We caught up with Jakub Bednarz, who graduated last year from Gobelins, l’école de l’image in Paris. Where he spent two wonderful years on a Master of Arts in Character Animation and Animated Filmmaking Program in English.


In the second year we dedicated the whole 9 months to our graduation project. At Gobelins only a few people each year are allowed to do a personal graduation film – mostly participants of the exchange program with CalArts. The rest of us, team up in groups of 4-7 people, develop original stories, visuals, and produce a short film with the help of teachers and professionals.

The year is divided into segments. After each one we present our progress to the jury and receive comments. From the beginning of the year until Christmas we work on our storytelling and visual direction, and in the second part of the year we craft the animation, backgrounds, assist at recording the sounds, foley, and we compose the film together. Nothing ever goes exactly as planned, so I wanted to share with you my personal experience of working on ‘DOGS’ in a team of 7 men.

The first mistake I made, right at the very beginning of September, was to overthink my initial concept.

Conception

Concept for the failed pitch


The first mistake I made, right at the very beginning of September, was to overthink my initial concept. We started with a short pitching session. We were supposed to present our ideas to find people with matching tastes or expectations for the team. Soon enough I had to learn, that in a group, where everyone is equal and where there’s no clear hierarchy, you have to leave space for other people to express themselves. You need to trust that they’re also as competent in what they’re doing as you believe you are.

I came in with an already built story, that I was thinking about during the holidays. It talked about a stray dog and an angel—a messenger of hope. Only three things stayed in from that initial concept—the character of the dog (although consistent only in its species), the narrative style of magical realism (where in a realistic setting irrational events happen not as sign of psychosis in the mind of the character, but rather as a poetic commentary on reality), and an aim—to focus on telling a meaningful story first, rather than creating a film as a pretext, to showing off creative abilities.

Our team was loud. We were all people with strong opinions, and we held daily debates on big subjects—truth, love, war, society.

Few people wanted to commit to an already created story that early on, so my first pitch failed. The team that we gathered later was composed of people, who either joined on condition of abandoning the first storyline, or didn’t have anywhere else to go. We were seven men, who knew each other from last year, but weren’t especially connected by any bonds different, than those of the current predicament. Nevertheless, very quickly I found out that working with Mohammad Babakoohi, Benjamin Berrebi, Mathinus van Rooyen, Diego Cristófano, Karlo Pavičić-Ravlić, and Théo Lenoble was a blessing.

A color and texture moodboard. We prepared these to get a hold of the atmosphere of what we're heading for in the film.

Our team was loud. We were all people with strong opinions, and we held daily debates on big subjects—truth, love, war, society. We created among us a peculiar balance. I remember feeling distinctly more confident when the whole team was there to discuss a topic, than when someone was absent. I loved the ideas of Diego, but I knew they sometimes needed to be countered by Marthinus. Benjamin was a type of pragmatic. He often brought the elevated discussion down to the ground with raising doubts. It which was annoying, but also essentially helpful.

It took us a long time to settle on what we really wanted to touch upon in the short film.  It was finally ‘the human conscience in the face of violence’. Some of the biggest challenges at this point were to keep the balance between telling a story, that is true to the characters, and telling a story with a pre-established idea, which we wanted to convey. Falling too much to one side risked having a shallow plot devout of meaning. Falling to the other would mean telling a morality play. We watched too much Tarkovsky (which made us look extremely pretentious) and we were all afraid that writing down certain story points could risk over-defining our storyline, thus loosing a sense of mystery. Even so, we had to accept, that, as authors, we needed to know what happens behind the veil of mystery, even if our spectator would never discover it himself.

Things look very different in the script, then on the screen, especially to untrained eyes, like ours. While writing, it’s easy to make a film more literary than cinematic. At one point we had a scene where our protagonist, sitting in the tree, looks around and appreciates the serenity of nature. He's trying to convince himself, that, maybe, he could escape the war and stay there forever, living a purified life of a Rousseau’s bon sauvage, the noble savage. This idea didn’t translate into the screen. It was boring and hardly readable.

Historic photo moodboard to help visualize the setting and composition.

Our plot went through twenty different versions before finding any ground of what will happen in it, who’s the protagonist or where the action takes place. It moved from a village in the Middle East, to Poland under the Soviet occupation, to Warsaw during Warsaw Uprising in 1944. It went way over the 5-minute limit of length imposed by the school. It wasn't unexpected, because so did every other graduation project at that time. Some of them later got shortened enough to fit the designated timespan. Ours got shorter, though not to 5, but 7 minutes.

I was the person who exploded with anger and threw objects flying across the classroom.

