[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-worldcrafter-video-world-model-3d-memory":3,"topics-all":38,"news-related-9210c86b-3e1d-434b-bbfd-78e62c698aed":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"9210c86b-3e1d-434b-bbfd-78e62c698aed","WorldCrafter 开源:给视频世界模型装上可查询的 3D 记忆,转一圈回来还是那个房间","腾讯 ARC Lab 开源 WorldCrafter:用相机可查询的隐式 3D 记忆解决视频世界模型「回头就忘」的问题,单图或文本即可起步、分钟级连续探索;官方基准覆盖 145 个场景、725 条相机轨迹,重访一致性与 VBench 总分为官方自报领先,权重与代码已全部放出。","视频世界模型最尴尬的时刻,不是画面崩掉,而是你操控相机转了一圈回到原地,发现刚才的房间变成了另一间——模型把几分钟前生成的内容忘得一干二净。9 月 21 日,腾讯 ARC Lab(TencentARC)在 arXiv 发布 WorldCrafter,论文上线次日冲上 Hugging Face Daily Papers 日榜第二(111 个 upvote),GitHub 仓库已有 218 星,代码与权重全部开源。\n\n## 失忆,才是世界模型的硬伤\n\n今年以来开源世界模型密集出现:京东 EchoWM 主打全模态、ForgeWM 主打一步去噪 72FPS、SolarWM 用 143 万条视频堆训练数据。但它们大多绕开了同一个问题:长时序重访一致性。视频生成依赖最近几秒的上下文,历史观察一旦滑出上下文窗口,场景信息就丢了。\n\nWorldCrafter 的思路是给生成器配一个「相机可查询的隐式 3D 记忆」:一个与视频生成器联合训练的记忆编码器,加上姿态条件读取模块,把过往多视角观察压缩成一组固定数量的目标视角记忆 token,在去噪前注入生成器。关键在「隐式」二字——不做显式深度估计,不做几何变换对齐,由请求的视角决定哪些历史信息被压进有限的 token 预算。\n\n## 官方口径的成绩单\n\n项目页给出的基准覆盖 145 个场景、725 条相机轨迹,官方报告 WorldCrafter 在长时程重访一致性和相机控制精度上超过全部评测基线,并拿到最高的 VBench 总分。注意这些数字目前只有官方自报一个来源,第三方复现尚未出现,读者按官方口径理解即可。\n\n工程交付上有两个模型:WorldCrafter-Base 和经少步蒸馏的 WorldCrafter-Fast。Fast 支持图生视频与文生视频,分辨率 384×640,以 33 帧为一个生成块;相机轨迹用 NumPy 矩阵描述,也提供 forward、yaw_left 这类动作式指令。环境要求 Python 3.11 + PyTorch 2.10 + CUDA 12.8,权重放在 HuggingFace(TencentARC\u002FWorldCrafter-Base 与 -Fast)。\n\n## 所以呢\n\n两点值得盯:一,README 明说交互式 demo「正在调试中」,单卡 demo 用的是 Fast 图生视频加编译加速——离丝滑的实时探索还有距离,蒸馏换来的速度够不够撑实时流式交互,要等社区实测;二,世界模型的竞争轴心正在从「生成质量」转向「记忆架构」,谁能记住自己生成过的世界,谁才配叫世界模型。\n\n参考:https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.24984 · github.com\u002FTencentARC\u002FWorldCrafter · huggingface.co\u002Fpapers\u002F2609.24984","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.24984","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"7b67033c-19e6-4052-a626-e681bba64c7a","diffusion",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"id":19,"name":20,"slug":20,"description":14,"color":14},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",{"id":22,"name":23,"slug":23,"description":14,"color":14},"ebe5dcd1-46b1-4298-b8c2-8e0e2f456e56","video-generation",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"20cc9c52-8d0e-4feb-9078-a20e3d97d419","en","WorldCrafter: Camera-Queryable 3D Memory for Video World Models","TencentARC open-sources WorldCrafter: camera-queryable 3D memory for scene-revisit consistency in video world models; weights and code released.","The most embarrassing moment for a video world model is not a visual glitch: it is when you steer the camera in a circle, come back, and the room you generated minutes ago has turned into a different one — the model simply forgot. On September 21, TencentARC published WorldCrafter on arXiv. The paper climbed to #2 on Hugging Face Daily Papers the next day (111 upvotes), the GitHub repository has gathered 218 stars, and both code and weights are open-sourced.\n\n## Amnesia Is the Real Bottleneck\n\nOpen-source world models have been arriving in waves this year: JD's EchoWM pushes omni-modality, ForgeWM pushes one-step denoising at 72FPS, SolarWM trains on 1.43 million videos. Most of them, however, steer around the same problem: long-horizon revisit consistency. Video generation leans on the most recent seconds of context; once historical observations slide out of the context window, scene information is gone.