XR SURVEY

Visual Generative AI for Extended Reality

A survey of assets, scenes, dynamic worlds, and XR systems.

Explore

Loading demonstrations…

Recorded research demos from their authors. Sources & credits ↗

Paper collection

Download catalogue ↓

204 methods and systems · 37 datasets and benchmarks · Updated October 2026

Loading catalogue…

A curated narrative review with targeted retrieval through 23 October 2026.

About the survey

How do generated objects, scenes, and dynamic worlds become usable XR experiences? This survey connects four content families with the representations and evidence needed for viewing, editing, and interaction.

Read the full abstract and download the paper →

Browse the method taxonomy

Loading the taxonomy…

Review framework & overview

Compare what is generated, how it is generated, the representation exposed to applications, and when generation enters the interaction loop.

From generated content to XR operations. Source examples: SceneScape, 4D-fy, and WorldPlay. Figure credits ↗
Open questions for XR evaluation

Proposed evaluation directions, rather than new experimental results.

    Resources & references

    Source-linked capability evidence for 31 selected methods and systems.

    Selected datasets & benchmarks

    Cite this survey

    Huang, G. (2026). Visual Generative AI for Extended Reality: A Survey. Preprints. https://doi.org/10.20944/preprints202610.2630.v1

    BibTeX
    @misc{huang2026visualgenerativexr,
      title  = {Visual Generative {AI} for Extended Reality: A Survey},
      author = {Huang, Guoxi},
      year   = {2026},
      doi    = {10.20944/preprints202610.2630.v1},
      url    = {https://www.preprints.org/manuscript/202610.2630}
    }

    Visitors: —

    Expanded research figure