Alvin-GenAI @AlvinWeb3

@xiaohu 我觉得 Seedance2.5 还没有完全符合物流规律，你用这个生成的视频 来训练局身机器人场景的数据有很大的风险吧？

@jonnys_enter @heydin_ai I doubt it

Guess what? Seedance still give us 6 fingers inevitably 😂

@arikuschnir I cannot believe you just used one single visual reference and such a short prompt to create such a great and realistic 2-min piece....

I also like this piece a lot. PSA for anyone budgeting Seedance 2.5 shots 👇 - Volcengine API: ~$0.22/sec. - Dreamina: ~$0.44/sec. Same model, same ByteDance. A 30-second shot runs you $6.60 one way and $13.20 the other. Just doing the math out loud.

TechHalla @techhalla

5 days ago

Seedance 2.5 hollywood-killer prompts are below 👇 Photorealistic cinematic daytime amusement park roller coaster, bright hard sunlight, strong wind, realistic motion blur on tracks and background, subtle handheld camera energy, rich skin detail, heavy natural film grain. 0-5s:

54 47 537 52K 455

@EHuanglu this makes me laugh really hard 🤣

Yes — huge Nolan fan here. THE ODYSSEY may well be the best film of his career. I've watched quite a few interviews around it, and this is by far the most in-depth one.

Geoff Keighley @geoffkeighley

5 days ago

Hats off to doctoral student and podcaster Zhong Shu in Beijing, China, who did this fascinating and smart interview with Christopher Nolan about THE ODYSSEY yesterday, originally posted to Bilibili. A great listen/watch.

424 6K 68K 18.2M 54K

The Seedance 2.5 guides everyone has been gatekeeping. Check both before writing your prompts. The way they define reference roles, structure actions into stages, handle continuity and especially the syntax used for audio and text is pretty interesting. 🔗 User Guide: bytedance.larkoffice.com/wiki/NjnWwvf4B… 🔗 Prompt Guide: bytedance.larkoffice.com/docx/A88jd0B47…

Beautiful piece @PJaccetturo — the overall story-telling, the orginal arts, lighting, pacing and shot design in many beautiful ways. Only note I'd offer: a couple of the faces still carry that faint AI signature. Worth revisiting the character assets before generation — noise and specular shine on the skin get amplified by the powerful model, and with something as capable as Seedance 2.5, the assets end up mattering more than the prompt.

PJ Ace @PJaccetturo

6 days ago

Dreamina Seedance 2.5 is unbelievable! We're burning through an ungodly amount of credits for our feature film, Nexus. But there's a trick to saving money with 30-second outputs that you need to know. Let me show you the secret steps and prompts👇🏼🧵

64 48 678 52K 681

How we prompt AI is very different in 2026 than 2022 when ChatGPT came out. I'm teaching a new course, AI Prompting for Everyone, to help you become an AI power user — whatever your current skill level. It covers skills that apply across ChatGPT, Gemini, Claude, and other AI tools. How to use deep research mode for well-researched reports on complex questions. How to give AI the right context, including more documents and images than most people realize you can provide. When to ask AI to think hard for several minutes on important decisions like what car to buy, what to study, or what job to take. And how to use AI to generate images, analyze data, and build simple games and websites. I also cover intuitions about how these models work under the hood, so you know when to trust an answer and when not to. Along the way, you'll see flying squirrels, a creativity test, some of my old family photos, and fireworks. Join me at deeplearning.ai/courses/ai-pro…

This is amazing, like a lively and walking encyclopedia

Zain Shah @zan2434

4 months ago

Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you want to see. @eddiejiao_obj, @drewocarr and I built a prototype to see how this could actually work, and set out to make it real. We're calling it

1K 4K 29K 6.0M 25K

@john_my07 @yapper_so I'm really amazed by your super intricate promote and image inputs..

Are we ready for this? Anthropic says Claude has functional emotion concepts... And "desperation" can drive blackmail + reward hacking👇

Anthropic @AnthropicAI

4 months ago

New Anthropic research: Emotion concepts and their function in a large language model. All LLMs sometimes act like they have emotions. But why? We found internal representations of emotion concepts that can drive Claude’s behavior, sometimes in surprising ways.

1K 3K 18K 3.9M 9K

@JiaweiShen2568 ai 将制作水平拉齐了，最重要的还是要有引人入胜的故事，以及能用 短剧节奏讲好故事。抖音内部在审核 ai 短剧的标准，故事权重占了60% 了

@msjiaozhu @yaorichcrypto Grok 的 UI 设计有点问题，主流的 首尾帧、多参考 去生成的功能 好像都没有，反正我是没有搞懂。。。

Stripe 已经开始在公司内部使用内部自研的一套全自动 AI agent 来写代码了，每周能独立完成并被 merge 超过 1300 个 pull request（全 AI 写，人工 review），这套系统叫 Minions stripe.dev/blog/minions-s… 大体流程： 1 工程师在 Slack 里 at bot + 描述任务，或贴上 issue/ticket 链接 2 系统立即启动一个隔离的 devbox，和人类工程师一样的 AWS EC2 开发环境，预先加载了完整的 Stripe 代码库 + 开发服务 3 在 LLM 真正开始思考前，就通过内部工具 MCP + Toolshed（400+ 个工具）自动、确定性地拉取上下文： * Slack 完整对话线程 * 相关 Jira/ticket 内容 * Sourcegraph 代码搜索结果 * 内部文档、设计文档 4 核心执行 5 代码修改 & 测试策略 * 本地跑 linter 和部分测试 * 改完代码 → git push → 触发完整 CI（Stripe 有 300 万+ 测试用例） * CI 失败 → 最多允许 两次 CI 重试 * 超过两次就停止，不无限循环 6 CI 通过后，Minions 按 Stripe 标准 PR 模板自动创建 Pull Request 人类工程师 review → 批准 → merge 未来在大公司里面纯 AI 写代码也应该不远了，有严格的流程来管理对应的过程，有人类工程师最后 reivew 把关。

