Core Insights - Sora2, an AI video generation product launched by OpenAI, is set to tap into a trillion-dollar market, significantly impacting the industry chain [1][2][6] - The technology has evolved through various stages, now dominated by the Diffusion Transformer (DiT) architecture, enhancing video generation quality and controllability [1][2][17] - Sora2 achieved rapid success, topping the U.S. iOS app charts shortly after launch, indicating strong market demand and user engagement [1][6][30] Technology Development - Video generation technology has progressed from early GAN and VAE architectures to the current DiT architecture, which combines the strengths of Transformer and diffusion models [1][17][29] - Sora2 has not made significant technical breakthroughs but has optimized training with large-scale video data and improved controllability through prompt rewriting and audio-visual synchronization [1][32][36] Market Potential - The AI video generation market is projected to be substantial across three segments: - Professional creators (P-end) with a mid-term market of 26.2 billion yuan and a long-term potential of 88.8 billion yuan - Business applications (B-end) focusing on film and advertising, with mid-term and long-term markets of 50.1 billion yuan and 66.6 billion yuan, respectively - Consumer applications (C-end) expected to reach 76.3 billion yuan in the mid-term and 155.4 billion yuan in the long term [2][7][8] Product and User Engagement - Sora2 employs a social product loop strategy, simplifying the creation process to just a text input box, allowing users to generate videos with a single sentence [1][6][39] - The app's features, such as "Remix" and "Cameo," enhance social sharing and user interaction, contributing to its viral growth [1][6][55][56] - The app's initial success is attributed to its invitation-only model, which creates exclusivity and encourages user sharing among friends [1][45][46] Cost and Collaboration - Sora2 incurs high computational costs, estimated at $14 million per day, leading to an annual cost exceeding $5.12 billion, highlighting the importance of computational power in AI applications [2][8][36] - OpenAI has partnered with NVIDIA and AMD to secure computational resources necessary for Sora2's operations [2][8]

作者 | 微卷的大白 编辑 | 自动驾驶之心 原文链接： https://zhuanlan.zhihu.com/p/1952449084788029155 点击下方 卡片 ，关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近30个 方向 学习 路线 >>自动驾驶前沿信息获取 → 自动驾驶之心知识星球 前两天看到李飞飞 Worldlabs 新工作Mrable的时候，提到后面想多看一看 3DGS / 重建相关的工作。 不过如果真的有小白要踩坑 ，gsplat 的文档和维护其实比gaussian-splatting 要稍微好一些，个人更推荐这个库。 相比3DGS 论文对应的 gaussian-splatting 库，nerfstudio-projectgsplat 是对官方库做了一些优化，可参考https://docs.gsplat.studio/main/migration/migration_inria.html 的 说明。 但是知乎搜了一下发现，讲 3DGS 论文原理、改进的不少，我自己上半年也回顾过cuda kernel 源码：重温经典之 3DGS CUDA 源码解析 ，但是另一个常用的gsplat ...

Core Viewpoint - The article discusses the development of EasyCache, a new framework for accelerating video diffusion models without requiring training or structural changes to the model, significantly improving inference efficiency while maintaining video quality [7][27]. Group 1: Research Background and Motivation - The application of diffusion models and diffusion Transformers in video generation has led to significant improvements in the quality and coherence of AI-generated videos, transforming digital content creation and multimedia entertainment [3]. - However, issues such as slow inference and high computational costs have emerged, with examples like HunyuanVideo taking 2 hours to generate a 5-second video at 720P resolution, limiting the technology's application in real-time and large-scale scenarios [4][5]. Group 2: Methodology and Innovations - EasyCache operates by dynamically detecting the "stable period" of model outputs during inference, allowing for the reuse of historical computation results to reduce redundant inference steps [7][16]. - The framework measures the "transformation rate" during the diffusion process, which indicates the sensitivity of current outputs to inputs, revealing that outputs can be approximated using previous results in later stages of the process [8][12][15]. - EasyCache is designed to be plug-and-play, functioning entirely during the inference phase without the need for model retraining or structural modifications [16]. Group 3: Experimental Results and Visual Analysis - Systematic experiments on mainstream video generation models like OpenSora, Wan2.1, and HunyuanVideo demonstrated that EasyCache achieves a speedup of 2.2 times on HunyuanVideo, with a 36% increase in PSNR and a 14% increase in SSIM, while maintaining video quality [20][26]. - In image generation tasks, EasyCache also provided a 4.6 times speedup, improving FID scores, indicating its effectiveness across different applications [21][22]. - Visual comparisons showed that EasyCache retains high visual fidelity, with generated videos closely matching the original model outputs, unlike other methods that exhibited varying degrees of quality loss [24][25]. Group 4: Conclusion and Future Outlook - EasyCache presents a minimalistic and efficient paradigm for accelerating inference in video diffusion models, laying a solid foundation for practical applications of diffusion models [27]. - The expectation is to further approach the goal of "real-time video generation" as models and acceleration technologies continue to evolve [27].

Investment Rating - The report maintains a "Buy" rating for the AI video generation industry, highlighting the accelerated commercialization and strong performance of domestic manufacturers [2]. Core Insights - The AI video generation industry is entering a commercial development fast track, with significant advancements in technology and diverse application scenarios. The global market size is projected to reach approximately 25.63 billion USD by 2032, with a compound annual growth rate (CAGR) of 20% from 2025 to 2032 [4][40]. - The industry is driven by both pricing and model capabilities, with current API prices ranging from 0.2 to 1 RMB per second. The cost advantages of AI video generation compared to traditional video production methods are substantial [46][47]. - Domestic manufacturers, such as Kuaishou and Meitu, are showing outstanding performance in the competitive landscape, with products like Kuaishou's Kling and ByteDance's Seedance leading the market [58][62]. Summary by Sections 1. Technology Path - The evolution of AI video generation technology has progressed from static image sequences to GAN, Transformer, Diffusion Model, and DiT, enhancing content richness and controllability [4][7]. - The DiT architecture, which combines diffusion models with transformers, has emerged as a key direction in the industry, validated by the Sora model's performance [23][31]. 2. AI Video Generation Industry 2.1 Driving Factors - The growth of the AI video generation industry is fueled by both pricing and performance improvements, with significant cost advantages over traditional video production methods [46][47]. - The current mainstream generation duration is 5-10 seconds, with advancements allowing for longer video generation, enhancing narrative capabilities [47]. 2.2 Industry Applications - The industry has diverse applications in B2B sectors such as film content creation, commercial advertising, e-commerce marketing, and education, as well as in C2C scenarios that enhance user engagement [51][54]. 2.3 Product and Competitive Landscape - Domestic manufacturers like Kuaishou and ByteDance are leading the market with their advanced models, achieving high usage and web traffic [58][62]. - The competitive landscape shows that products like Seedance1.0 and Veo2/3 are among the top performers, indicating a strong domestic capability in AI video generation [58][62]. 3. Investment Recommendations and Related Stocks - The report suggests focusing on Kuaishou (1024.HK) and Meitu (1357.HK) as key investment opportunities in the AI video generation sector, given their strong commercial performance and growth potential [64][75].