WHAT NEW IN X11.01 SUN

X 上的 AI 合集档案 · 22 个合集 · 64,204 条

gpt-6VIDEO

Opus 5.5 与 GPT-6 Astra 视频生成对比

作者拿 Opus 5.5 和 GPT-6 Astra 比视频生成,认为真正的突破在于整段场景的可信度,而不只是单帧画质。

Tolik
@tolik1230818 天前收录原推发布于 2026.10.14
Tolik @tolik12308原推发布于 2026.10.14
在 X 上打开原推

OPUS 5.5 IS BEING PUT TO THE TEST AGAINST GPT 6 ASTRA. BUT THE REAL QUESTION ISN’T WHICH VIDEO LOOKS BETTER. IT’S HOW CLOSE AI IS GETTING TO CREATING REALITY. Look at what these models are trying to generate. Animals moving through natural environments. A bird flying over the ocean. Sharks moving underwater. Entire scenes with motion, lighting and details that need to stay consistent from one moment to the next. And this is where I think AI video becomes fascinating. Creating one realistic image is already impressive. But video is a completely different challenge. The model has to keep the subject consistent. It has to make movement feel natural. It has to understand how objects interact with their environment. And it has to do all of that across hundreds of frames without the scene falling apart. That is much harder than generating a beautiful picture. I think the biggest breakthrough will come when we stop judging AI video by individual frames and start judging it by how believable the entire scene feels. Because realism is not just about what we see. It is about whether everything moves the way we expect it to. We are moving toward a point where creating a scene may no longer require cameras, locations, actors or expensive production equipment. And that could change much more than filmmaking. Advertising, education, gaming and online content could all be transformed. The most interesting question is no longer whether AI can generate a realistic frame. It is how long it will take before we can generate an entire world that feels real.

这条是引用着另一条写的:x.com/skaly__bull/status/2109615299494289830

x.com/tolik12308/status/2110126236781695385采集于 2026.10.14 11:12