OpenAI Video Generator Just Shocked EVERYONE(Sora)

OpenAI just released something that I can't wrap my head around. And I spent a bunch of time talking to people and across the Internet. These are all my thoughts, okay? This is literally the most revolutionary release in AI since GPT4, I think. It's not really a release, right? It's a research preview. But this changes more than we could cover in a 10 hour video. I'm going to do my best to give you the perspective of somebody who spent the last decade working in video and film production. Five YouTube channels. I've worked with corporations and clients of most sizes. And this changes everything in video. OpenAI's video model Sora. Here we go. So look, this thing is just too good to be true. So let's just start with what they showed off and what this is, okay? It's a new foundational model by OpenAI, the creator of ChatGPT. And if up until now, AI video was like a 2 out of 10 and know Hollywood production quality video is like a 10 out of 10, this is like a 8, okay? And I don't say that lightly. Now you could argue that it's seven or something like that, but are you serious? Look at some of these examples. These drone shots are flawless. These are commercial grade shots where even the hands look all right, like I have a hard time talking. I don't understand how we could have made such a big leap in such a short amount of time. How did we go from this about a year ago to this today? Look at the hands. They're not terrible, they're not perfect. But this is one that has been generated by Sam Altman on Twitter here. Okay? People are giving him prompts and he's just running it for the model. Now look, this is not something that's available today. This is a research preview of our foundational model. But we need to talk about this. So first things first. If you go to their website, you're going to find that this is not under their blog. This is not a new ChatGPT feature. This is not a new upgrade to what they have. This is a brand new thing. Okay? If you go to research, you can see Sora is the new thing here. So what does it do? Well, it generates video. And not just any video, it smashes all the limitations that were present up until now. Okay, so what were the limitations? I mean, let's list them briefly. First of all, you could only do short clips, right? One minute long clips can be generated with this. Secondly, you couldn't really do realism, especially of humans. I mean, that's kind of over. Look at this trailer. Thirdly, you could only do minimal camera movements. These previews are incredible. Okay, so I'm gonna try to tone down my excitement, but I just can't contain myself. This literally changes the entire video production, film industry. Many companies like Netflix and Apple, and this just literally changes everything. I think a lot of people are gonna have these clickbaity YouTube titles, but in this case, you can't make it clickbait enough. This is insane. This does change everything. This is mind blowing. Let's look at some of these examples and what this actually means and what we will be able to do with this. Because the clips by themselves are impressive. But I don't know if you're already seeing the bigger picture here that I have in my head, but this is going to change industries. This is going to change the world, because this is a brand new category of product. Okay? This is not a 3D engine where you have things and you can recreate materials and you have things like ray tracing where the light bounces around. They put this also. Well, at the bottom of the article they say SORA serves as a foundation for models that can understand and simulate the real world. A capability we believe will be an important milestone for achieving AGI. This simulates the real world, so you're going to be able to simulate animation, movies, but also documentaries about nature and scenarios that you maybe couldn't afford. So let's start talking about some of the industries and areas that I see this actually disrupting. Because something like stock video is kind of over, right? And I say this as somebody who's been selling stock video footage since a decade now, but why should anybody be buying footage of a cat walking through a forest if you can just generate it and make it exactly the way you want? And my guess is the cost of this will be very, very low. One stock footage clip of Santorini like this would cost you about 50, maybe $100. Maybe on some of the cheaper sites, 10, $20. If you have a subscription, it might cost you $50 or $100 a month to get these. Well, we don't have any details on the pricing yet, but I would bet that this shot is not going to cost us $50, right? Maybe the only comparison here is the other video generators that are not even close. I don't even want to compare them. It would be unfair and just unproductive. I mean, what's the point? But in their case, you pay like 10, 20, $30 for a subscription where you can generate hundreds of Clips, right? Or what about this one? And there's somebody who's actually documented constructions with time lapses. I can tell you this is incredibly hard to achieve. Now in this case it's a drone shot, but you could easily also do a time lapse. And let me tell you, I worked on one project for Coca Cola hbc, their factory in Central Europe, and I created a three month long time lapse of their rooftop, okay? They put solar panels all across the biggest factory building. And I had two cameras set up over three months taking pictures every five minutes. That was the highest effort video I've ever created. That one clip of the solar panels being laid out took me months to produce. And in the middle of it, actually when they were moving something, the construction, construction workers unplugged one of the plugs and replugged it, which made my camera shut off. And then I was missing a week of footage. So all of a sudden there's like a jump in the time lapse here. You're just going to be able to prompt this out of nothing because you have an engine to simulate reality and then you can do crazy things like this, right? But I think the crazy edge cases are not a revolutionary thing here. I think things like this, this looks like a scene out of Harry Potter, right? This is what changes the game because with language models you can generate scripts and with this you'll be able to turn them into reality. Now, text is a powerful thing, but video is a whole different deal. There's probably a reason why over time all social media channels trend towards video, right? Instagram, originally about pictures, is now all about reels. TikTok has been the biggest rising star among social media platforms over the last few years. YouTube goes without saying. And if you go on Facebook, most of the feed is short videos. It's just the richest media format we have. And as you know, the other component to video, audio, has been kind of solved already. If you look at tools like Suno, you can create excellent quality music. And if you look at human voices, well, those are virtually indistinguishable at this point. And with this shot, would you be able to tell that this is AI generated? Now look, there's going to be some weirdness