NEW AI Video Tool Claims To Be As Good As Sora!

While we're still waiting for Sora, we got another generative AI video platform. This new one is called Hotshot and you can find it over at Hotshot Co. In fact, this one is so close to Sora that their press release was literally titled How a four person team Built Sora. They put together a pretty in depth article here going into how this model was built, explaining that their very first model, Hotshot XL, generated one second clips at eight frames, frames per second. And that was only 13 months ago. And now they're getting video quality like this. And here's the coolest thing. Anybody can try this out right now. However, before we test this out, let's take a peek at some of the demo videos that they've put out already showing what it's capable of. Starting with the one that we can see on the screen right now, which used the prompt. A woman with braids peeking above water. A man with an afro sitting on a cloud. Aerial view of a couple on a picnic blanket. A woman looking worried in front of a refrigerator. We could see on their website they have some categories where we can select different videos and take a peek. So let's look at some others here. A kid with a determined expression. A person looking concerned. A person smiling. Now, as far as text to video generation goes, this is some of the most realistic video I've seen. I've seen better image to video generation, but for text to video, this is pretty dang good. Here's a vintage camera. A time lapse of a hand painting a bonsai tree. A barista creates latte art. Snow capped mountain peaks reflect reflected in a lake erupting geyser in a geothermal area. A fly through between mountains in the ocean. Dense rainforest canopy with sunbeams. A blonde woman on her phone at the subway. A woman looking out a train window at the mountains. A flight attendant walking down an airplane aisle. Here's a tsunami coming through an alley. I mean, that looks pretty dang realistic. A ball of flames in a market. Wisps of smoke. Come on, that looks pretty cool. A news anchor talking directly at the camera. A focused man with a headset making conversation that is ultra, ultra realistic. There might be a little funkiness in the teeth and maybe the microphone looks a little off, but other than that, the person himself looks really good in my opinion. A woman looking scared at the dentist office. It's cool because you can just see the expression in the eyes without even seeing her mouth. Corgi sitting in a school classroom. Anthropomorphic cat taking a Selfie in Paris. I mean, the cat kind of has like three human looking fingers, but hey, that's a high video for you. Golden retriever puppies cooking in the kitchen. There's some pretty good golden retriever puppies. Honestly, a bear in a grocery store holding up a very good sign. Yeah, as soon as they start adding text and signs in there, it definitely looks a little bit janky again. Sunny side up, eggs, sourdough, toast and tomatoes. A jug pouring milk into a glass. Anthropomorphic cat as a butcher cutting salmon. Hey, this time it didn't try to give it human hands. I mean, it doesn't look like cat paws, E but it didn't give it human hands at least. And those are all the examples they have on their website. Probably fairly cherry picked. I mean, that's kind of what they do. But they do give us the option to make your own. So let's go ahead and test it real quick ourselves. I'll click make your own. And here's the deal. Right now it's going to let you generate two images per day for free. So you can see up here it says due to high demand, we're temporarily limiting free tiers to two generations per day. Paid tiers have no wait times or daily limits. I haven't used either of my two for today. I wanted to wait till I record this video. So you're going to see what I generate with me. I'm going to turn to my old buddy Claude here and have it suggest some ideas for videos for me. I'm going to give it the prompt, give me a couple interesting prompts to test a text to video model. Each prompt should be less than 10 words and not too complex. And the only reason I put that was because when I was showing off the demo prompts that they showed us, none of them were super long, super detailed prompts. They were all, you know, 10 words or less. Let's see what Claude recommends. Number one is a cat riding a skateboard through a neon city. I'm actually going to tweak that a little bit. If you've watched my past AI videos, I always like to do a monkey on roller skates. So let's do a monkey on roller skates in a neon city. I'll keep the neon city from Claude suggestion just so Claude doesn't get offended. And let's generate and see how it does. I have no idea how long these generations take, but when this is done, I'll let you know how long it took. So I'm actually starting to get some static now. So I'm not sure if that's just like a static video, it's playing in the background, or if I'm actually seeing the diffusion process in real time. All right, so that took about a minute and a half. And here is the Monkey on Roller Skates video that we got. Not great. The arms disappear a little bit and it looks like it's more of a skateboard looking thing than roller skates. They look connected. The output's not very good. One thing that was very cool though, was the process as it was making this video. It was showing the diffusion happening. So you saw those grainy pixels all start to sort of form into a video until we saw the final output, which is the video that you're seeing on the screen right now. So now I only have one prompt left to try. So let's go ahead and use Claude's other suggestion. Blooming flowers emerging from melting snow in time lapse. Let's go ahead and submit that prompt and see what it gives us. I'm going to speed up the video this time so you can actually see it diffuse into the video that we're going for because it does look really cool to see it actually start to form the video for you. And here's our video. This one's actually a little bit more impressive than the last one. It doesn't really look like a time lapse to me, but I mean, the colors and the flowers and the snow all look pretty dang good. Wonder what would happen if I clicked Regenerate. Will it regenerate for me even though I'm out of