---
title: "Tencent's WorldCrafter Remembers Generated Worlds 47.6% Better on Revisit"
subtitle: "A new video world model uses a camera-queryable 3D-aware memory to keep generated scenes consistent across long, revisiting camera paths."
url: "https://alphasignal.ai/news/tencent-s-worldcrafter-remembers-generated-worlds-47-6-better-on-revisit"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-09-21T17:57:06.000Z"
updated: "2026-09-26T01:58:02.012Z"
type: "Paper"
topics:
  - "Video"
  - "Llms"
subtopics:
  - "World Models"
  - "Video Generation"
  - "Vision Language"
upvotes: 10
source_url: "https://arxiv.org/abs/2609.24984"
comments: 0
truncated: true
---

# Tencent's WorldCrafter Remembers Generated Worlds 47.6% Better on Revisit

> A new video world model uses a camera-queryable 3D-aware memory to keep generated scenes consistent across long, revisiting camera paths.

Interactive video world models generate a scene as a user moves a virtual camera through it. Their limited context windows create a persistent memory problem: after the earliest frames leave the window, revisiting a location can produce different walls, objects, or textures. [WorldCrafter](https://drexubery.github.io/WorldCrafter), developed by researchers at Peking University and Tencent’s ARC Lab, addresses that drift with an implicit 3D-aware memory queried by the requested camera pose.

The [paper](https://arxiv.org/abs/2609.24984) reports a 47.6% relative improvement in revisit consistency over its strongest baseline, better camera-control accuracy, and generation at 16 frames per second on four GPUs. The system joins work such as Google DeepMind’s Genie 3, Tencent’s HY-WorldPlay, NVIDIA’s Lyra 2.0, and Matrix-Game 3.5, all of which aim to turn video generation into controllable simulation.

## Why generated worlds drift

Video world models predict future observations from user actions, usually by conditioning a diffusion transformer on recent frames and camera movement. A transformer can attend to a finite number of tokens, so long explorations eventually push early observations out of context. Existing memory designs manage that limit in three main ways:

* **Context memory:** Retain earlier frames or cached attention states. Full-history attention raises compute and memory costs, while selective retrieval can omit useful views.
* **Spatial memory:** Estimate scene depth, construct an explicit 3D representation, and warp it into the target view. Tight geometric alignment can restrict moving subjects and other scene changes.
* **Implicit memory:** Compress history into learned tokens. Geometry-aware systems such as CineScene and GIM-World improve spatial recall, although geometry-estimation pretraining can preserve less appearance detail than reconstruction training.

WorldCrafter follows the implicit-memory approach and initializes its encoder from LagerNVS, a multi-view representation trained for novel-view reconstruction. That task requires the encoder to preserve scene geometry, color, texture, and object appearance across viewpoints.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [WorldCrafter: Consistent Video World Model with Implicit 3D-aware Memory](https://arxiv.org/abs/2609.24984)
- [WorldCrafter paper HTML](https://arxiv.org/html/2609.24984)
- [WorldCrafter Project Page](https://drexubery.github.io/WorldCrafter)
