报告来源 报告导读： 字节跳动近期推出新一代 AI 视频生成模型 Seedance 2.0 。该模型生成视频 稳定性上升，或可大幅降低制作成本，视频工业迎来 GPT3.5 时刻，行业有望加速发展。 投资建议： 近期，字节旗下的即梦团队发布 Seedance2.0 全新视频生成大模型，真正实现了从 " 能生成 " 到 " 能商用 " 的跨越式突破，该模型首次实现 了文字理解与字幕动效生成，可自动解析参考图中的文字并添加合理动态效果。此进展标志着 AI 正从 " 单模态理解 " 向 " 全双工连续感知 " 与 " 跨模态深 度创作 " 跃迁。 Seedance2.0 的自主创作能力不仅重塑了内容生产与交互模式，更催生了涵盖视频生成、实时交互、设计工具、端侧智能等环节的全新产业 链投资机会。我们看好 AI 应用加速落地的长期发展空间。 Seedance2.0 发布，控制精度提升至"导演级"。 其四项巨大突破，使得 Seedance2.0 的模型能力产生质变： 1 ）自分镜和自运镜，模型可以根据用户描 述的情节自动规划分镜运镜。 2 ）全方位多模态参考，用户可以同时给它最多 9 张图片、 3 段视频和 3 段音频，总 ...

Core Insights - The emergence of Seedance 2.0 by ByteDance marks a significant transformation in AI video generation, enabling creators to produce high-quality content with minimal effort, leading to a surge in the stock prices of related companies in the media sector [1][8] - Industry experts express both excitement and concern regarding the implications of this technology, particularly regarding the potential for misuse and the erosion of intellectual property rights [2][10] Industry Impact - The release of Seedance 2.0 has led to a notable increase in stock prices for companies in the A-share media sector, with several stocks hitting the daily limit up [1][8] - The technology is seen as a major leap forward in AI video creation, with capabilities that allow for more sophisticated storytelling and visual presentation [3][7] - The competitive landscape is shifting, as Seedance 2.0 and its counterpart, Kuaishou's AI 3.0, are evolving to become more adept at understanding cinematic techniques and narrative structures [3][4] Technical Advancements - Seedance 2.0 demonstrates improved stability, visual quality, and detail handling compared to previous models, showcasing its ability to generate content that meets commercial standards [4][8] - The model's advancements include better understanding of camera movements, narrative pacing, and multi-modal interactions, indicating a shift from mere image generation to a deeper comprehension of video creation [6][9] Market Reception - The initial market response to Seedance 2.0 has been overwhelmingly positive, with industry professionals noting its potential to democratize video creation by lowering the technical barriers for creators [8][9] - However, there are concerns about the implications of easily generated realistic videos, which could lead to a crisis of trust in digital content [10][11] Future Considerations - Experts suggest that while Seedance 2.0 excels in short-form content, it may not yet be suitable for long-form cinematic productions due to the complexity involved in narrative and pacing [13][16] - The industry is urged to address ethical concerns surrounding content creation, including copyright issues and the potential for misuse of generated content [12][16] - For broader application, advancements are needed in areas such as long video consistency, physical interaction accuracy, and the development of a robust ethical framework for AI-generated content [16]

全天候科技从一名接近字节跳动生态的应用商人士处独家获悉，Seedance 2.0预计最快有望于2月24日对 外开放API接口。 尽管这一具体时间表尚未获得字节跳动官方的确认，但多方信号显示Seedance 2.0的对外输出已进入倒 计时阶段。 此前，Seedance 2.0的核心能力主要局限于字节跳动内部生态，作为底层技术支撑着创意平台"即梦 AI"及营销工具"小云雀"的视频生成功能。 这意味着，这一此前仅服务于字节内部产品的核心能力，将很快转化为通用的技术服务，输出给第三方 的开发者与企业用户。 不仅如此，据全天候科技从行业人士处了解到，已有部分下游公司获悉相关动向并制定了接入计划，拟 抢占AI视频生成的红利。 然而，大规模开放API并非简单的接口对接，这对于字节跳动的底层算力储备将是一场极限压力测试。 继Seedance 2.0火出圈后，字节跳动正在加快该模型的商业化步伐。 视频帧的连续性计算、高分辨率渲染以及复杂的物理规律模拟等都对算力基础设施提出了极高的要求。 据近期全天候科技的实测数据显示，Seedance 2.0虽然表现惊艳，但在实际的用户体验层面仍面临挑 战：目前生成速度相对较慢，且存在一定的生 ...

