KT - Assets Unit Tests

At Kylotonn, we faced a recurring issue: the lack of visibility over asset quality and memory usage across projects
One of our major problems was the presence of highly uniform textures consuming unnecessary memory due to the way GPU compression formats work

To address this, I developed a lightweight analysis tool called Texture Optim Ratio

Texture Optim Ratio tool

This tool evaluates texture efficiency using PNG compression (file sizes) as an indicator of average pixel information density, while engine-compressed formats (BC1, BC4, BC7) are resolution-dependent rather than content-dependent

Because PNG compression benefits from color uniformity while GPU formats do not, this comparison reveals how uniform a texture is

High ratio = good entropy (ex: noisy textures)
Low ratio = likely uniform / wasteful textures that could be lowered in resolution

This tool provided a simple metric to:

• Detect poorly optimized textures
• Identify incorrect compression settings

And flag issues such as:

• Missing alpha in BC7 textures
• sRGB misusage (normal maps, masks)
• Other pipeline inconsistencies

---

Houdini PDG / TOPs

Some time after the previous tool, we expanded the idea into a scalable asset validation pipeline using Houdini PDG / TOPs
The goal was to provide centralized, automated asset analysis across multiple projects, accessible through a simple web interface for artists and developers
I was responsible for the design, implementation, and production pipeline integration of the tool

Using Houdini PDG allowed us to:

1. Process large volumes of assets in parallel thanks to PDG Services
2. Handle both light (textures) and heavy data (meshes, levels) thanks to Houdini versatility
3. Rapidly iterate and be fully extensible, allowing new validation rules to be added easily by adding custom attributes

With that system, it exposes key metrics such as:

• Mesh data: polycount, LOD count, bounds
• UVs: count, range, texel density
• Materials: count, usage density
• Collisions: count, complexity
• Texture metrics (including Optim Ratio)

It also generates automated warnings, for example:

• Improper LOD reduction
• Incorrect naming conventions
• Invalid texel density
• Poor material distribution

This tool helped identify memory waste and improve overall asset quality, contributing to better runtime performance