《第九区》导演 Blomkamp 发布 AI 短片《Nightborne》，全片由字节跳动 Seedance 2.0 生成

周一，《第九区》和《赛车总动员》导演尼尔·布洛姆坎普（Neill Blomkamp）公布了他的最新项目：一部13分钟的科幻短片《夜生者》（Nightborne），灵感大致来源于彼得·沃茨（Peter Watts）2014年的小说《回声行为》（Echopraxia）。这部短片出自布洛姆坎普的新人工智能初创公司/制作公司 Barley Studios，并且片中的角色的声音和面孔都基于真人演员建模。但《夜生者》的每一个镜头都是使用字节跳动的 Seedance 2.0 文本到视频生成器制作的。

在关于这部短片的X帖子中：https://x.com/NeillBlomkamp/status/2079078300660769193，布洛姆坎普将其描述为一个“测试开始”，旨在展示生成式人工智能的能力，他还表示，希望未来某个时间“以这种形式制作一部完整的长片”。布洛姆坎普似乎认为，《夜生者》是一个有力的概念验证，可能会引起人们对他下一步创作的兴趣。但即便这部“影片”相比如今网络上漂浮的大多数AI生成视频更加精致，它仍然带有许多我们通常会认为粗糙的特征。即便制作过程中有一支真实的团队参与，《夜生者》依然糟糕到让人很难称其为电影制作的未来。

《夜生者》的背景设定在美国与叙利亚之间的军事冲突中，讲述了一位美国飞行员的故事，她的直升机被击落后身受危及生命的伤害。虽然飞行员的大部分身体严重受伤，但她仍然在技术上存活，头部未受损伤。这些条件使她成为“夜生者计划”（Project Nightborne）的理想候选人——这是一个将脑死亡士兵改造成远程控制武装僵尸的政府计划。而由于军方可以无限获取尸体，这名飞行员只是众多正在秘密接受战争训练的制服化食尸鬼之一。

像以前所有的 AI 生成的电影一样，《夜生》实际上是由一些非常短的视频片段剪辑在一起，以讲述一个简单的故事。《夜生》经常将其僵尸化的主角镜头切换到关于“夜生计划”的纪录片风格采访镜头。这样的剪辑方式给由奥斯丁·丹尼斯（Austyn Daines）剪辑的影片提供了一种掩盖事实的手段——Seedance 2.0 每次只能生成几秒钟的视频素材。影片对白的混音方式几乎（但不完全）让人觉得你在听真实的人说话。

布隆坎普的创作声音几乎不存在

当你仔细观察《夜生》的视觉效果，并认真倾听其音频时，很容易注意到其中包含了大量生成式 AI 元素。背景中的标志通常是完全的乱码，角色的台词强调方式不寻常，使他们听起来像是完全不知道自己在说什么。根据布隆坎普的说法，“真正的概念艺术家”为该项目贡献了作品，还有 32 名人类演员同意在屏幕上使用他们的肖像。但这些人类输入似乎远远不足以让《夜生》感觉更像是一件艺术品，而非机器生成的内容。

除了出自同一位导演，他曾构思过《Joburg 生存》： https://www.youtube.com/watch?v=iNReejO7Zu8——这是一个运用视觉特效探讨种族隔离时期偏见思想的创意练习：https://www.indiewire.com/news/general/neill-blomkamps-alive-in-joburg-224248/ ——《夜生》本身没有什么特别之处。随着短片故事的发展，布隆坎普的创作声音几乎不存在，如果没有 Seedance 2.0 的参与，人们大概也不会谈论它。

尽管 Seedance 2.0 显然能够生成人物情绪视频，《夜生》却凸显了该模型仍然无法真正理解使人类表演自然、生动的细微细节。短片中你可能看到的任何情感细微之处实际上只是 Seedance 2.0 被训练的视频的反映，并不完全是它能够自豪地宣称的能力。

除了僵尸之外，《夜裔》与其原始素材并没有太多共同点。但它非常像贾樟柯的短片《贾樟柯的舞蹈》和Ruairí Robinson的病毒式AI视频，视频中展示了汤姆·克鲁斯和布拉德·皮特的对打——这两部作品都是用Seedance 2.0制作的：/ai-artificial-intelligence/883615/seedance-bytedance-tom-cruise-brad-pitt-jia-zhangke。这些项目都表明，传统电影制作人往往知道如何淡化生成式AI的技术局限，但他们的技巧只能在一定程度上弥补这种技术的固有弱点。

尽管文本到视频模型可能变得越来越复杂，但它们还没有发展到能够制作出观众期待观看的完整作品的程度。《夜裔》的发布几乎立刻引发了一波反对浪潮：https://www.yahoo.com/entertainment/movies/articles/district-9-director-neill-blomkamp-200423166.html，许多人将这部影片视为布洛姆坎普被“淘汰”的象征。尽管布洛姆坎普可能认为《夜裔》会为自己带来一些积极的讨论，但现在看来，他更像是不幸卷入了一个类似达伦·阿伦诺夫斯基因为其关于独立战争的糟糕AI系列而制造的局面：https://deadline.com/2026/02/darren-aronofsky-auteur-ai-youtube-backlash-1236709982/。

关于《夜裔》，最有趣的事情是其故事可以被解读为对制作它所使用技术的评论。《夜裔》项目讲述的是掌权者在未征得死者同意的情况下重新利用他们的身体，以廉价地获取劳动力。生成式AI并不像杀人僵尸士兵那样（立即：https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117 或明显：/news/818508/chatbot-eating-disorder-mental-health）具有破坏性，但推动这一技术使用的人，其动机与《夜裔》中的反派所驱动的欲望相似。

On Monday, District 9 and Gran Turismo ：/23707049/sony-gran-turismo-movie-trailer director Neill Blomkamp unveiled his latest project: a 13-minute sci-fi short titled Nightborne that’s loosely based on Peter Watts’ 2014 novel Echopraxia . The short comes from Blomkamp’s new AI startup / production company, Barley Studios, and features characters whose voices and faces are modeled after human actors. But every single one of Nightborne ’s shots was made with ByteDance’s Seedance 2.0 text-to-video generator：/ai-artificial-intelligence/877931/bytedance-seedance-2-video-generator-ai-launch.

