StateFlow：構建、演化與存取用於預視覺化的 3D 世界狀態

摘要

預視化是電影、遊戲、建築與都市設計中，介於構思與製作之間的中間層。它讓創作者能以疊代方式精煉場景、動作、攝影機與時空動態。然而，現有的生成方法依賴簡單的提示詞，透過一次性的影像或影片合成來共同控制所有這些因素，導致可控性薄弱，且對疊代編輯的支援有限。從根本而言，一個世界由多個具備幾何、外觀及其他屬性的元素，以及攝影機所組成。不同的畫面是透過對這個共享狀態進行局部修改或重組而產生的，而這個狀態在大多情況下會被重複使用。因此，我們認為所缺少的是一個明確且持久的工作狀態。為了解決這個問題，我們提出 StateFlow，一個以狀態為中心的生成式預視化框架。StateFlow 並非一次性生成影片，而是使用可編輯的 3D 世界來組織場景結構、演化過程與攝影機，同時在需要更高保真度時，借助現成的影片模型來提升視覺品質。這個世界被維護為一個由場景元素與攝影機配置所構成的持久化結構化 3D 狀態，作為預視化的核心工作表徵。基於此洞見，StateFlow 包含三個階段，分別用於建構、演化與取用世界狀態。狀態建構階段透過先驗引導、衝突感知的雙視圖初始化，將生成的 2D 內容提升為連貫的 3D 世界。狀態演化階段則在使用者意圖轉化為結構化狀態轉換的同時，保留世界記憶，避免每次編輯都需重新生成整個場景。狀態取用階段利用渲染回饋的反思機制，將攝影機規劃精煉為視覺上可行的軌跡，避免僅依賴 VLM 的語意理解。實驗結果顯示，StateFlow 能為影片創作與類遊戲原型開發產生高品質的 3D 世界。

English

Previsualization is an intermediate layer between ideas and production in film, games, architecture, and urban design. It lets creators iteratively refine scenes, actions, cameras, and spatial-temporal dynamics. Yet existing generative methods rely on simple prompts to jointly control all of these factors through one-shot image or video synthesis, offering weak controllability and limited support for iterative editing. Fundamentally, a world comprises multiple elements with geometry, appearance, and other attributes, together with cameras. Different frames are produced through local modifications or recombinations of this shared state, which is otherwise largely reused. Therefore, we argue that the missing component is an explicit and persistent working state. To address this, we present StateFlow, a state-centric framework for generative previsualization. Rather than generating videos in one shot, StateFlow uses an editable 3D world to organize scene structure, evolution, and cameras, while off-the-shelf video models enhance visual quality when higher fidelity is desired. This world is maintained as a persistent structured 3D state of scene elements and camera configurations, serving as the core working representation for previsualization. Built on this insight, StateFlow has three stages to construct, evolve, and access the world state. State construction lifts generated 2D content into a coherent 3D world through prior-guided, conflict-aware dual-view initialization, while State evolution translates user intent into structured state transitions while preserving world memory, avoiding full-scene regeneration for each edit. State access uses render-feedback reflection to refine camera plans into visually feasible trajectories, avoiding reliance on VLM semantics alone. Experiments show that StateFlow produces high-quality 3D worlds for video creation and game-like prototyping.