国泰海通发布研报称，近期，字节旗下的即梦团队发布Seedance2.0全新视频生成大模型，真正实现了 从"能生成"到"能商用"的跨越式突破，该模型首次实现了文字理解与字幕动效生成，可自动解析参考图 中的文字并添加合理动态效果。此进展标志着AI正从"单模态理解"向"全双工连续感知"与"跨模态深度 创作"跃迁。Seedance2.0的自主创作能力不仅重塑了内容生产与交互模式，更催生了涵盖视频生成、实 时交互、设计工具、端侧智能等环节的全新产业链投资机会。该行看好AI应用加速落地的长期发展空 间。 国泰海通主要观点如下： 根据极客公园测算，Seedance2.0生成15秒视频的可用率或达到90%，相较此前行业内平均值大概20%， 提升幅度较大。当生成的视频可用率提升后，可降低实际成本，以做90分钟的视频项目为例，成本有望 从1万多降低到2000元左右，将提升行业的使用量，其巨大的成本压缩，或可改变整个行业的底层逻 辑。 Seedance2.0标志着AI视频生成从"技术可行"向"商业可用"的关键跨越 过去一年，视频生成领域已完成从512像素静态图像到10秒电影级短片的代际跃升。此次升级的多镜头 叙事与角色一致性保障能 ...

2026年2月初，字节跳动旗下视频生成模型Seedance 2.0以"多模态参考""原生音画同步"等突破性能力引 爆市场，其生成的电影级视频画面流畅度与细节真实感引发行业惊叹。 然而，技术光环之下，游戏科学CEO冯骥、科技博主"影视飓风"Tim等业内人士公开表达担忧：当逼真 视频的制作门槛趋近于零，假视频泛滥、肖像权侵犯、信任体系崩塌等风险正以前所未有的速度逼近。 Seedance 2.0的突破性在于其将AI视频生成从"概率性抽卡"变为"确定性生产"。据测试者反馈，模型可 根据单张静态照片自动生成人物动态形象与声音，甚至能推断建筑物背面结构。这种能力使得传统视频 鉴伪手段（如摄像机运动轨迹分析）几近失效。冯骥指出，技术飞跃将导致"一般性视频制作成本趋近 于算力边际成本"，但同时也意味着"逼真假视频将毫无门槛"。 更令人担忧的是模型对个人生物信息的深度复现能力。科技博主"影视飓风"Tim在测试中发现，仅凭一 张照片，模型即可生成与其本人高度相似的语音语调，且未经任何授权使用其公开视频作为训练数据。 这种"无提示词克隆"能力，使得公众人物与普通用户均面临肖像权、声音权被滥用的风险。 面对舆论压力，字节跳动在模型上 ...

请务必阅读正文之后的免责条款部分 股 票 研 究 Seedance2.0 发布，AI 视频迎来创作平权 与产业奇点 | [姓名table_Authors] | 电话 | 邮箱 | 登记编号 | | --- | --- | --- | --- | | 杨林(分析师) | 021-23183969 | yanglin2@gtht.com | S0880525040027 | | 吕浦源(分析师) | 021-23183822 | lvpuyuan@gtht.com | S0880525050002 | | 钟明翰(研究助理) | 021-38031383 | zhongminghan@gtht.com | S0880124070047 | 本报告导读： 字节跳动近期推出新一代 AI 视频生成模型 Seedance 2.0。该模型生成视频稳定性上 升，或可大幅降低制作成本，视频工业迎来 GPT3.5 时刻，行业有望加速发展。 投资要点： 计算机《从能力领先到入口级产品：阿里押注模 型、生态与 AI 基础设施》2026.02.10 计算机《计算机周观点第 33 期：底层基础设施迭 代加速，AI 原生力量重塑软件产业格 ...

