
How InstaLOD proved to be an indispensable tool for CVEDIA, streamlining their 3D asset optimization pipeline for AI simulation and training.
CVEDIA, a company specializing in the creation, adaptation, and optimization of 3D assets and environments for AI simulation and training, faced a significant challenge in managing the high demand for optimized 3D content. The process of model optimization, crucial for their workflow, was highly complex, time-consuming, and often manual, taking days to complete. By integrating InstaLOD into their pipeline, CVEDIA successfully automated this critical phase, drastically reducing optimization times from days to mere minutes while maintaining uncompromised visual quality.
The core of CVEDIA's operations relies on a continuous supply of highly optimized 3D assets. However, their traditional workflow was hampered by several inefficiencies:
- High Demand vs. Manual Processes: The need for a large volume of assets and environments clashed with the predominantly manual and delicate nature of 3D model optimization.
- Time-Consuming Optimization: Procedures such as mesh reduction, UV unwrapping, and material consolidation often required many hours, sometimes extending to days for a single asset.
- Partial Automation: While some 3D software offered partial automation, a comprehensive, single-solution approach was lacking, leading to fragmented and inefficient workflows.
- Maintaining Visual Quality: A critical requirement was to achieve drastic polygon reduction without any perceivable loss in visual fidelity, which manual processes struggled to guarantee consistently.
CVEDIA turned to InstaLOD to tackle these challenges head-on. InstaLOD's robust features provided the comprehensive automation they needed:
- Advanced Polygon Reduction: InstaLOD's remesher allowed CVEDIA to reduce models with over 3 million polygons down to 150,000 polygons without visual degradation.
- Automatic UV Unwrapping: A significant time-saver, InstaLOD automatically unwrapped models during the optimization process, eliminating a historically manual and tedious task.
- Material Consolidation: InstaLOD enabled the team to merge a variety of complex materials into a single, efficient set of PBR textures, simplifying asset management and improving performance.
- Versatile Import/Export: InstaLOD's compatibility with various file types, such as FBX, ensured seamless integration into CVEDIA's existing pipeline.
- Configurable Settings: The ability to adjust maximum triangle counts and configure texture output settings provided precise control over the optimization results tailored to their AI and machine learning requirements.

The implementation of InstaLOD brought transformative results for CVEDIA:
- Time Savings: The most significant impact was the reduction in asset optimization time from days to mere minutes.
- Uncompromised Visual Quality: Despite aggressive polygon reduction and material merging, the visual quality of the optimized assets remained completely intact.
- Streamlined Workflow: InstaLOD consolidated multiple manual steps—mesh reduction, UV unwrapping, and material merging—into a single, automated process.
- Increased Productivity: By automating complex tasks, CVEDIA's team could produce high-quality, optimized 3D content at an unprecedented pace, meeting the high demand for AI simulation and training data.
- Efficiency in AI/ML Applications: The ability to output low-polygon models with single PBR materials made the assets ideal for real-time AI simulation and machine learning environments, ensuring optimal performance.

InstaLOD proved to be an indispensable tool for CVEDIA, empowering them to overcome the bottlenecks of 3D asset optimization. By automating highly complex and time-consuming procedures, InstaLOD enabled CVEDIA to accelerate their content creation pipeline, deliver high-quality assets efficiently, and significantly enhance their capabilities in AI simulation and training.
"Using InstaLOD, we can reduce models with more than 3 million polygons and a variety of materials down to 150,000 polygons with a simple PBR material in a matter of minutes, and most importantly without compromising any visual quality." — Matthew from CVEDIA
https://youtu.be/j1rHG_haxgM?si=ILrHnAUnBfWGpzDd
First published: 6th August 2021