like this, obviously, but look at this. Why would anybody buy stock footage? Everything except of maybe the hands here, in some select cases, when you watch them really closely, is, I would argue, perfect. The lighting on her, the details, the expressions, it's all consistent even though the camera moves. And again, this is one of the weaker ones Here. And yes, these are just research previews, but the fact is there's so many of them. And Sam Altman actually going ahead and like creating these on the fly as people give him different prompts on Twitter just solidifies this. Look at that. Two golden retrievers podcasting on top of a mountain. Yeah. Could you tell this is AI maybe by the little artifact on the shoulder here? But guess what? If it's in a music video or in a movie where things are fast paced in the middle of an action scene at this quality level, there's no way you're gonna notice. And sci fi films, gosh. I mean, look, one of the first things I did was actually pull up a list of all movie genres and you can do this just by googling movie genres and then tell me which one of these is not gonna be affected by this technology at this stage. And also remember, this is the worst it's ever gonna be. Right? Action movies like these fast paced action scenes where you have cuts at this pace, gosh, of course they're gonna use this science fiction. That might be the most obvious one. You can create things that were only possible with animation or extremely high budgets up until now on a shoestring budget. This might just do to movies what digital cameras did to video production. Twenty years ago, you needed a hundred thousand dollar camera to produce a commercial. Now people shoot them on these and one can't even tell, right? So everything here, I mean horror, by the very nature, it just doesn't reveal a lot and it's often grainy. It's gonna be amazing. For that we could go through all of these. You can make up your own mind. But with things like fantasy, where you have a little bit of leeway in terms of realism and or westerns, where you can have that slight, slightly grainy style, you're gonna be able to generate movies with this. I'm not kidding. I don't think I'm exaggerating here. Now look, I'm open to all types of comments and criticism in the comment section and I think there's still a lot of value to film. No doubt. But this does change the game. There's no denying that for documentary movies, you're gonna be able to create images like this or this one. This is indistinguishable, right? What am I missing here? My brain is telling me I'm missing something here. So is this gonna completely demolish the stock footage industry and all the people that are making money through that right now? Yeah, absolutely. I don't think that's much of a discussion. If it's this quality level, is it going to affect the film industry? Absolutely. Now, it's not going to replace it. Right. But all of a sudden you might be able to produce movies that would have cost 100, 150 million for hundreds of thousands of dollars maybe. I mean, you're not going to be able to skip the scripting and the editing and the post production and all that, including the distribution of it, obviously. But the production part, the part where you go out there and you turn on a camera and you simulate scene so that the camera sees them a certain way. Well, what I see here is not a video generator. It's a reality simulation engine that is going to be available to everybody. And you could create animations, B roll drone shots at a level of quality that will be indistinguishable to most people. And examples like this really show off how good the physics simulation here is. By it being trained on a bunch of footage of water where the physics are playing out in real time, it's able to recreate it at a level that we just haven't seen before. I think there's so much to talk about here, even though we don't have access to this as of today. But first we need to kind of accept that we're in a new reality. A reality that no human ever had the privilege of experiencing. One where you can recreate your thoughts, your dreams, anything that you can imagine with a few words and the click of a button. And we talked about film briefly, but I don't think it just stops there. The entire video production industry, where it comes to commercials or the animation industry, where people create things like this with. With very high effort and very high budgets, it's just going to change it all now. There are definitely some pieces that will be missing once this comes out. We don't know when it's exactly coming out. They just state in this article that currently they're testing this with their Red Team, which is responsible for making this safe. And I think they're right to do so. Let's be real, this is a bigger deal than text generation. In many ways this is going to democratize the production of films, but it's also going to give people without a budget the ability to manipulate reality and create. Create scenes that would have never occurred. And so this video doesn't just become a hype fest. I'm going to add a few comments of things that this obviously can't do because they're avoiding them in the examples. One of them would be text. There's not a single one where text is involved, so they're obviously avoiding that for a reason. Another one would be character consistency. So you might be able to generate this minute long shot of a woman walking through a city that looks like a commercial. But how good is the character consistency on it? Is this face, body posture and outfit consistency consistent with what's happening back here? Well, I actually don't think so. If you look very closely, you can see that on the glasses you don't have the logo here on the side. Whereas if I go to the end, it's clearly there. What about her blemishes? This looks pretty spotless, whereas here in the back there is some. But again, this is the first preview of an actually capable model that can actually be used in something like movie production. Consider this is the worst it will ever be. And also consider that not 100% of a commercial or a movie has to be generated with this. It could create a few specialty shots that would have been really expensive to get. You're going to be able to give it a picture and say, okay, now create a 365 day time lapse of this that concretely I also did. And I had to set up a camera for 10 days straight four times in a year in the exact same spot. And then I had to process around 40,000 images. You're going to be able to type a few words and hit Enter. So yeah, this is a real game changer if I've ever seen one. I can't wait to get my hands on this. If you want more details, you can look at the blog post and all the prompts that are involved here. I'm going to be covering this more because obviously with my video production background and my passion for AI and technology, this is exactly at the sweet spot where I want to be spending my time. So subscribe for more content on this and leave your thoughts below. What is this going to change? How is this going to affect you concretely? I would love to hear it. All right, I'll see you in the next one.