Veteran Disney animator Eric Goldburg does not use the motion blur technique when he directs computer-aided animation. “I want the actual form of the object or character itself to give you as much fluidity as possible, as opposed to hoping that motion blur would make it fluid,” he says in Episode 30 of the Animation Podcast.

Goldburg pinpoints a common criticism of computer-aided animation, an issue that also happens to thread through all my animation. My scenes tend to veil shoddy animation with copious amounts of motion blur. Now I realize that motion blur can be to a bad animator like auto-tune is to Rebecca Black.

A simple example of the type of motion blur-less animation that Goldberg was talking about comes from this frame of his well-received short, Rhapsody in Blue.

The tennis ball is squished into an oval and the line trailing shows where the ball has been. It means that the lines still look solid. It also means that with one frame the viewer knows exactly where the ball is coming from. Using motion blur in my own animation yields the following result.

Here, it’s not easy to tell which direction the ball is going, the shape of the ball, or what we can expect the ball to do when it hits an object.

To practice, I used this piece of instructional hand-drawn animation.

This image provided a guideline for my first experiences using the squash and stretch technique in computer animation. There are two steps. The first step is to map the ball’s location and the second step is to deform the mesh.

By default, Blender smooths the motion path between keyframes. This means that simply mapping keyframes at the peaks and troughs of a bounce would create a sinusoidal-like curve. The yellow cuve below represents the vertical location of the ball and means that the ball slows down when it gets close to the ground.

To make the ball speed up towards the bottom of the bounce, I tweaked the vertical motion curve until the point where the ball hits the ground looked like a point in the motion curve as well. Now the yellow line looks appropriate.

The squash and streach follows the hand drawn example. When the ball is close to the ground, the ball stretches to simulate a blur. At the very bottom, the ball stretches to simulate weight and bounce.

I really like the result. The motion looks a little choppy, but it reminds me of hand-drawn animation. Most importantly, the animation speaks for itself: it does not need to hide behind motion blur.

Squash Ball Bounce Animation Test from Ian Elsner on Vimeo.

Many traditionally-animated characters express themselves with eyes that are able to deform from perfect spheres into egg-shapes. In most computer-animated films, though, character eyes are perfectly spherical. This is purely a trait of anatomy and the medium: it is much more difficult to represent a ball-joined sphere on paper than on a computer. But spherical eyes tend to convay expressions less expressively. A lot of emotional information travels through the shape of the eyes.

Realizing this, I made one final change to Laika’s model. I added facial deformers to the skin around her eyes, instead of just the lids and brows. The difference can be seen in the image above: on the left, I try to show surprise using only eyebrow deformations and on the right, I show surprise by adjusting the skin around the eyes to make them look taller and narrower.

Confident with the range and fidelity of Laika’s expressions, I completed the film’s animatic and began to set up the final animation and renders.

As I mentioned before, the relationship between Laika and her flight engineer rescuer relies on the visual metaphor of a game of fetch. The act of running free — it is essential that Laika is never leashed or tied — allows me to convey the trust that the flight engineer gives Laika and the loyalty she returns.

After the first scenes I describe in my last post, this dynamic guides the film through its second minute and its only montage. The montage is necessary because this sequence spans the time between Laika’s rescue in the winter of 1956 and the launch in the autumn of 1957. (For context, this is also the approximate length of time it takes to make both a human and this film.)

As Laika continues to play fetch, her doomed rocket steadily comes together around her. She is completely oblivious to its construction and instead cherishes the metals and pins which are cerimouniusly attached to her flight suit.

I designed the scenes so that the face of the Soviet engineer is never in the shot, a characteristic which is both convenient and artsy.

The night before the launch, the montage ends to a real-time shot of Laika and her flight engineer as the sun sets over the rocket. Every betrayal needs a Last Supper.

The next morning, the engineer throws Laika ball into the tiny capsule. This film is critical for story and animation. Laika hesitates at the entrance to her capsule, reluctant to go into a tight space, but eager to bring the ball back. As soon as she enters, the airlock doors slam shut. Almost immediately afterword, the rocket launches.

Laika does not realize what happend until she is perplexingly and inexorably in orbit.