We also never really stopped discussing the script, even very far into the actual production. While animating, we discovered that a lot of ideas, that we talked about in preproduction, risked being altered, if the character acted even slightly the wrong way. We reiterated unceasingly on the killing scene. We acted it out in front of the camera, changed its dynamic and gestures. I remember a heated discussion around one Diego’s shot where the character, carrying a rifle on his shoulder, jumps over a tall fence.

Martinus wanted to have the strap of the rifle go from the shoulder across the chest of the protagonist, like every normal person would wear it. This would prevent the rifle from wiggling around, while climbing. To me it felt too much like Indiana Jones, and I was firmly decided that the protagonist should carry the rifle on one shoulder. That way it could fall from it clumsily, when jumping over the fence. It would make him look less powerful and less competent. I don’t think we resolved this issue very peacefully. In fact, this was the only time when I just stubbornly stuck with my opinion until the other side gave up.

Early concept art by Mohammad. In the early production it was essential not to think too much, whether a style is doable in animation form or not.

We were all quite stubborn, but not totally uncritical about our own ideas. We chose carefully which ones to hold on to, even if they were criticized from all sides by our professors. We understood, that some good concepts are delicate, before they are properly developed and grounded in the story. I think it’s not always wise to throw away a fresh idea that you feel good about, not necessarily knowing why, just because it doesn’t immediately fit in your current version of the script.

This was the case of the burning tree, the centre of our mystical scene. Even now, when we talk about it, each of us has a favourite interpretation of what happens in it. These interpretations don’t go against each other. Rather they emphasize on different aspects of what might be understood from the scene. Although they’re also not all the same and none of them explains the sequence holistically.

Production

My colourscript for 'DOGS'. Upper part contains thumbnails of selected shots in the sequences. Below are all the characters with colour pickers and their progression of colours throughout the film.

The visual development and shot production phases of our short film were a time of brotherhood. We never held any loud debates on our stylistics, but agreed on the visual direction quickly and without any problem. That doesn’t mean other groups didn’t face issues with their visual direction. It was just the specific dynamic of our group. Everyone brought in what he knew he could do best. For example, the backgrounds were clearly the domain of Mohammad.

We left him a lot of freedom to dictate how they’re supposed to look like. We established a hierarchy over different parts of the production. We divided responsibilities in animation, storyboard, background and compositing. I had the last say in animation, I fabricated the colour script and collaborated on visual direction with Mohammad. I gave up my responsibility over storyboard, editing, and compositing. Later on, due to approaching deadlines, it all got mixed anyways, and we worked on a little of everything.

I learned one important lesson later. It was something our teachers repeated continuously—‘expect the unexpected’

The biggest stylistic challenge on ‘DOGS’ was how to merge our characters, animated in TvPaint, with Mohammad’s oil painted backgrounds. We decided to abandon the line art and play with turning our characters into shapes of colour—thus hiding and uncovering features of the face or clothing as we please. We hid the eyes under the eyebrow shadows, to emphasize on the one story moment, when a direct light from the burning tree uncovered them to the viewer. This graphic solution felt immediately attractive, but it also meant we had to iterate a few times on the face design of our protagonist. The lack of facial features made him look  characterless and uninteresting, so it needed more complexity.

A background painted by Mohammad in oil and retouched in Photoshop.

For some time we were trying to develop an innovative pipeline, that integrated machine learning technology, to render our images closer to the oil paint. In concept, we would animate the character in flat colours and then hand repaint the keyframes, adding texture and colour transitions. Then the machine would render the rest of the images, by copying the style of the keyframe onto the rest of images. Even though the project showed promising results, we had to abandon it early in production. We were lacking the time and computing power. After the release, we discovered, that someone else was working on a similar technological solution. They released it too late for us to take advantage of it. One might say—we were just thinking ahead of our time!

Test of the machine learning pipeline. On the right is my flat coloured walk cycle test, on the left is Karlo's machine learning result. Background is a Mohammad's oil painting.

We counted the workload of animation, then the remaining time, and assigned a 6-day average length to finishing a single shot. This meant, we could allow ourselves for 1, max. 2 feedbacks on a shot from our teachers. We had to make a lot of decisions quickly and on our own. Rational thinking, planning, and spreadsheets were a helpful remedy to a spreading panic. We owe for these a lot to production students, Elisa and Daniel, who were assigned to help us with the project.