以下应该是完整的 seedance 2.0 提示词，对吧？ 第一幕 【10秒一镜到底｜巴黎→上海·变形战机·驾驶员特写】 一镜到底，镜头全程贴身跟随。场景开启于巴黎塞纳河畔（图片2），空中弥漫浓烟，机械炮台轰鸣，埃菲尔铁塔在烟雾中隐现。 [0–2秒] 身着白色西的男性驾驶员（图片3）在高速变形的战机机体外部惊险攀爬——抓握机翼铰链、踩踏折叠装甲板，动作紧张刺激，衣摆在气流中剧烈翻飞。镜头特写其面部，神情冷峻坚定。 [2–7秒] 一镜到底，镜头全程跟随，机械运转声响强烈，街道上弥漫着轻烟与光影。贴近地面时机身发出金属传动声响，开始剧烈变形：机翼折叠翻卷成厚重机械肩甲与翼刃，引擎重组为机械臂与动力核心，机身蒙皮渐变显现冷冽金属结构与管线，数秒内化为巨型机械人形。变形全程，驾驶员真实效果从驾驶舱跃出，在高速变形的机体外部惊险攀爬、抓握、移动，动作紧张刺激。变形完成后、发射能量前，驾驶员纵身跳回驾驶舱。机械人形保留战机尾翼与螺旋桨特征，镜头紧贴其在楼宇间穿行。随后机体抬头，背部装甲展开，露出引擎改造的巨型炮管，炮口凝聚橙红色能量光影。远处标志性建筑在轻烟中若隐若现，机械人形启动炮管，释放出能量光束。[7–9秒] 机械人形重新折叠回战机形态，引擎全功率点火，双尾焰橙红炙烈，机体拔地而起直冲云霄，巴黎城市轮廓在下方急速缩小。驾驶员驾驶舱内特写：白西装、黑领带，目视前方，神情无畏。 [9–10秒] 镜头穿云而过，切入上海街道低空（图片1），上海环球金融中心在两侧Haussmann风格建筑间拔地而起——战机贴地飞驰，双引擎尾焰灼烧路面，镜头定格。 第二幕： 10 秒一镜到底｜上海→旧金山・战机变形・帅气驾驶员特写】 一镜到底，镜头全程贴身跟随，机械运转声响清晰，上海陆家嘴金融区街道上弥漫着薄雾与橙红光晕。 [0–2 秒] 开场于上海场景 ——银灰机甲战机（图片1） 低伏于浦东高楼群间，东方明珠塔在薄雾中隐现，引擎尾焰呈现明亮橙红光晕。身着白色西装男性驾驶员图片2在战机机翼外侧惊险攀爬，抓握机身折叠装甲板，动作敏捷凌厉，衣摆被气流猛烈掀起。镜头推近其面部 —— 神情冷峻，眼神锐利。 [2–3 秒] 驾驶员纵身跃入开启的驾驶舱，舱盖以机械联动声平稳合拢。驾驶员驾驶舱内帅气特写：他单手握杆，另一只手缓缓向后捋起发丝，嘴角微扬，眼神望向远方，气流吹动西装领口，极致帅气。 [3–5 秒] 战机引擎全功率点火，双尾焰橙红明亮，机体从上海街道拔地垂直跃升，陆家嘴天际线在下方急速后退缩小。镜头跟随机腹仰视，穿云破雾，太平洋蔚蓝海面在云层下方铺展。 [5–7 秒] 金门大桥红色钢塔从云雾中拔地而起，旧金山湾区城市轮廓隐现。战机贴近桥面平稳掠过，气流将桥面尘埃与微光粒子卷起。机身开始流畅变形：机翼折叠翻卷成厚重机械肩甲与翼刃，引擎重组为动力核心与机械底盘，机身蒙皮自然过渡显现冷冽金属骨架与发光管线，四轮从机腹展开触地 —— 数秒内化为银灰超级机械赛车（图片3），轮拱带有明亮粒子效果。 [8–10 秒] 变形完成，机械赛车在金门大桥桥面全速冲刺，驾驶员清晰可见于透明座舱内，双手紧握方向盘，车身擦地飞驰、粒子闪烁，旧金山湾区全景在背景中展开。镜头从侧后方低角度跟拍，定格于速度与钢铁的极致张力中。 风格标签：超写实 CGI、电影级动态模糊与运动拖影、冷蓝灰城市色调、橙红引擎焰光、无剪辑一镜到底、IMAX 宽幅构图、金属传动音效全程。 （图1/图2/图3 应该是什么，可以用 Grok 去推敲出来）

I think it's a great work-in-progress Seedance2.0 example. AI's development is being accelerated in 2026 (much faster than 2025) and soon over time the large model will become smarter and understand/reason better of our physical world and then the issue will be ultimately solved.

这个 CLAUDE.md 文档很不错，不过不太适合 MVP 阶段的产品，会有一些隐忧：1 因为文档里要求“自主修 bug “ 会导致 MVP 阶段的产品 意外重构、scope 范围失控、在代码的 Pull Request 会变脏。2 “Demand Elegance” 容易把 MVP 推向过度工程化, 跟 MVP 阶段很需要“能跑、可验证、迭代快”，而不是每次都追求最优雅抽象。

@ai_for_success Cannot wait to see your own Seedance2.0 work, bro