Implementing X3DCPPSAIL with AI-Assistance

Abstract

Extensible 3D Graphics (X3D) is a specification defined by ISO/IEC 19775-1, a standard for representing web-publishable, three-dimensional (3D) models and interactive spaces (ISO/IEC 2023). ISO/IEC 19775-2 defines the Scene Authoring Interface (SAI), which specifies how applications can programmatically inspect and manipulate the X3D scene graph (ISO/IEC 2015). ISO/IEC 19777 is a sibling standard to 19775 that offers language binding definitions that map a variety of programming languages to the SAI standard. Prior to the work discussed here, the most recent C++ language binding, ISO/IEC 19777-4 (part 4), only implemented support for the X3D v3.3 node set (ISO/IEC 2025). The draft 4.1 version of the X3D standard is in an advanced stage of development, and will add material, rendering, and geometrical features that align much better to the emerging OpenUSD framework for authoring 3D scenes (ISO/IEC 2026). This paper describes X3DCPPSAIL, an updated ISO/IEC 19777-4 scene authoring interface library supporting the X3D 4.1 node set. The library is intended to be the foundation for the creation of a file translator between the X3D and OpenUSD formats. This paper documents the X3DCPPSAIL architecture and the contribution of the Artificial Intelligence (AI) assistant (Anthropic 2026) used throughout the development process to interpret the X3D specification, generate repetitive code structures, explain unfamiliar concepts, and validate design decisions. The paper describes the resulting development workflow and the strengths/weaknesses of the AI assistant that emerged throughout the X3DCPPSAIL development process. The final API implementation takes advantage of these rapid prototyping results and design-pattern experimentation to deterministically produce the final implementation by using the X3D Unified Object Model (X3DUOM).