Fictures

May 18, 2026 / Evaluation

Evaluating Lightweight 3D Worlds for Static Web Publishing

This article examines evaluating lightweight 3d worlds for static web publishing as an engineering constraint in Fictures. The central claim is practical: public character worlds need assets that are repeatable, inspectable, and cheap to serve, not merely impressive in an isolated generation demo.

Abstract

This article examines evaluating lightweight 3d worlds for static web publishing as an engineering constraint in Fictures. The central claim is practical: public character worlds need assets that are repeatable, inspectable, and cheap to serve, not merely impressive in an isolated generation demo.

1. Background

3D scene generation seeks to synthesize spatially structured, semantically meaningful, and photorealistic environments for applications such as immersive media, robotics, autonomous driving, and embodied AI. Early methods based on procedural rules offered scalability but limited diversity. Recent advances in deep generative models (e.g., GANs, diffusion.

Optimize GLTF/GLB for Fast 3D Model Loading - LotifyAI provides implementation context.

2. Fictures Context

Fictures world pages are not full games: they are cached, ad-safe browser scenes that must load quickly, communicate story context, and expose a small number of reliable interactions.

The operational question is therefore not whether a model can produce a plausible demo artifact. The harder question is whether the output can enter a daily publishing loop where readers see stable character identity, fast pages, and enough technical provenance to make the archive auditable.

3. Method

The daily blog job searches arXiv and the open web, records the sources used for the article, and then writes a static page. This mirrors the product architecture: expensive or unstable work happens before publication, while the public site serves cached HTML, GLB, image, and metadata artifacts.

4. Evaluation Lens

What should be measured when a generated scene must be both narrative media and web infrastructure?

For Fictures, a useful answer combines measurable asset properties with editorial constraints: file size, mesh stability, material consistency, humanoid compatibility, browser behavior, source license risk, and whether the result supports a story beat rather than only a thumbnail.

5. Limitations

The sources below are used as supporting context, not as a claim that any single model or format fully solves production character generation. Generated meshes still need evaluation, simplification, rig checks, and public-page tests before they become durable media assets.

References

  1. 3D Scene Generation: A Survey (2025-05-08): 3D scene generation seeks to synthesize spatially structured, semantically meaningful, and photorealistic environments for applications such as immersive media, robotics, autonomous driving, and embodied AI. Early.
  2. SPATIALGEN: Layout-guided 3D Indoor Scene Generation (2025-09-18): Creating high-fidelity 3D models of indoor environments is essential for applications in design, virtual reality, and robotics. However, manual 3D modeling remains time-consuming and labor-intensive. While recent.
  3. Controllable 3D Outdoor Scene Generation via Scene Graphs (2025-03-10): Three-dimensional scene generation is crucial in computer vision, with applications spanning autonomous driving, gaming and the metaverse. Current methods either lack user control or rely on imprecise, non-intuitive.
  4. Optimize GLTF/GLB for Fast 3D Model Loading - LotifyAI
  5. Optimize 3D Assets with Khronos' New glTF-Compressor Tool
  6. Optimizing 3D Models for the Web using Draco and other tools
  7. Faster glTF Loading with Needle & model-viewer