[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fYwgn16GvWGP61Y_5p7bjC-wP4-9PoUBsU7n6DY_FLyc":3},{"code":4,"msg":5,"data":6},200,"操作成功",{"id":7,"title":8,"content":9,"digest":10,"source":10,"coverPath":11,"thumbsCoverPath":12,"isTop":13,"isShow":14,"baseClick":13,"clickCount":15,"createTime":16,"typeId":17,"isNewest":18,"newsInfoTypeRespVo":19,"voiceUrl":22,"voiceSize":23,"taskId":24,"releaseTime":25,"titleEn":26,"contentEn":27,"voiceUrlEn":28,"taskIdEn":29,"voiceSizeEn":30},1290,"AI视频生成走向规模化应用","\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">花300多元成本，就能制作一部科幻短片？过去需要一整支团队才能完成的视频制作，如今只需一张图、一句话。从玩具成长为生产工具，AI视频生成技术正以惊人的速度走向规模化应用。\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">330元做出影视级科幻短片\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">“密码0821，快去重启，消灭他们。”“收到！”当男主角接到指令，历尽艰难输入重启控制台的密码，在暗无天日如同世界末日般的场景下，给人类带来巨大威胁的异形生物瞬间崩成碎粒、消失不见……8月21日，国际知名视效指导姚骐展示了一段自己用AI制作的2分钟科幻短片《归途》。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">传统影视制作中演员、场地、设备租赁、后期配音、特效追加等环节费用不菲，一直是行业成本居高不下的原因，如今这些环节都能通过AI生成实现。“它让好莱坞级大片镜头不再需要百万元预算。”姚骐透露，《归途》有40多个镜头，每个镜头生成3次，共用AI生成超120个片段素材，其中包括18个10秒一体化的有声片段和102个5秒片段，最终花费约一周时间制作完成。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">姚骐创作《归途》所用的AI工具是百度“蒸汽机”（MuseSteamer）音视频一体化模型。当天，该模型宣布重大升级。与业界大量AI生成视频大多为“默片”不同，“蒸汽机”首次实现多人有声音视频一体化生成。百度商业体系商业研发总经理刘林介绍，使用“蒸汽机”制作的《归途》，整体成本仅330.6元。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">先行者开启收获期\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">从百度“蒸汽机”、字节跳动即梦、快手可灵AI，到智谱清影、生数科技Vidu、MiniMax海螺AI，互联网巨头与初创企业都相继布局AI视频生成模型。放眼国际市场，视频生成模型领域也是竞争激烈。记者注意到，在AI视频生成模型从“玩具”成长为生产工具的过程中，一些玩家已进入收获期。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">近期，一部由热门IP故事改编，主要由AI视频生成的同名漫剧《明日周一》成为“爆款”，在抖音平台上线5天播放量累计突破500万。记者从生数科技获悉，该公司视频大模型Vidu在上线的8个月内已实现年化经常性收入突破2000万美元，用户覆盖全球200多个国家和地区。快手21日发布财报显示，可灵AI的收入在二季度超过2.5亿元，相比一季度大幅增长。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">相比于产生直接收益，百度则将AI视频生成模型视为撬动其整体生态活跃度的重要支点。“如果‘蒸汽机’能对百度搜索流量、用户留存带来1个百分点的贡献，收益都将十分可观。”百度副总裁、移动生态商业体系负责人陈一凡说。近期，百度移动搜索结果页中AI生成的内容占比已从4月的35%提升至7月的64%，AIGC（人工智能生成内容）的涌现大幅提升了多模态内容创作的自由度。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">QuestMobile最新发布的《2025中国移动互联网半年大报告》显示，由AI技术驱动的变革正深刻重塑行业格局，AIGC行业使用时长同比飙升393.9%。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">进一步破圈仍待技术突破\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">尽管发展迅猛，AI视频生成模型若要完全支撑起长视频内容的工业化生产，仍需突破一系列技术瓶颈。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">一家互联网大厂AI视频产品运营负责人李杨表示，当下AI创作尚未达到一键生成、拿来即用的程度，需要创作者通过AI工具反复写prompt（提示词）、调prompt、抽卡、调视频，非常考验创作者的创造力与韧性，因此目前其消费群体主要是相对专业的创作者。AIGC视频生成市场的进一步爆发，还需凭借技术和体验的突破，实现使用群体从专业创作者向大众群体的跨越。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">技术瓶颈还体现在视频长度上。当前，行业内视频生成模型提供的视频生成长度大都以5秒、10秒为主，且难以做到实时生成。百度商业研发首席架构师李双龙分析，这主要跟当下业内普遍采用的模型架构有关，在这一架构下，视频长度一旦翻倍，就可能导致成本100倍的增长。要解决这一问题，必然需要实现技术的代际革新。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">【新闻来源】北京日报 记者 孙奇茹 \u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fwww.sohu.com\u002Fa\u002F928602157_163278?scm=1019.20001.0.0.0&spm=smpc.csrpage.news-list.5.1756260668043EiNIYPu\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(187, 187, 187);\">https:\u002F\u002Fwww.sohu.com\u002Fa\u002F928602157_163278?scm=1019.20001.0.0.0&spm=smpc.csrpage.news-list.5.1756260668043EiNIYPu\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">（本网转发此文章，旨在为读者提供更多的信息资讯，所涉内容不构成投资、消费建议。文章事实如有疑问，请与有关方核实，文章观点非本网观点，仅供读者参考。）