Monthly Devlog: Integrating 3D Assets into a 2D Hand-Drawn Workflow

Overview

This month’s primary focus was on blending 3D environmental assets—particularly terrain and overworld elements—into a predominantly 2D hand-drawn sprite workflow. The challenge was to leverage Blender’s procedural and semi-procedural tools to enhance visual fidelity and production efficiency without losing the handcrafted aesthetic. Alongside technical experimentation, I also addressed workflow infrastructure improvements and began laying groundwork for formalizing the business side of development.

Highlights & Progress

Asset Integration & Visuals

• Developed a hybrid approach for grass assets by drastically reducing density (fivefold) and switching from fully procedural 3D blades to 2D polygon sprites with rotation limited to 0° or 180°, improving visual consistency and production speed.
• Enhanced rock textures by combining low-poly models with noise texture displacement and decimation; planned a new low-poly rock creation method using assembled cubes on a grid to streamline asset creation.
• Explored shader-based edge detection for vegetation to replace Blender compositor effects, aiming for more efficient and flexible real-time visuals.
• AutoTerrain system successfully produces organic terrain up close but reveals a cartoonish style when zoomed out, highlighting a need for further refinement.
• Encountered difficulty modeling buildings in Blender; decided to limit Blender use to terrain and environmental elements, handling buildings manually or through rough 2D shapes to maintain artistic control.

Procedural Systems & Challenges

• Utilized Blender’s geometry nodes for procedural grass, cobblestone textures with noise, and water ripple effects, pushing procedural tools while balancing hand-crafted look.
• Created an auto path system in Blender generating grid-based paths with Bezier lines and randomized noise; instanced tiles by grid coordinates.
• Faced a significant challenge with path point classification: height-based classification risks exponential growth in tile variants (from 13 to over 200), threatening manageability.
• Blender’s bug preventing path redraws required a workaround, adding complexity to the pipeline.
• Investigated ray casting and shader-based solutions to prevent grass blades intersecting with rocks, as density reduction alone was insufficient.

Game Systems & Design

• Conceptualized a dynamic difficulty system for a Metroidvania-style game that adjusts challenge based on player progress and deaths, including tweaks to enemy movesets, health/stamina scaling, and resource availability.
• Began prototyping subsurface modifiers on flat terrain to add organic variation without losing discrete height changes.

Workflow & Infrastructure

• Upgraded hardware to an AMD AM5 motherboard, 64GB DDR5 RAM at 5600MHz, and a PCIe5 GPU, significantly boosting performance for Blender and rendering tasks.
• Switched remote work software from TeamViewer to Parsec to improve remote collaboration and responsiveness.
• Planned a work trip to Taiwan, anticipating some development slowdowns but preparing to maintain workflow stability.
• Initiated steps toward forming an LLC for game development, targeting mid-summer completion to formalize business operations.

Reflections

• Prioritizing hand-crafted assets over purely procedural methods for grass proved effective, improving both visual quality and production speed.
• Limiting grass rotation angles simplified visual consistency and reduced complexity without noticeable loss in variety.
• Blender’s procedural tools are powerful but have practical limits, especially for complex pathing and building modeling; adapting the workflow to leverage Blender for terrain while handling buildings manually is a pragmatic shift.
• The exponential growth in tile variants due to path classification was an unexpected complexity, highlighting the need for more scalable rule sets or alternative approaches.
• Shader-based edge detection and ray casting hold promise to solve visual intersection and compositing issues, but require further experimentation.
• Hardware and software upgrades have stabilized the workflow, but upcoming travel poses a test for maintaining productivity.
• Formalizing the business structure is a necessary step that adds focus and accountability but requires balancing with ongoing development demands.

Next Steps

• Continue refining the balance between procedural and hand-crafted assets, focusing on maintaining aesthetic quality without sacrificing efficiency.
• Prototype and evaluate shader-based edge detection for vegetation to replace Blender compositor effects.
• Develop strategies to manage or reduce tile variant complexity in path classification, possibly through rule simplification or alternative data structures.
• Experiment with manual or semi-manual building asset workflows to complement terrain modeling.
• Progress with LLC formation and establish best practices for business management alongside development.

Keynotes

Decisions & Observations

• Decided to limit Blender use to terrain and environmental elements; buildings will be handled manually or with rough 2D shapes to maintain artistic control.
• Initiated LLC formation targeting mid-summer completion to formalize business operations and improve management alongside development.

Issues & Next Steps

• Needs to resolve exponential growth in tile variants caused by height-based path point classification, threatening manageability.
• Will explore shader-based edge detection for vegetation to replace Blender compositor effects and improve real-time visuals.
• Investigating ray casting and shader solutions to prevent grass blades intersecting with rocks, as density reduction alone is insufficient.
• Will develop strategies to manage or reduce tile variant complexity in path classification through rule simplification or alternative data structures.
• Will experiment with manual or semi-manual workflows for building assets to complement procedural terrain modeling.

Uncategorized Notes

• Upgraded hardware and switched remote collaboration tools to stabilize workflow; upcoming travel to Taiwan may impact productivity and requires preparation.

Generated with Nucleate by Watchlight Studio — distilled from full transcription.