and there was no difference in quality between the backgrounds.
Philipp also made sure that the film look remained technically consistent. To achieve this, he wrote tools that made the same effects available to all artists – e.g. Nuke gizmos that simulated the camera filters used for incamera shots. This meant that all full CG shots could be optimised to match the shots that had been shot. Despite the good filming preparation, shots still had to be corrected or partially relit so that they fitted into the edit.
Another major challenge was the interaction of the actors with CG elements – a task that Thomas was nevertheless happy to take on in favour of the storytelling: “We tried to avoid these cases as far as possible. But a piece of debris that the protagonist caught was so ingrained in the story that it was unavoidable not to show these scenes.”
Opening with radio play
“Right at the beginning of the concept phase, I had the idea of having the audience sit in the cinema in complete darkness for the first minute of the film. I came up with this idea because I had heard of an orchestral piece in which the musicians don’t play for several minutes at the beginning, but simply sit there quietly and all the sounds that arise during this time form the piece. The initial darkness deprives the audience of what they expect to see in the cinema: images on a screen. The audience should be trapped in the darkness like the protagonist. No image, just incredibly loud sound.
In the best case scenario, the audience is irritated, only to wake up from their swoon together with the protagonist with the first image. That’s why I decided to open the film only with radio play without images – even at the risk of confusing the audience. I see it as a kind of experiment,” explains Thomas.
To create a film world that completely captures the viewer, the team decided to work with Dolby Atmos. This made it possible to create spatiality with sound.
Volker Armbruster, who realised the sound design and final mix for the film, built a microphone rig especially for the wind recordings. These recordings were then mixed with procedural wind noises until the right sound was found. Dolby Atmos made it possible to tell completely different sound perspectives: With a spectrum that ranged from extremely loud and very direct wind noise to very subtle, distant wind noises.
Experience gained
The main thing Thomas Sali learnt from the project was that as a director you have to hand over tasks and responsibilities: “That was the biggest thing I learnt from the project. Being a director, art director and VFX artist at the same time is a lot of fun, but it is difficult to coordinate. It’s difficult because a director has to keep an eye on the overall concept and at the same time can’t spend days digging around in technical details. Today, I think that as a director you need distance from the realisation. Not because I don’t value these processes enough, on the contrary, but because you can look at the film more from the viewer’s perspective.”
Thomas would also approach a few technical things differently today, such as the use of Ncam: “We did do tests with Ncam in advance, but never with so many people and moving objects. That’s why we couldn’t always shoot with Ncam, because it didn’t track properly. I will still consider the technology for further projects, but I would carry out the stress tests more extremely.” Overall, Thomas is very satisfied with the process and the end result: “All the production processes built on each other well, which was due to the fact that most of the team members had already worked on film productions at the Film Academy and our preparation was very meticulous.”