\u003C\u002Fspan>\u003C\u002Fp>","","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F3d727dc1ec364b16937682be83d9f1be\u002FAI领域.jpg","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002Fthumbs\u002F3d727dc1ec364b16937682be83d9f1be\u002FAI领域.jpg",0,1,217,"2025-08-28 11:02",2,false,{"id":17,"name":20,"enName":21},"芯位视野","Xinwei Vision","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3Ad719a866-83e8-4749-86c2-12fac866ddb5%3A0.wav?Expires=1756379396&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=MfYAwcP3hAqDgAPq4iR%2BWwbTKnY%3D",8288150,"d719a866-83e8-4749-86c2-12fac866ddb5","2025-08-28 11:00","AI Video Generation Moves Toward Large-Scale Application","\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">Spending over 300 yuan, you can produce a science fiction short film? What used to require an entire team can now be done with just one image and one sentence. From a toy to a production tool, AI video generation technology is rapidly moving toward large-scale application.\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">Create a film-level sci-fi short film for 330 yuan\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">“Password 0821, go restart it and eliminate them.” “Understood!” When the male lead received the order, he struggled to input the password to reboot the control panel in a dark scene like the end of the world, and the alien creatures threatening humanity instantly collapsed into particles and disappeared... On August 21, the internationally renowned visual effects director Yao Qi showcased a two-minute sci-fi short film \"Return Journey\" he made with AI.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In traditional film production, costs for actors, locations, equipment rentals, post-production voiceovers, and special effects are high, which has always been the reason for high industry costs. Now, these stages can all be achieved through AI generation. \"It no longer requires a budget of millions for Hollywood-style shots.\" Yao Qi revealed that \"Return Journey\" had over 40 shots, each generated three times, resulting in more than 120 AI-generated clips, including 18 10-second audio-visual segments and 102 5-second segments, and was finally completed in about a week.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The AI tool Yao Qi used to create \"Return Journey\" is Baidu's \"MuseSteamer\" audio-video integrated model. On that day, the model announced a major upgrade. Unlike most AI-generated videos in the industry that are silent, \"MuseSteamer\" has achieved the first integration of multi-person audio-visual generation. Liu Lin, general manager of Baidu Commercial R&D, introduced that the overall cost of \"Return Journey\" created using \"MuseSteamer\" was only 330.6 yuan.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">Pioneers Enter the Harvest Period\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">From Baidu's \"MuseSteamer,\" ByteDance's Ji Meng, Kuaishou's Ke Ling AI, to Zhipei Qingying, Shengshu Technology's Vidu, MiniMax's Hai Luo AI, internet giants and start-ups have all successively entered the field of AI video generation models. Looking at the international market, the video generation model sector is also highly competitive. Journalists noticed that during the process of AI video generation models evolving from \"toys\" into production tools, some players have already entered the harvest period.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Recently, a popular IP story-based animated series \"Tomorrow Monday\" adapted mainly by AI video generation became a hit, with cumulative views exceeding 5 million on the Douyin platform within five days of its release. According to information from Shengshu Technology, the company's video large model Vidu has achieved annualized recurring revenue exceeding $20 million within eight months of its launch, with users covering over 200 countries and regions globally. Kuaishou's financial report released on the 21st showed that Ke Ling AI's revenue in the second quarter exceeded 250 million yuan, a significant increase compared to the first quarter.