World Building BtS

Last time, we mostly completed the Cellar Door Bar map, so we thought we'd share the creation process with you.

For "fault 2b", our key concept is "experiencing the story while walking through the world".

While it's crucial to create captivating concept art that serves as the foundation for the stage design, this process is quite a bit different from completing a single illustration.

Creating the Map

Since the players will progress through the story by walking on a 3D map, no matter how good the design is, it can't be difficult to navigate or have important elements off-screen during movement.

If a map is too large, it becomes tedious and difficult to fill the space, while if it's too small, it becomes hard to navigate once objects are placed, not to mention unsatisfying to play. We needed to really get a feel of the scale and then finalize the details.

To start, we used a 3D mockup tool called Probuilder to decide on a rough layout.

During the process, we found that viewing things from afar often led to scaling issues. We'd inadvertently create furniture that are either too large or small.

It was here that we realized just how crucial it is to always place a character in the map for size reference while designing it.

What seemed nice from an overhead view often turned out quite differently when we set the camera and walked through from a player’s perspective.

▼Player view during adjustments. Initially, the ceilings were too high, so the viewpoint was set way too low in an attempt to show the ceiling.

At this stage, no drawings were made; we only collected reference images. I imagined elements like "an aquarium under the table" and "a large window overlooking the sea". Munisix and I would discuss and confirm these directions using the rough layout created in Probuilder. This brainstorming process was incredibly fun.

We also established some basic camera work at this stage. The key elements I focus on are "the excitement of entering a room", "the ease of talking to characters", and "the ability to arrange various compositions within a single scene".

Especially, layouts that only look good from one specific angle should be avoided. Making 3D scenes allows for dynamic compositions that should not be hindered by static layouts.

Keeping these points in mind, we walked through the map created in Probuilder and tested various camera angles. Once everything seemed okay, we finally started the concept art process.

Creating Concept Art

After thorough research and adjustments, here is the concept illustration we came up with:

The details are modeled based on reference images, so the rough finish here is acceptable as long as the general idea comes through. Honestly, I still wrestle with how detailed to make these, but spending too much time on textures and fine details at this design stage isn't feasible. Since I'm handling both design and modeling for this project, this level of detail was sufficient.

At this point, my 3D skills were still developing, so often I was just going with gut feelings like, "I don’t know how to do this...but it looks good like this, right!?!"

I entrusted the future me with effects like sea illumination and bubble animations.

Despite feeling quite nervous inside, I kept pushing forward, often wondering, "Can I actually finish this?"

Now, Onto 3D Modeling! 💪

We took the model created in Probuilder and imported it into Blender.

To ensure there were no size discrepancies, we created detailed parts while reusing some components to save on effort where possible.

However, when arranging the parts, we realized that reusing too many parts left the space feeling empty. Quick fixes like creating makeshift shelves often ended up making the space feel sparse and oversized.

While I aimed to add uniqueness, working without enough reference material led to unsatisfying results. Here, the advice from 3D tutorials to gather around 100 reference images made sense.

Furthermore, I realized the importance of creating detailed design sketches before modeling. Jumping straight into 3D modeling leads to a lot of trial and error!

▼Size confirmation and rough material assignment for checking the overall impression

Texturing (Painting the 3D Models)

This map is predominantly made of wood and metal, with wood being particularly challenging.

The sheer number of material variations for wood is staggering—types of wood, polishing levels, oil finishes, paints, wear and tear from aging, etc. So, I gathered as many photos capturing "bar-like wood textures" as possible.

▼Some search results for "Wood" from Poly Haven, a 3D material resource site. As you can see, there are countless types of wood materials.

In the end, the concept art required over 100 reference images. Collecting these images took a significant amount of time, emphasizing the importance of ongoing research and reference gathering. While fantasy settings like fault allow for some leeway, historically accurate projects would require even more meticulous research.

Revisions and Adjustments

Once the model and texture were roughly completed in Blender, it was time to export the 3D model into Unity. This process required back-and-forth between Blender and Unity for adjustments—an undeniably tedious part of the work. Frequently, mysterious errors popped up, causing a lot of frustration, particularly when lacking specific knowledge.

Stumped by certain error messages, I often turned to ChatGPT for help. Helpful answers were provided about 30–40% of the time, so here's hoping it becomes more accurate!

Placing objects as per the Probuilder layout sometimes revealed issues, such as a missing window where the character should sit, leading to positional adjustments and the removal of some shelves (green arrow parts).

Additionally, due to insufficient light sources in the back, we added small items like sediment tubes.

Even at this stage, I had no confidence in how to create the aquarium under the table, which was quite stressful.

Despite these challenges, the concept art served as a general reference for the ambiance and silhouettes, while adjusting layouts during actual gameplay and character dialogues remained a priority.

Aquarium and Ocean

Creating water effects had been a persistent challenge. I wanted an ambiance where light diffuses and shimmers underwater, so my focus was to at least make the scene presentable. Shader effects would likely enhance it further...

For now, placing a blue-textured plane at a distance from the window surprisingly created a convincing look. I'd love to add swimming fish at some point.

▼Left: In-game appearance. Right: The sea is just a plane when viewed without walls.

We used tweaked assets for foam, seaweed, and coral-like models.

The successful light map bake ensured the vivid blue of the water.

While textures included colors and gradients, the final outcome heavily depended on lighting setups. Understanding how to use light and post-processing effectively could mask lower texture resolutions, making the overall scene look impressive.

Preparing decent visuals in Blender, then quickly moving to Unity for lighting and post-processing, and finally identifying areas needing improvement seems the most efficient, despite appearing cumbersome.

In Conclusion

Though the final look deviated from the initial concept art, we managed to preserve the core essence.

Reflecting on the process, I realized how much my drawing skills and spatial perception need improvement—the final size and color often differed from initial plans. Experienced concept artists likely have a clearer vision of the end result.

Working with Munisix helped identify unnoticed inconsistencies and shortcomings, fine-tuning the areas to focus on.

Thank you for reading this long journey. I hope this gives you an idea of how we are creating our games!

Hare Konatsu