minWM: A Full-Stack Open-Source Framework for Real-Time Interactive Video World Models

2026-05-28

AI Summary

This work addresses the challenge of transforming existing high-quality video diffusion models into controllable, causal, and low-latency real-time interactive world models. We propose a modular full-stack framework that enables end-to-end conversion from general-purpose text-to-video or image-to-video foundation models into streaming autoregressive world models. Our approach integrates camera-control fine-tuning, a novel Causal Forcing++ training pipeline—featuring autoregressive diffusion, causal ODE/consistency distillation, and asymmetric DMD—and few-step generation techniques. The method is successfully deployed on backbone models such as Wan2.1-T2V-1.3B and HY1.5-TI2V-8B, enabling low-latency rolling prediction. We release code, checkpoints, and comprehensive system ablations, establishing the first complete pathway for converting general video generation models into real-time interactive world models.

Abstract

Recent video diffusion foundation models have achieved remarkable progress in high-quality video generation, yet turning them into real-time interactive video world models remains challenging. Interactive world models require controllable, causal, and low-latency rollout, which in practice demands a full pipeline spanning data construction, controllable fine-tuning, autoregressive training, few-step distillation, and streaming inference. In this work, we present minWM, a full-stack open-source framework for building real-time interactive video world models. minWM provides an end-to-end pipeline that converts existing bidirectional T2V/TI2V video foundation models into camera-controllable few-step autoregressive world models. Specifically, minWM first fine-tunes a bidirectional video diffusion model with camera control, and then applies the Causal Forcing / Causal Forcing++ pipeline, including AR diffusion training, causal ODE or causal consistency distillation, and asymmetric DMD, to distill it into a few-step autoregressive generator for low-latency rollout. The framework is modular and architecture-extensible: we instantiate it on representative open backbones, including Wan2.1-T2V-1.3B and HY1.5-TI2V-8B, covering both cross-attention-based condition injection and MMDiT-style architectures. minWM also supports adapting existing video world models, such as HY-WorldPlay, to new data distributions, training recipes, and latency targets. Beyond releasing runnable scripts, checkpoints, documentation, and inference code, we provide practical ablations on camera trajectory quality, controllability training steps, and minimal batch-size requirements. We hope minWM serves as a reproducible and extensible recipe for building and adapting real-time interactive video world models. Project Page: [https://github.com/shengshu-ai/minWM](https://github.com/shengshu-ai/minWM)

Problem

• interactive world models
• real-time video generation
• controllable rollout
• low-latency inference
• video diffusion models

Innovation

• real-time interactive video world models
• causal forcing
• few-step distillation
• camera-controllable generation
• autoregressive video diffusion