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Compared to direct revenue, Baidu sees the AI video generation model as an important lever to drive the overall ecological activity. \"If 'MuseSteamer' contributes 1 percentage point to Baidu search traffic and user retention, the revenue would be very considerable,\" said Chen Yifan, vice president of Baidu and head of the mobile ecosystem commercial system. Recently, the proportion of AI-generated content in Baidu's mobile search results has increased from 35% in April to 64% in July, and the emergence of AIGC (Artificial Intelligence Generated Content) has significantly enhanced the freedom of multimodal content creation.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">According to the latest \"2025 Half-Year Report on the Chinese Mobile Internet\" released by QuestMobile, the transformation driven by AI technology is deeply reshaping the industry structure, and the usage time of the AIGC industry has surged by 393.9% year-on-year.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">Further Expansion Still Requires Technological Breakthroughs\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Although development is rapid, AI video generation models still need to overcome a series of technological bottlenecks to fully support the industrial production of long-form video content.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Li Yang, the operations director of an internet giant's AI video product, said that current AI creation has not yet reached the level of one-click generation and immediate use, requiring creators to repeatedly write prompts (prompt words), adjust prompts, draw cards, and adjust videos, which greatly tests the creativity and perseverance of the creators. Therefore, the current consumer group is mainly relatively professional creators. The further explosion of the AIGC video generation market still needs to rely on technological and experiential breakthroughs to enable the transition of the user base from professional creators to the general public.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Technical bottlenecks are also evident in video length. Currently, the video generation lengths provided by industry video generation models are mostly 5 seconds or 10 seconds, and real-time generation is difficult. Li Shuanglong, chief architect of Baidu commercial R&D, analyzed that this mainly relates to the widely adopted model architecture in the industry. Under this architecture, if the video length doubles, the cost could increase by 100 times. To solve this problem, a generation of technological innovation is necessary.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">[News Source] Beijing Daily, Reporter Sun Qiru \u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fwww.sohu.com\u002Fa\u002F928602157_163278?scm=1019.20001.0.0.0&spm=smpc.csrpage.news-list.5.1756260668043EiNIYPu\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(187, 187, 187);\">https:\u002F\u002Fwww.sohu.com\u002Fa\u002F928602157_163278?scm=1019.20001.0.0.0&spm=smpc.csrpage.news-list.5.1756260668043EiNIYPu\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">（This article is reprinted by this website to provide readers with more information and news. The content does not constitute investment or consumption advice. If there are any questions about the facts of the article, please verify with the relevant parties. The views of the article do not represent the views of this website and are for reference only.）\u003C\u002Fspan>\u003C\u002Fp>","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3A5dce29bf-b04f-4590-9606-c7f5f7f88d6f%3A0.wav?Expires=1774838483&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=shl2qqFe5a38CfbWb4EqUcvABW4%3D","5dce29bf-b04f-4590-9606-c7f5f7f88d6f",11368828]