I had initially planned to follow Laika’s journey from her lift off to her death, but I realized that the most compelling part of Laika’s story occurs before she enters the capsule. She was found the winter before the launch on a Moscow street. Her hardiness against the Russian cold was one of the major factors that lead her to her selection.

Surely an abandoned dog would have some sort of connection with her rescuer, with whomever showed her affection in the warm intior of a Soviet research lab. (This angle follows one of the storylines in Nick Abadzis’s graphic novel about the event.) The same researcher would use her trust to lure her into the capsule when the time came, betraying Laika

The idea appeals to me because it contains a story arc and two characters reacting to each other, both of which my initial version lacked. The visual metaphor of their interaction is a game of fetch, which they play throughout Laika’s training. The betrayal continues the metaphor: Laika haltingly enters the capsule to retrieve a ball and the door seals shut as soon as she enters.

With the complete storyboard laid out, I began creating the animatic. The animatic is a low resolution visualization of the entire film. Rendering is extremely time consuming, even more so with a thick layer of fur and realistic lighting, so the low resolution and the exclusion of shadows, anti-alising, and radiocity raytracing. Creating an animatic helps me outline the camera movement, the rhythm of the edit, and, most importantly, the character animation without the superfluous expense and distraction of a good render.

One of the most animotionally demanding scenes calls for a run cycle as Laika plays fetch with her rescuer. Rendering it as an animatic made it much easier to see problems in the animation.

Studying the motion, I realize that the run cycle needs to be more about the dog’s body than the legs. There should be a crunched but tuck in keyframe 2 and a full extension in keyframe 4.

And finally, here is an animatic of scene three, which shows Laika’s first consciousness after being rescued from the bitter cold. The lighting and coloring are inconsistent (the full render will look more like top image) and the arm model is a stand-in, but Liaka’s model and animation are nearly complete.

So far, my animation has leaned on situations rather than expressiveness to drive the story forward. My characters tend to eerily glide around my sets, forcing the audience to use musical and situational clues to guess how a character is feeling.

My latest project aims to eliminate this problem through its design. The protagonist pooch is confined to a tiny space and the animator is confined to her face to show the audience how she is feeling.

I approached Laika’s model with these restrictions in mind. Since she can’t walk or turn around, making armatures and detailed models for her feet is unnecessary. Instead, I was free to focus on her face, which requires the right mix of realism and cartoony expressiveness.

Initial wireframe model

The facial structure of a dog is elongated and low, as you can see in this reference image.

Everest, the reference husky

With expressiveness in mind, I raised the area around the eyes into only a slight slope down the muzzle. For the skin I used UV mapping, which I painted over a furry texture in the GIMP. The UV texture maps a 2D image onto a 3D object, as seen below, and ensures that the distribution of brown and white fur is seemingly random.

UV Mapping

I then duplicated the mesh. One became the skin and the other, with help from Blender’s hair simulator, became the fur.

Frizzy Dog

I had a hard time with the fur, in part because of limitations within the Blender simulation and in part because of the tricky guesswork involved in varying lengths of the hair over different regions of the head.

Slightly Better Fur

Much of the realism of digital fur depends on the lighting, so I turned my gaze to her eyes. They are based on the same models I have been using for years, with concave irises to collect as much light as possible from the rest of the scene. After experimenting with brown eyes, I decided on green eyes to match the beige green “70′s refrigerator color” of the hardware inside the spacecraft. I made the irises much bigger because dog’s don’t seem to show much white and to exploit their cuteness.

Big, green eyes

The nose is visually delicate. It still looks like it was simply glued onto the rest of the mesh, which is essentially what happened. One problem is that the skin was visible behind the nostrils, which can be easily fixed by a shadow of blackness. The harder problem was figuring out how to represent the pattern of tiny holes near the mussel.

Bigger eyes, wider tongue

I like the idea of imposing limits on my work to force a better film. There is no reason the audience should not read the dog’s face to figure out exactly how she feels about being launched into space against her will.

Much of the remaining work on the dog revolves around the animation. I need to decide whether to focus on realism or toonisim in Laika’s restricted motions. And so I expect the final mesh will undergo some minor motion-inspired tweaks before the final version.