During the previous production of Charge, the creation of assets was more straightforward (with some exceptions). There were a bunch of assets that needed to look and feel realistic in proportion. I modelled them and used our main character as reference, created basic UVs, and moved onto the next task.
But with Pet Projects I had to do a 180 and start skewing, tapering, scaling, and rotating objects I wasn’t used to. Creating stylised assets was a process that I had to familiarise myself with. Thankfully, Vivien Lulkowski, our 2D art director, created many draw overs and line drawings to help set the visual identity.
We knew early on that our goal was to emulate a 2D animated show in 3D. This is why in my experience, the intention of creating and placing 3D assets was more specific compared to Charge. When working on a task, Vivien and I had many back-and-forths addressing any big changes or the smallest iterations to refine the design. One of such iterations are the “improvised wings” where function, construction, and animation needed to be be established before finalising the design.
The improvised wings received different configurations of materials where the function and construction were established.
Rik created this animation to illustrate the movement of the wings. Especially useful to know which parts would bend and the up/down -ward motion of the base.
After the animation test, I blocked the wings based on sketches from Vivien.
Here is the draw over created by Vivien with design notes.
After minor feedback here and there, the wings are wrapped up and made ready for shading.
Before moving on, let's mention that at start of the production, I wanted to use Geometry Nodes to see how far I could push flexibility in a stylised production. This has its pros and cons which I will discuss later in this article.
The director on Pet Projects, Rik Schutte, initially gathered references and inspiration from other animated series and productions. With annotations he referred to the specific elements he envisioned for our world building and rendering of backgrounds, characters, and FX. Some of the points for the backgrounds and props are:
It took a while to recognise which elements of the model needed specific adjustments to make it look stylised. While Vivien was developing designs, I was mocking up models that were adjusted based on new design notes. Some assets required more iterations due to the complexity or to accommodate changes from the layout.
The rocket exterior is a good example of big proportional changes. The top section at the back of the rocket was lowered to taper downward, this made the front of the rocket a better point of interest. The exhausts were fanned outward to exaggerate the design further.
The most fun note from Vivien was on the rocket exterior, by drawing a beluga in a corset to illustrate how the rocket should taper towards the exhausts.
Exaggerating shapes and elements of an asset are important to achieve a level of stylisation. These include increasing the thickness and avoiding thin/detailed geometry, gradually inflating/bending/tapering objects, and rotating or scaling elements like handles or levers to fit the bigger asset they are sitting on. How much and where you apply these modifications affect the stylisation and general design.
The pilot chair went from a thin to an inflated design.
Changes were made to the back support, removed the headrest and refined the model overall.
This was the first non-character asset that we liked and felt fitting to the world.
Applying this principle to all assets wasn’t a must, but a good mantra to break away from circular and symmetrical static design. For example, the steering wheel, speakers, and some buttons are more oval than circular. The levers have a skewed ball shape with variations to the amount of skewing.
Taking a look at the toolbox we can see how symmetrical the initial version was. In the second iteration, with the help of Vivien’s design, I tapered the geometry in several places and added slight curving lines. Notice that the amount of tapering isn’t the same everywhere, adding to the stylisation.
Draw over and design by Vivien.
From a symmetrical straight rectangle to a skewed and stylised toolbox that fits the world.
Deciding which objects need more geometrical detail or textures was usually discussed during our daily meeting. The advantage is knowing how long we need to spend time modeling and what will be covered with texturing or in our case, Grease Pencil strokes to create detail.
Grease Pencil disabled.
Grease Pencil enabled, also, all surfacing is done by Simon Thommes.
Especially in the rocket interior we have assets that “respect” or adhere to each other's shape and directionality. Most elements placed on the consoles flow naturally along its surface. This flow creates a nice sense of rhythm in the design, some of these are the dashboard elements, “techpanel”, and the levers next to the steering wheel. But the opposite can be useful too, to break monotony, like the two consoles on the right hand side or the rocket air refreshner.
Before modeling and spending a lot of time, it is useful to communicate ideas using Grease Pencil. For the exterior of the rocket we needed panels and hatches that are indistinguishable from another. Using Grease Pencil to visualise a design is a good way to save time, plus any frustrations that might come with redoing everything.
Red lines are base panels, black lines are the hatches. Since we see the right side of the rocket most of the time, we didn’t spend time adding more detail on the left side.
Adding colors takes time but might help communicate or translate a design better. This was one of the first iterations of the rocket.
For the rocket exterior we have panels that fly/rip off showing the skeletal structure. Before adding all of the wires and adding smaller changes I did a small section first. After discussing this with Vivien we came to the conclusion that fewer wires were better to avoid visual clutter. Creating a small section first increases efficiency and saves time!
Example of a small section.
Final result.
As mentioned before, I wanted to use Geometry Nodes where possible to enhance the flexibility of our production. Creating setups that were somewhat intuitive but have limits to specific art direction. In the beginning of the production we had more time and so I was trying to create a preliminary list of props and start noodling.
A setup for switches and a stick shift.
Strictly speaking for creating assets, the advantages of Geometry Nodes are linked to the time you will save later on. That is why it is wise to determine if spending time creating procedural setups is the way to go. Some positives when using Geometry Nodes:
The bike belt mechanism was one of the few setups that is still driven by Geometry Nodes.
Not the most efficient but a quite flexible setup to generate a power cable.
Some assets were quite useful up to a certain point where applying the setup and manually changing things was faster. This was apparent with the warning light asset that required specific wires and to optimize it for production. With the twisted wires we didn’t have enough control for the animators. Applying the setup and rigging the wires was an easier solution. Some downsides to using Geometry Nodes this way are:
This setup allows changing the amount of loops, resolution, rods position, scaling of the bulb and more.
Twisted wires with a lot parameters, more flexible but also more complex to work with.
Activating these switches by proximity was fun but rigging them would have been more animator friendly from the start.
Adopting a style or design influences the creation of assets. In our case we used a large amount of art direction for every aspect of the production. This means:
Interested in learning Geometry Nodes? Watch this course by Simon Thommes
Thank you for sharing so many learned lessons and experiences and that chair does look pretty darn appealing.
cool!
Join to leave a comment.