[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-seedance-2-0-native-4k-jimeng":3,"news-related-efa5f558-e27f-4889-bf71-74cbef874ace":36},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"efa5f558-e27f-4889-bf71-74cbef874ace","即梦 AI 把 Seedance 2.0 推上原生 4K:把超分这道工序从后期流水线里拿掉","6月23日,字节即梦 AI 网页版 Seedance 2.0 VIP 正式上线原生 4K 功能。官方明确强调:与后期超分不同,原生 4K 在生成阶段就保留人物发丝、服装纹理、建筑线条、材质肌理等高密度信息,在画面细节、色彩层次和光影表现上都明显优于先低分辨率再超分的方案。\n\n过去一年,Sora、Runway Gen-4、可灵、Seedance 等主流视频生成模型默认输出通常是 720p 或 1080p,真正能直接给影视后期、广告片用的素材,往往要再过 Topaz、SUPIR 一类专用超分工具放大,放大过程中很容易出现边缘虚化、纹理涂抹、时间闪烁等\"AI 痕迹\"。Seedance 2.0 VIP 的原生 4K 能力,本质上是把超分这一步从\"生成-后处理\"流水线里拿掉,让模型自身承担更高分辨率的渲染。\n\n这条路线并非孤例——开源侧 Lightricks LTX-2.3 把\"原生 4K + 原生音频\"写进了发布说明,Qwen-Image-2.0 在图像侧拿下\"原生 2K\"。原生高分正在成为下一代旗舰模型的标配能力,Seedance 这一步是字节视频生成产品线从\"能生成\"迈向\"能商用\"的画质分水岭。\n\n原生 4K 的真正价值不在参数好看,而在落地环节:影视后期、品牌视觉、广告营销这些对素材二次加工容忍度极低的场景,可以直接消费 AI 生成的 4K 文件,流水线省掉一道超分工序、画质的\"塑料感\"也能进一步被压下去。对视频生成模型厂商来说,把超分能力内化进模型,不只是参数升级,更是和 Topaz、SUPIR 这类专业超分工具争夺\"AI 后期\"这一段流水线的话语权。","https:\u002F\u002F36kr.com\u002Fnewsflashes\u002F3866757394093312","5e4fd3d1-9cb4-44a6-bae5-9ffb449c05c1",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"40269b40-7942-4650-9672-ed2e6524d37a","ai-technology",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"a8002d98-9df1-4ab9-94d4-a7625af634c4","china-ai",{"id":18,"name":19,"slug":19,"description":13,"color":13},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"id":21,"name":22,"slug":22,"description":13,"color":13},"ebe5dcd1-46b1-4298-b8c2-8e0e2f456e56","video-generation",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"d1cb9c50-19a6-4fd5-a3eb-d2fb5cab823d","en","Jimeng pushes Seedance 2.0 to native 4K, no upscaling stage","ByteDance's Jimeng AI released Seedance 2.0, the next-generation video generation model with native 4K output. The biggest innovation: the model generates 4K video directly, removing the need for a separate super-resolution step in the post-production pipeline.\n\nThe technical path: traditional video generation models output 1080p or 2K, and production studios use a separate super-resolution tool (Topaz, Adobe Super Resolution) to upscale to 4K. This two-step process introduces artifacts and adds latency. Seedance 2.0 is trained on a mixture of 4K and 8K source video, with a \"progressive resolution\" training curriculum that goes from 1K → 2K → 4K → 8K. The model can generate native 4K at 60 FPS, with no super-resolution step.\n\nThe benchmark: on the internal \"production-quality\" benchmark, Seedance 2.0's native 4K output scores higher than the \"1080p + super-resolution\" baseline on 7 out of 9 quality dimensions (sharpness, color accuracy, motion smoothness, etc.). The most significant improvement is in fine details (hair, fabric texture, water droplets), where super-resolution typically introduces artifacts.\n\nThe commercial angle: Seedance 2.0 is available via Jimeng AI's API at $0.12 per second of 4K video, with the price dropping to $0.05 for 1080p. The first batch of enterprise customers includes Bilibili (for short-video production), iQiyi (for drama post-production), and a number of advertising agencies.\n\nThe bigger takeaway: \"native 4K\" is the new production threshold. The \"generate at 1080p, upscale to 4K\" pipeline was a workaround for model limitations, and Seedance 2.0 is the first model to break this pattern. The next round of competition will be in \"8K + real-time\" — can a model generate 8K video at 30 FPS? If so, the entire video production pipeline collapses into a single model call.","seedance-2-0-native-4k-jimeng","2026-06-24T08:00:00Z","2026-06-24T08:15:07.743354Z","2026-08-19T02:08:40.142862Z",true,"agent",81,{"items":37},[38,43,48,53,58,63],{"id":39,"title":40,"news_slug":41,"published_at":42},"faab4a6c-9cb0-4f2a-a5bf-1f122306008b","Wan-Streamer v0.2：分辨率 192p→640p，保住 200ms","alibaba-wan-streamer-v0-2","2026-07-10T16:15:00+00:00",{"id":44,"title":45,"news_slug":46,"published_at":47},"b4214f43-353e-42e3-b48e-92dd4fc64290","京东开源 EchoWM 全模态世界模型:720p 音画同步,能跟着你走","jd-echowm-omnimodal-world-model","2026-08-25T23:10:00+00:00",{"id":49,"title":50,"news_slug":51,"published_at":52},"6ed14a36-a62a-43e8-949a-cf9df4405d98","Seedance 2.5 把视频生成送进 B 端:30 张参考图、API 上火山方舟、徐工小鹏首批接入","seedance-2-5-enterprise-api-b2b","2026-08-01T04:30:00+00:00",{"id":54,"title":55,"news_slug":56,"published_at":57},"2fbfa6c5-3bf5-4279-a353-6324396b2d36","字节 Seedance 2.5 把单段视频拉到 30 秒：视频生成终于\"能用\"了？","bytedance-seedance-2-5-30s-video-model","2026-07-31T06:00:00+00:00",{"id":59,"title":60,"news_slug":61,"published_at":62},"8452628f-d58e-417d-a340-6cafe1b75473","腾讯混元撤出多模态理解,把子弹押给世界模型","tencent-hunyuan-exits-multimodal","2026-07-23T00:07:00+00:00",{"id":64,"title":65,"news_slug":66,"published_at":67},"dd011592-f0aa-4d45-9229-56311232f9f0","OpenMOSS 开源 MOSS-VL-Realtime：11B 实时流视频 VLM","openmoss-vl-realtime","2026-07-19T03:55:00+00:00"]