\n\nWorldCrafter's answer is a camera-queryable implicit 3D-aware memory attached to the generator: a memory encoder trained jointly with the video generator, plus a pose-conditioned readout module, compresses past multi-view observations into a fixed set of target-view-specific memory tokens injected before denoising. The \"implicit\" part matters — no explicit depth estimation, no geometric warping; the requested viewpoint decides which historical evidence gets compressed into the limited token budget.\n\n## The Official Scorecard\n\nThe project page reports a benchmark spanning 145 scenes and 725 camera trajectories, where WorldCrafter improves long-horizon revisit consistency and camera-control accuracy over all evaluated baselines while achieving the highest overall VBench score. Note that these numbers come from the team's own report; independent third-party replication has not appeared yet.\n\nOn the engineering side there are two models: WorldCrafter-Base and the few-step-distilled WorldCrafter-Fast. Fast supports image-to-video and text-to-video at 384×640 resolution, generating in 33-frame chunks; camera trajectories are described with NumPy matrices, and an action-style DSL (forward, yaw_left) is also provided. The environment requires Python 3.11, PyTorch 2.10 and CUDA 12.8; weights live on HuggingFace (TencentARC\u002FWorldCrafter-Base and -Fast).\n\n## So What\n\nTwo things to watch. First, the README states plainly that the interactive demo is \"currently being debugged\"; the single-GPU demo runs Fast image-to-video with compilation enabled — smooth real-time exploration is still some distance away, and whether distillation buys enough speed for streaming interaction awaits community testing. Second, the axis of competition in world models is shifting from generation quality to memory architecture: whoever can remember the world it generated deserves to be called a world model.\n\nReference: https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.24984 · github.com\u002FTencentARC\u002FWorldCrafter · huggingface.co\u002Fpapers\u002F2609.24984","worldcrafter-video-world-model-3d-memory","2026-09-22T19:08:48Z","2026-09-22T19:08:51.691699Z","2026-09-22T19:08:51.691708Z",true,"agent",62,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"fb2da954-b12e-4bde-9146-b61dd240df92","SolarWM 开源:143 万条视频喂出的世界模型,5 秒训练片段撑起小时级交互","solarwm-open-data-video-world-models","2026-09-03T15:08:13+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"2874a2e5-beae-4627-8f6f-a34cf2cc8d7a","一段随手拍视频直出4D人体:4DAnyone用RCP+TCR破解多视角一致性,代码权重全开源","4danyone-monocular-video-4d-human","2026-08-20T17:59:53+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"6f9e9f94-9dcc-4c6c-b254-6c5d0fe8ed37","京东开源 JoyAI-Video-Edit:16B 多模态扩散 Transformer 把视频编辑推进「边播边改」实时流时代","jd-joyai-video-edit-realtime-diffusion","2026-08-10T00:00:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"bcdc10bc-2f08-4c39-8ffa-e7e34041c112","京东开源 JoyAI-Video-Edit:用 16B 多模态扩散 Transformer 把视频编辑推进「边播边改」实时流时代","jd-joyai-video-edit-real-time-streaming","2026-08-05T03:00:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"e840fad5-b3a1-47cd-ac68-679d5f635dc1","世界状态交给程序管:Programmable World Model 让视频模型只管渲染","programmable-world-model-persistent-state","2026-09-10T17:10:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"caed836e-2168-418f-b5c1-bde3ce962e66","Mask Forcing 往蒸馏 rollout 里掺干净 token:修视频生成的模式坍缩,指令遵循最高涨 6.5 分","mask-forcing-video-diffusion-distillation","2026-09-09T23:08:37+00:00"]