In an X post about the short：https://x.com/NeillBlomkamp/status/2079078300660769193, Blomkamp described it as a “test start” meant to demonstrate what generative AI is capable of, and he said that he wants “to tackle a full feature in this format” at some point in the future. Blomkamp seems to think that Nightborne is a strong proof of concept that might get people intrigued to see what he prompts next. But as polished as the “film” might be compared to most of the AI-generated videos floating around the internet today, it still bears many of the hallmarks we associate with slop. And even though a team of real people was involved in the project’s production, Nightborne is such a terrible watch that one would be hard-pressed to call it the future of moviemaking.

Set against the backdrop of a military conflict between the US and Syria, Nightborne tells the story of an American pilot who sustains life-threatening injuries when her helicopter is shot down. Though most of the pilot’s body is badly injured, she’s still technically alive and her head is intact. Those things make her the ideal candidate for Project Nightborne — a governmental program that turns brain-dead soldiers into remote controlled, weaponized zombies. And because the military has an endless supply of bodies at its disposal, the pilot is one of many uniformed ghouls being trained for war in secret.

Like basically every other AI-generated film that has come before it, Nightborne is really a collection of very short video clips that have been edited together to tell a simple story. Nightborne frequently cuts from shots of its zombified main character to documentary-style interviews with people involved in Project Nightborne. These cuts give the film — which was edited by Austyn Daines — a way to obfuscate the fact that Seedance 2.0 is only capable of generating a few seconds of video footage at a time. And the way the film’s dialogue is mixed almost (but not quite) makes it seem like you’re listening to real people speaking.

Blomkamp’s creative voice feels almost nonexistent

When you really take a good look at Nightborne ’s visuals and listen closely to its audio, it’s easy to pick up on how much gen AI is in the mix. Signs in the background are often written in complete gibberish, and characters deliver their lines with unusual emphases that make them come across like people who don’t actually know what they’re saying. According to Blomkamp, “real concept artists” contributed work to the project, and 32 human actors agreed to have their likenesses used on-screen. But that human input doesn’t seem like it’s anywhere near enough to keep Nightborne from feeling more like machine-made content than a piece of art.

Aside from it coming from the same director who dreamt up Alive in Joburg ：https://www.youtube.com/watch?v=iNReejO7Zu8— an imaginative exercise in using VFX to explore ideas about apartheid-era bigotry：https://www.indiewire.com/news/general/neill-blomkamps-alive-in-joburg-224248/ — there’s nothing remarkable about Nightborne . Blomkamp’s creative voice feels almost nonexistent as the short’s story unfolds, and if Seedance 2.0 wasn’t part of the equation, people probably wouldn’t be talking about it.

Though Seedance 2.0 can clearly churn out videos of people emoting, Nightborne highlights how the model still doesn’t have any sort of real understanding of the minute details that make human performances feel organic and truly alive. Whatever emotive nuance you might see in the short is really just a reflection of the videos Seedance 2.0 has been trained on, and not exactly something that it can take credit for.

Aside from zombies, Nightborne doesn’t have all that much in common with its source material. But it’s a lot like Jia Zhangke’s Jia Zhangke’s Dance short and Ruairí Robinson’s viral AI video depicting Tom Cruise and Brad Pitt fighting — both of which were created with Seedance 2.0：/ai-artificial-intelligence/883615/seedance-bytedance-tom-cruise-brad-pitt-jia-zhangke. Each of these projects has demonstrated that traditional filmmakers tend to know how to downplay gen AI’s technical limitations, but their skills can only do so much to offset the technology’s innate weaknesses.

Though text-to-video models may be getting more sophisticated, they haven’t advanced enough to produce finished works that audiences are excited to see. Nightborne ’s announcement was almost immediately met with a wave of backlash：https://www.yahoo.com/entertainment/movies/articles/district-9-director-neill-blomkamp-200423166.html from people who saw the film as a sign of Blomkamp being washed. And while Blomkamp might have thought that Nightborne would generate some positive buzz for himself, it seems much more like he’s wandered into an unfortunate situation similar to the one Darren Aronofsky created with his terrible AI series about the Revolutionary War：https://deadline.com/2026/02/darren-aronofsky-auteur-ai-youtube-backlash-1236709982/.

The most interesting thing about Nightborne is the way its story can be interpreted as commentary on the technology that was used to make it. Project Nightborne is all about individuals in positions of power repurposing dead people’s bodies without their consent in order to generate labor on the cheap. Generative AI isn’t as (immediately：https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117 or obviously：/news/818508/chatbot-eating-disorder-mental-health) destructive as killer zombie soldiers, but those pushing for the technology’s use are driven by a desire that’s similar to the one motivating Nightborne ’s villains.