Group 1 - The anime and gaming industry is experiencing sustained growth driven by fundamental industry improvements and breakthroughs in AI video generation models, leading to increased market attention on the gaming sector and a significant rise in trading volume for the Huatai-PineBridge Gaming ETF (516770) [1][3] - On February 10, 2026, the trading volume of the Huatai-PineBridge Gaming ETF reached 65.63 million yuan, marking a 140% increase compared to the previous trading day [1][3] - The Seedance 2.0 video generation model, released on February 7, 2026, has gained attention for its ability to generate movie-quality videos from text and images, offering advanced features that could enhance content production efficiency in gaming and anime [1][3] Group 2 - The policy environment remains favorable, and the solid recovery of the industry fundamentals has highlighted the investment value of the gaming sector and related products like the Huatai-PineBridge Gaming ETF [2][4] - In January 2026, a total of 182 gaming licenses were issued, continuing the normalized issuance trend established in 2025; the actual sales revenue of China's gaming market reached 350.79 billion yuan in 2025, a year-on-year increase of 7.68%, surpassing the 350 billion yuan mark for the first time [2][4] - The Huatai-PineBridge Gaming ETF closely tracks the CSI Animation and Gaming Index, which includes listed companies involved in animation, comics, and gaming, providing a comprehensive investment opportunity in the industry [2][4]

Group 1 - ByteDance has launched its latest AI video generation model, Seedance 2.0, which supports multi-modal input and intelligent storyboarding, capable of producing film-quality short videos with native audio and coherent scenes [1] - The model has been described as an "all-around AI director" and signifies a shift in AI video production from a playful tool to an industrialized production phase [1] - The CEO of Game Science, Feng Ji, noted that the ability of AI to understand and integrate multi-modal information has made a significant leap, suggesting that the production costs for general videos will no longer follow traditional film industry logic [1] Group 2 - The film market has shown increasing effects of scheduling and the Matthew effect, with audiences still willing to watch quality films, especially with the upcoming long Spring Festival holiday in 2026 [2] - The recent internal communication from the CEO of Reading Group highlighted the company's strategic focus on "evergreen content, IP + AI, and globalization," aiming for breakthroughs in various areas including AI [3] - The partnership between Disney and OpenAI for a three-year strategic collaboration to use classic IP for content generation indicates a new model of authorization and revenue sharing, which could benefit companies like Fubo Group that manage digital rights [3]

Core Viewpoint - The launch of Seedance 2.0 by ByteDance's subsidiary Jimo has generated significant excitement, being hailed as the strongest video generation model currently available [1][3]. Group 1: Market Reaction - Following the introduction of Seedance 2.0, the media and entertainment sector experienced a notable surge, with stocks such as Zhongwen Online, Rongxin Culture, and Jiecheng Co. reaching a 20% limit up [1]. - On February 10, major capital inflows were observed in the film and television, publishing, digital media, and gaming sectors, with net inflows of 3.09 billion, 1.168 billion, 1.141 billion, and 992 million respectively, ranking them among the top four sectors for capital accumulation that day [1]. - Individual stocks like Light Media saw over 1 billion in net capital inflow, while several others, including Aofei Entertainment and Shanghai Film, also experienced significant inflows exceeding 200 million [1]. Group 2: Seedance 2.0 Features and Limitations - Seedance 2.0 is currently available only to members, offering various modes such as theme reference and intelligent multi-frame, with the ability to use multiple images for reference in the all-purpose mode [5]. - The platform has implemented restrictions on generating videos involving celebrities or well-known IPs, indicating a focus on safeguarding against potential misuse of AI technology [7]. - The generation of videos requires points, with a 10-second video costing approximately 60 points and a 15-second video costing 90 points, while membership fees range from 659 yuan to 5199 yuan annually depending on the level [6].

2月10日，Seedance概念股多股涨停。这场由字节跳动掀起的文生视频革命，其真实价值在于重构内容 产业的成本曲线与生产范式。当电影级短片可由AI一键生成，哪些赛道将迎来变革？谁又能真正接住 这场技术红利？且看机构最新研判。 从技术演进看，Seedance 2.0的核心突破在于其"可控性"的大幅提升。传统AI视频生成存在"抽卡"痛 点，即多次生成才能获得可用画面，导致成本高企。Seedance 2.0通过多模态参考、角色一致性保持、 智能分镜与音画同步，显著降低了重复生成次数。据机构测算，即便在保守情景下，其也可降低总生成 成本约5%，若抽卡频率减半，成本降幅可达37%。 从落地场景看，Seedance 2.0最直接利好影视与漫剧板块。该模型可实现"单张角色图生成剧情视频"， 大幅降低传统影视IP开发的时间与资金门槛；AI漫剧制作成本有望从万元级降至千元级，用户规模也将 大幅提高。同时，多模态生成能力的提升也有望赋能营销、电商、游戏等内容创作环节，推动AI在广 告素材、短视频营销、互动叙事等场景的规模化应用。 从产业趋势看，2026年被视为AI应用商业化拐点年。一方面，北美云服务厂商（CSP）持续上修资本开 ...