If you’re not paying attention to AI video, today is the day to start. Hey, it’s Kyle Shannon with AI Learning Lab. This is an AI walking talk. And I’m walking around a parking lot in south Denver. The misses is at a doctor’s appointment. Is there a lot going on with video. So I’ll talk a little bit about that. There’s also a bonus if you wait till the end, there’s some new stuff with ChatGPT you need to know about. So let me give you a little history. There were kind of two major players in the text to video game. There was Runway ML and PICO Labs. And there was another one, I can’t remember the name of it, that was decent. They were all decent, but they were, you know, glitchy, weird ass videos. And then OpenAI announced Sora and Sora blew everyone’s mind. And the way they described it, it’s not just a video maker, but it’s actually something that understands and simulates reality. And that’s why the video is so good. And that was in February and we’re in June. And other than Astrian Kutcher, I don’t know that anyone has access to it. Like it’s very few pee. It’s not out. They haven’t launched it. And then 3 weeks ago, 2 weeks ago, 3 weeks ago, doesn’t matter, this Chinese version came out. But it begins with a k, but not everyone has access to that. But that looks really good. And so it just seem like, you know, people are talking about these cool video models, but they weren’t actually out. And then this last week, 3 days ago, 4 days ago, Luma Labs just drops out of nowhere. So one of the lead developers left Google who was working on Vo was another video model that was announced and he left Google working on the Vo project because he said the chances of it being released to the public, you know, anytime soon were low. So he went to Luma Labs where they could launch something in three months. After he got there, they launched this thing. So it’s called Dream Machine. And so if you go to Luma Labs dot AI, you can make videos now. They’re severely rate limiting the free tier it for a while it was anyone could do anything and then their servers got completely overloaded and then they dropped it down to 10 generations a day. And then I read this morning, five generations a day. But it’s, it’s really good. Still janky, still kind of a mess.
Then yesterday, Runway ML, one of those early players announces Gen 3. So if you haven’t seen the Gen 3 announcement, go look on Twitter for Runway ML Gen 3 and look at the examples. I think they’re better than the Sora examples. And then Luma Labs also announced that in the coming, I don’t know, weeks, they’ve got a whole bunch of new editing tools that it’s gonna get really good. And then apparently it’s just bonkers. There’s a new video model launching today. I don’t know what it is. I haven’t seen it yet. Hasn’t launched yet, but that’s the rumor is that there’s another one coming today. Plus, at some point, Sora will come out and at some point this, you know, thing will come out.
Not to be outdone, Google Deep Mind announced yesterday that they’ve got a video to audio model that is in some decently advanced form of development. It needs, they need to do red team testing on it, which is where they test for safety and efficacy and stuff. But that’s one where you can take a video clip of, say, you know, a car during doing a burnout in a parking lot. And you can just say car doing burnout in a parking lot, you know, with cinematic music behind it. And it’ll just make all the sound effects and the music for that clip and actually pay attention to what’s going on in the clip and match the audio to what’s going on. It’s bonkers. The stuff is just kind of, in the past week, it’s all just exploded. So one of the models of the AI Salon is play first. So I would strongly encourage you if you haven’t played with these video models yet, go play with them. Play with Runway ML if you can. I don’t remember if they had a free tier or not. Luma Labs dot AI definitely has a free tier. And the other stuff I don’t think you need to play with, but pay attention to Twitter, what’s going on with this. So the Gen 3 from runway should be live. They said in a few days. So I’m thinking this week, today is Tuesday. The, I don’t know, 14th or 17th or 18th, I don’t know it, the date is, it doesn’t matter. A good play. Go play. Just see what these things are capable of. Go look at the example videos, start paying attention. Alright, and just hang on once. Okay, the final thing is, if you’ve got a ChatGPT plus account, go make some images. It looks like they secretly upgraded Dali to maybe the new GPT four Omni Image model, or it’s an updated version of Dali. I don’t know what’s going on. But it looks like it just got way better at text generation and prompt coherence and things like. that so go check that out hope this helps bye。