Researcher working across generative video world models and multimodal game-playing agent benchmarks
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-08) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Publication-volume data is not available in this snapshot: total paper counts, venue record, and yearly production could not be computed from the two-paper Evolve catalog, so this axis is left unscored. It would measure aggregate research productivity over time.
Impact—
Citation and influence metrics are absent: no citation counts, h-index, or downstream adoption signals exist in the provided data, so impact is not scored. It would measure how widely the work is cited and built upon by other researchers.
Focus0/100
The focus score is 0, reflecting an even 50/50 split between two distinct fields — generative video world models (computer vision) and multimodal game agents and benchmarks (machine learning). Both 2026 catalog papers fall on opposite sides of this split, indicating breadth across two lines rather than concentration in a single area.
Collaboration62/100
Collaboration scores 62 from a network of 22 distinct coauthors, with 12 recurring collaborators including Yiran Wang, Ying Shan, and Xiaodong Cun. The first-author WorldCrafter paper and the middle-position GameHorizon Suite both involved sizeable teams, suggesting the researcher operates comfortably within multi-author group projects rather than solo work.
Activity—
Recent-activity data is insufficient: the snapshot provides only two 2026 arXiv preprints and no multi-year publication cadence, so this axis is not scored. It would measure publishing rhythm and sustained research output across recent years.
Reproducibility38/100
Reproducibility averages 38 across the two reviewed papers (reproMean 38, 2 papers). WorldCrafter rates low: its claimed 47.6% revisit-consistency gain over 8 baselines rests on a self-built benchmark despite strong ablations. GameHorizon Suite rates medium, with 5k hours of gameplay across 21 games and 47 models, though its VLM-generated ground truth needs scrutiny.
Research summary
Wangbo Yu is a researcher whose cataloged work sits at the intersection of two fields: generative video world models in computer vision and multimodal game-playing agents and benchmarks in machine learning, with the Evolve field distribution split exactly 50/50 between the two. This is an early-stage profile built from a small public bibliographic snapshot: two 2026 arXiv preprints in the Evolve catalog, no institution, country, or ORCID on record, and identity resolved by name from public bibliographic data. The first-authored (non-corresponding) WorldCrafter proposes a consistent video world model with implicit 3D-aware memory, targeting long-horizon consistency and camera control in autoregressive video diffusion with few-step distillation; Evolve's review credits a claimed 47.6% revisit-consistency gain over 8 baselines with strong ablations, while noting the evidence rests on a self-built benchmark. The middle-position GameHorizon Suite contributes a large multi-horizon AAA gameplay dataset and evaluation suite — 5k hours of gameplay, 21 games, and 47 evaluated models — though VLM-generated ground truth and the lack of online AAA evaluation warrant scrutiny. Recorded collaboration spans 22 distinct coauthors with 12 recurring partners, yielding a collaboration dimension score of 62, while reproducibility averages 38 across the two reviewed papers. Output, impact, and activity axes remain unscored for lack of citation and production data.
Collaboration pattern
Wangbo Yu's recorded network spans 22 distinct coauthors, with 12 recurring partners — Yiran Wang, Ying Shan, Xingyilang Yin, Junfu Pu, Guangzhi Wang, Kaifeng Li, Mingyu Ouyang, Huiqiang Sun, Lingen Li, Cheng Cheng, Honghao Chen, and Xiaodong Cun. Both catalog papers are multi-author efforts: a first-authorship on WorldCrafter and a middle position on GameHorizon Suite, with neither role corresponding-author, consistent with team-based work bridging video generation and game-agent groups (collaboration dimension: 62).
Research keywords
Highlights
First-authored WorldCrafter: 3D-aware memory for consistent video world models
GameHorizon Suite: large-scale AAA gameplay dataset and benchmark
Timeline
- 2026First-author arXiv preprint WorldCrafter proposes a consistent video world model with implicit 3D-aware memory, claiming a 47.6% revisit-consistency gain over 8 baselines.
- 2026Co-authors GameHorizon Suite in a middle author position — a multi-horizon AAA gameplay dataset and evaluation benchmark covering 5k hours, 21 games, and 47 models.
- 2026Cataloged output splits evenly across two fields — generative video world models and multimodal game-agent benchmarks — accumulated with a network of 22 distinct coauthors.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Dimension scores are computed from public bibliometric data only; no citation counts, h-index, or venue-tier metrics were provided.
- The profile rests on two 2026 arXiv preprints in the Evolve catalog; identity is name-resolved from public bibliographic data via co-author and topical consistency, with institution, country, and ORCID all null.
- Output, impact, and activity axes are unscored because the snapshot lacks publication counts, citation data, and multi-year cadence.
- Both papers are arXiv preprints with no peer-review venue recorded, and years lack month granularity, so within-2026 ordering is undetermined.