With Project Gold wrapping up earlier than anticipated, the studio team found themselves with some open time in their summer schedule.

I took this opportunity to propose an idea: to initiate small projects focused on Grease Pencil 3.0 development. The concept involved creating a series of short, 10-second animated clips, each exploring a unique style and technique using Grease Pencil 3. These clips would serve a dual purpose; identifying bugs by thoroughly testing the tools during production, and acting as a captivating teaser for the tool’s release in November. Additionally, the project files would be shared with the community, allowing others to explore and learn from them.

This was one of the rare occasions when studio projects could align with development, allowing newly implemented patches to be tested directly within production files. Typically, tools required for a production are only implemented after the project is completed, due to a pre-existing list of planned tasks. The goal of each project was to test a unique aspect of Grease Pencil, exploring its integration with other tools and environments in Blender. This approach ensured that Grease Pencil's compatibility and functionality were thoroughly evaluated in various contexts.

The articles to the other Grease Pencil projects: Cowboi, Color Void and Project Gold [links to come as they get published].

House of Chores

One of these projects is House of Chores.

With this project, I wanted to create a rich, animated diorama that could serve as a benchmark for performance testing while utilizing a variety of tools in combination with Grease Pencil. This concept ultimately earned its place as the splash screen for the Blender 4.3 Beta.

The idea behind House of Chores is to depict a self-sustaining house that takes care of all the chores automatically, allowing its owner to focus on making music and being creative. It serves as an analogy for the role AI can play—supporting human creativity and handling routine tasks, rather than taking over completely. I envisioned the style to be loose, with a hand-drawn quality reminiscent of illustrations in a children's book. Some of the visual inspirations include the video game Machinarium and the drawings of Philippe Fix.

Converting a 2d sketch to 3d artwork

After creating the initial sketch on a 2D plane, I began modeling simple 3D geometry to define the main shapes of each asset. The outlines (or inklines) were created entirely using Grease Pencil, with the lines projected onto the 3D geometry. A particularly useful feature of the “drawing on surface” option is the ability to select specific objects for projection mapping. This is especially helpful when dealing with overlapping geometry, allowing you to control where the lines are applied and avoid drawing on unwanted surfaces.

Texturing

The initial color pass was done using Blender’s texturing tools. The choice to use actual textures instead of Grease Pencil was made to enable wrapping the colors around the 3D geometry using UV coordinates. This approach isn't currently possible with the Grease Pencil fill tool, as it is designed only for filling flat surfaces. Ideally, a geometry-based fill tool could be developed in the future, allowing for easier filling of larger 3D shapes.

Using texture painting also provided the flexibility to create organic silhouettes for each object, rather than being limited by the geometry itself. Ultimately, the texturing and detailed coloring were done in Krita by Vivien Lulkowski.

Animation

The goal for animating assets was to utilize a diverse set of tools within Blender. This approach allowed me to thoroughly test each feature of the new Grease Pencil, as well as evaluate the conversion process to the updated version. Ultimately, I aimed to explore the full range of Grease Pencil's capabilities, identifying bugs and using this project as a benchmark for future testing.

Some objects were traditionally animated by hand-drawing each frame. For example, the trumpeter on the balcony was animated this way to suit the character's design and movement. Other elements, like the robotic arm stirring the kettle, as well as the smoke and fire from the kettle, were also hand-animated.

I used simple bone rigging for animating mechanical assets like the water bucket, catapult, and net. Some elements, such as the extended scissors and the paint roller, combined traditional animation with bone rigging support. Initially, the bone rigging wasn’t fully compatible with the new system, but after some adjustments, I was able to create rigs for animation.

Line Art Modifier

One of the few fully 3D animated objects is the waterwheel. My initial plan was to trace each frame and create a loop, but the perspective shift made this approach too complex and time-consuming to animate by hand. This challenge provided a good opportunity to use the “line art” modifier alongside the 3D waterwheel, blending it aesthetically with the overall style. To enhance the hand-drawn feel, I applied a noise modifier to the outlines.

Grease Pencil and Geometry Nodes

A major update with Grease Pencil 3 is the support for Geometry Nodes. I had to integrate this feature in a way that was manageable without turning it into a separate project. The chimney smoke plumes were an ideal candidate for this. I created a single stroke and applied an animated noise modifier with an upward direction. To achieve a semi-transparent effect, I added a color ramp with an alpha channel to the material. Using the "Step" math node, I converted the procedural animation to run on 2s, aligning with the style of the other animated objects.

The water surface was initially painted by Vivien and then enhanced with an animated shader distortion by Julien Kaspar to bring it to life. I also hand-animated ripples along the edges to integrate it smoothly with the rest of the scene. Additional hand-animated effects include the water on the waterwheel and the flow of water from the sewer.

A collaboration with development on Blender 4.3 alpha

When we began working with the new alpha build of Blender early on, we knew it would be highly experimental. This often meant that basic features would suddenly stop working or cause Blender to crash instantly, especially when new features or functionality were being added. While this could be cumbersome during project work, it was all for the greater good—identifying bugs and contributing to the improvement of Blender’s functionality.

We frequently communicated with Falk, sometimes on a daily basis, to report bugs and discuss user interface issues, tool behavior, and potential solutions. This close collaboration was mutually beneficial: it allowed us to stay informed about ongoing developments, and it gave the developers insight into the challenges we faced in a real production environment.

To keep track of bugs and features still in progress, we created a spreadsheet to differentiate between actual bugs and issues that were simply due to unfinished features. Working with the alpha version of Blender inevitably slowed down the project due to these unexpected hurdles, but that was one of the primary goals of the effort—to uncover bugs that wouldn't normally surface in a sterile testing environment with an empty blend file.

We hope you enjoyed this read, and feel free to share any questions or feedback you might have! If you want you can download the file below.