The key to animating efficiently was the preparation for each shot and recording good reference videos. In the style we chose, there weren't supposed to be any squash, stretch, or the shortcuts of  anime style snappiness, minimal movement, and static poses. We wanted to achieve believable motion through accuracy in body mechanics and attention to detail. We designed the motion for some sequences very early on. While storyboarding the killing scene f.e., we encountered many problems with the staging and expression. To solve them, we locked ourselves up in a reference room. For an hour we improvised on how this situation might unravel on the screen. We tried to grasp what feels natural for the characters to do in certain specific situation.

In a small production, like this, it’s nearly impossible to anticipate everything, that can go wrong.

I realized that the best shots, that I animated, were, surprisingly, the ones, where I was already used to the rhythm, the method of animating, and worked nearly mechanically. Preparing the shot and shooting the references took me from a few hours to a day. I then sketched a few loose poses, usually close to the reference, and filled in the gaps with quick gesture drawings. Once I was sure of my timing and body mechanics, I finished up the rough animation and exported it into PNGs. I then imported the sequence into Blender and traced the movement of the character with a digital model of the head and the rifle. We used these 3D models to avoid problems with deforming our characters. Both face and the rifle had specific volumes, that were easy to get wrong, while drawing without a reference.

After tracing the movement in 3D, I exported it into another PNG sequence and brought it into TvPaint. With the help of the digital model and my rough animation I then proceeded to draw the tiedowns. Firstly, I drew the keyframes, taking care of designing the shape of the shadow. Next, I filled in the gaps. All colouring was done on one layer, so there wasn’t need for differentiating between the body and shadow line art. I usually drew them in one colour on the same layer.

Our cleanup phase consisted of polishing the tiedown drawings wherever necessary. Lines wouldn’t be visible anyway and a little texture on the borders gave a desirable, handmade effect. With the finished lines, I coloured the animation, using TvPaint’s CTG layers. At the end, I used the Line Colourize effect to match the line colour to the fill.

In compositing a few more things was added to the appearance of our animation. Karlo created a template composition, that used the lineart and selective colours as masks to control the visibility of an After Effects plugin effect, Video Gogh. This added a bit of brush texture to our character. Shadows usually contained more of it, while in the light areas it was made less noticeable.

Karlo's 3D model of the head with examples of lighting setups, that helped with designing the face shadows in 2D.
The progression of animation in a shot I worked on.

Conclusion

They say hard times make strong men. The final weeks of our production were definitely a huge test of character and I don’t think I passed it that great. In these amounts of stress, it’s surprising how hard it is to let go of your steam. I spent days not being able to rest even while sleeping. Having breaks and leaving your desk for a while made little to no effect. Learning how to deal with one’s own stress was hard enough. Moreover, we also had to face the ways in which the others dealt with their stress too.

For one, I was the person who exploded with anger and threw objects flying across the classroom. For another, the only right way to finish the film was to rush for our deadlines. Yet a week before the end of production, some people in the group felt a different, revolutionary spirit.

They  suggested debating the final dates with the school. Such a solution, that meant moving the schedule of sound recordings, DCP conversions at external companies, and shifting the responsibility for our failures to the coordinators, was out of question. Still, it was something we had to discuss and make clear among us, while still keeping up our tempos.

One of Mohammad's comics about our team. The text says: 'The only team that needs more beers.'

I learned one important lesson later. It was something our teachers repeated continuously—‘expect the unexpected’. We were all too eager to claim the work finished after the deadlines for compositing. In fact, our compiled version of the film was all but completed. Every single shot contained different errors. Because we didn’t have any method for picking them out, we were constantly stumbling upon them it the worst desirable situations. Some of them we didn’t discover until they caused us a lot of trouble.

Others were discovered already after the premiere. In a small production, like this, it’s nearly impossible to anticipate everything, that can go wrong. Our last surprise was on the release day. The film was published simultaneously on YouTube, Vimeo, and Facebook. It was only then, that we realized, that a hidden setting in the final export made all three websites interpret the video as a VR and enable the 360º viewing.

This error made our film completely unwatchable. We had to immediately call the school to roll back the release and take down the videos. Then we had to rush to Gobelins to access the computers, re-export a correction, and make a new submission on the websites. Enough to give you several heart attacks.

I’m proud of what we finally achieved. ‘DOGS’ was an adventure and a great opportunity to learn, especially about people. Still, my first experience outside a school system, during my apprenticeship at Framestore in London, was to realize, that whatever I thought, I knew about teamwork, didn't necessarily apply to a VFX studio. I guess I’ll have to learn my whole life.

Gates of hell by Rodin. One of my early concepts for the ending scene.
You've successfully subscribed to Discover
Welcome back! You've successfully signed in.
Great! You've successfully signed up.
Success! Your account is fully activated, you now have access to all content.