Example of rigged character workflow using Blender, Unity, and Gimp.

This example is for a game with a stylized custom lighting workflow, and skips several PBR texturing steps in favor of non-standard approaches better suited for the specific shader and art-style.

Rigged Character WOrkflow

Base Mesh to Animated Asset

    • This workflow accommodates different pipelines, and workflows including photoreal PBR.

    • The pipeline being demonstrated uses the following softwares/processes:

      • Base mesh modification - Blender

      • Detailed mesh sculpting - Blender

      • UV Unwrapping - Blender

      • Texturing (Discrete) - Blender + Gimp

      • Texturing (Procedural) - Unity

      • Clothing Design - Blender

      • Clothing Simulation - Blender

      • Rigging - Blender

      • Shader Dev - Unity

      • Animation - Unity / C#

      • Animation Control - C#

Step 0: Base Mesh Strategic Assessment

Overview:

A licensed base mesh was selected and modified due to its alignment with project requirements, offering a cost-effective starting point over a bespoke sculpted solution. While creating a custom base mesh would allow more control, the marginal benefits did not justify the time investment.

Key Features:

The base mesh provided:

  • Adequate topological detail (3000 vert / 6000 tri; ideal: 2500 vert / 5000 tri)

  • Edge loop isolation for elbows and knees

  • Usable facial mask

  • Separate meshes for eyebrows, eyes, mouth, teeth, and tongue

  • Silhouette detail for features such as the philtrum, nostrils, and anatomical landmarks

Fixable Issues:

  • Highly stylized physiology

  • High surface tension geometry (nose, brow, jawline)

  • Closed-hull geometry for stylized eyelashes

  • Discontiguous edge loops around the torso

Impact:

The base mesh reduced development time while allowing targeted edits to address its minor shortcomings.

Step 1: Base Mesh Editing – Silhouette and Physiology

Overview:

The base mesh was modified using sculpting and proportional editing tools to align facial proportions with reference materials.

Specific Course of Action:

  • Adjusted eye shape, facial features, and skull shape for consistency with reference designs.

  • Used isometric camera views to simplify alignment and maintain consistent scaling across perspectives.

Reason:

Isometric views avoided complications with field-of-view distortions and camera positioning, enabling efficient adjustments and reliable comparisons.

Step 3: Clothing Design and Simulation

Overview:

To enhance stylized “toon” lighting, facial normals were edited to minimize lighting artifacts common in stylized shading approaches.

Specific Course of Action:

  • Constructed a simplified flat-shaded reference hull to transfer normal data.

  • Iteratively smoothed and relaxed stylized normals to reduce lighting artifacts without reintroducing literal normal issues.

Impact:

This workflow supported artifact-free shading at extreme angles and enabled fast iteration via Blender-to-Unity integration.

Step 3: Clothing Design and Simulation

Overview:

Clothing was designed and simulated using Blender for its faster creation workflow during early design stages, though Marvelous Designer (MD) was used for other projects requiring extensive garment modifications.

Key Features:

  • Undershirt: Modeled from body geometry using Boolean and Shrinkwrap modifiers, then manually adjusted for fit and UV unwrapping.

  • Jacket: Created with retopology tools and modifiers for fit, with manually sculpted drape and additional detail edits.

  • Pants: Modeled similarly to the undershirt, with simulated fit and manually unwrapped UVs.

  • Shoes: Designed with hard surface modeling techniques to reinforce panel designs, maintain sharp geometry, and optimize shading.

Impact:

Using Blender streamlined the creation of garments suited for unfinalized character designs, reducing time-to-iteration.

Step 4: UV Unwrapping

Overview:

UVs were unwrapped using a projection-based approach to optimize texture artist workflows and facilitate masking for clothing occlusion.

Specific Course of Action:

  • Forward projection allocated more UV space to the face for detailed painting while simplifying masking workflows for 2D editing tools like GIMP.

  • Scalp and neck UVs were adjusted to accommodate a variety of hairstyles.

  • The body’s UVs were forward-projected for easy masking and opacity-based clipping under clothing.

Reason:

Projection-based unwrapping minimized distortion and allowed easier body masking compared to standard UV techniques, which distribute texels more proportionally but complicate isolating areas.

Step 5: Hair Modeling

Overview:

Hair meshes were modeled to support stylized lighting and secondary animation, divided into static and rigged components.

Key Features:

  • Static Mesh: Includes bangs and skull-adjacent hair, modeled to preserve normal continuity and optimize lighting.

  • Rigged Mesh: Ponytail segments rigged with named bone chains for secondary physics-driven animation.

Step 6: Rigging

Overview:

Blender’s Metarig was used for rigging, providing standardization and reducing downstream issues in Unity integration.

Key Features:

  • Body: Automatically weighted and refined through manual weight painting, focusing on joint deformation control.

  • Face: Adjusted to eliminate skull volume distortion and deformation during typical movements.

Clothing:

Rigged using Blender’s Data Transfer modifier, with manual corrections for clipping during extreme deformations.

Step 7: Export/Import and Configuration

Overview:

Meshes were exported as FBXs with configurations to align Blender's conventions with Unity’s requirements, and a custom utility automated rig configuration for Unity's FinalIK system.

Specific Course of Action:

  • Export Settings: Ensured unit scale consistency, corrected axial mismatches, and excluded context objects from export.

  • Custom Utility: U_MBR2FBBIK automated FBBIK configuration by:

    • Mapping FK armature data to Unity components.

    • Generating IK targets and setting known good values.

    • Reducing manual, error-prone configuration steps.

Impact:

This utility streamlined rig setup, minimized configuration errors, and supported faster iteration.

Step 8: Animation

Overview:

Animation was fully procedural, leveraging parametric measurements and a state machine-backed controller.

Key Features:

  • Contextual behaviors like reaching and foot planting.

  • Adaptation across character sizes without manual adjustments.

  • Local simulation for network multiplayer compatibility.

Step 9: Animation Control

Overview:

Procedural animation replaced traditional baked animations, improving scalability and dynamic responsiveness.

Upsides:

  • Simplified character implementation for artists.

  • Clear state-event integration for game logic.

  • Profiles for character-specific animation expressions (e.g., swagger, sway).

Downsides:

  • Limited artist input for nuanced animations.

  • Transition control challenges due to FSM architecture.

Impact:

This approach facilitated responsive gameplay and procedural variety while maintaining scalability for future expansions.