credits? Nope. It wants me to upgrade. So it looks like typically you can generate up to 200 generations per month for 99 bucks a month. Or you can do up to 60 generations per month, but only two per day. I'm going to take one for the team here and upgrade so I can generate some more videos and see what this is capable of. But with the free model, hopefully you get what you're looking for within two generations. I'd say you have to be pretty dang lucky for that to happen, though. I had Claude generate a handful more ideas. Let's go ahead and see what these look like. Here's a chameleon changing colors on a rainbow background. I see the chameleon. I don't really feel like it's changing colors though. Here's a sand castle rising and crumbling with the incoming tide. I'm not really seeing the rising and crumbling, but I'm seeing the sandcastle and the Tide. This one's a paper airplane soaring through diverse landscapes. Yeah, this one's just cursed. Sorry. This one is campfire sparks swirling into constellations at night. Okay, so this one is robot learning to paint a masterpiece, which honestly, is probably the best one I've managed to generate so far. And that's not saying a whole lot, if I'm being honest. I haven't really been impressed with any of the generations that I've generated. The stuff that they had on their homepage, their examples clearly cherry picked because I haven't gotten anything close to that. However, most of the stuff that I was most impressed by on the homepage actually had people in them. It seemed like it did really good with faces and realistic people. So I'm going to try a couple more prompts. This time I'm going to be sure to include people in the prompts and see how those do. All right, so this one's a ballerina pirouetting in a field of floating bubbles. And, well, yeah, her leg disappears. She has three arms. At one point, both of her arms go to one side of her body. At another point. Yeah, not great. Let's try another person. And now we have a chef juggling vegetables in a busy kitchen. Now, the guy himself actually looks pretty decent, but obviously you can see the vegetables just look like they're floating around and disappearing and acting all funky. It's not a bad generation, honestly. I actually think the guy looks good. But, yeah, you can see what's going on with his hands. Doesn't look that great. And here's the last one that I'm going to show off is an astronaut planting a flag on a purple alien planet. The planet and the astronaut don't look too bad. The flag is just kind of floating there. There's nothing that says that he's planting the flag. Now, the day I'm recording this is actually the day that this hotshot tool was released to the public. And since it came out, I've been seeing some generations on X that are actually pretty decent. Better than any of the results I've managed to get. Here's one from Heather Cooper. A living tattoo on a person's skin that glitches and animates telling stories through shifting, pixelated imagery that spreads, retracts across the back. Cinematic realism, dramatic lighting. Maybe I just need more detail. I don't know. Here's one from Akash here. That's like a giraffe walking through a mall with a cell phone or something. That looks pretty dang Good. Keep in mind that Akash here is somebody that actually works at Hotshot. So they have the ability to kind of generate as many as they want and cherry pick the best ones, if we're being honest here. Here's one from Caitlin Holloway and you can see she put a pretty basic prompt, Chickens walking under the Eiffel Tower. And that one looks pretty dang solid. Here's another one from Akash here that says Hotshot can do styles too, like comic book styles, but. But they didn't share their prompt. So I don't know what prompt was used to get this comic book style. Here's another one of like a Mario ish character walking around. That actually looks pretty good. I think it's pretty funny. I can see that as being like a little meme in a video or something. Here's one from Dream Starter of somebody walking through a field of wheat or something like that. And finally, here's another one from Akash here of somebody putting out a fire in a kitchen. That doesn't look too bad either. Again, when I look at the prompts that they demo on their homepage, they're all pretty dang basic short prompts to get really good videos. But whenever I tried these types of similar prompts, I got videos nowhere near this. So either A, they're really cherry picking the videos or B, I haven't figured out the best way to prompt this to get the best results yet. So if somebody's watching this video and they have a tip or a trick to get the best results on this style of video, please leave it in the comments. Help me out. Help other viewers of this video out. Let's figure out together how to actually get the best possible videos out of this program. Because so far none of the videos I made came anywhere close to the quality that I'm seeing on the homepage. Which is kind of a bummer when you only get two a day. And if both of those kind of stink, you gotta wait till tomorrow to try again or pay a hundred bucks. Neither are ideal for most people. All right, before I wrap this video up, there's a couple little things I want to point out. This hot shot tool can generate up to 10 seconds of footage. Although I don't think any of the videos I generated were actually 10 seconds, most of them seem to be about 5ish seconds. It generates them at 720p. But the most interesting thing to me that they say in this article is that in the next 12 months, entire YouTube videos will be AI generated by creators control over every aspect of generations. Longer durations, higher resolutions, real time interactivity and more modalities like audio are just around the corner. The videos they showed were amazing. The videos I tried to make, not quite as amazing where they're headed. Really, really exciting. So that's Hotshot. It's over at Hotshot Co, not a sponsored video. They don't know I'm making this video. I came across this today. I saw the news release and went let's go test it. I'm sure a lot of people don't want to go spend the 100 bucks a month. I'll go spend the hundred bucks once and see what I can get out of it and show off what people can expect because I'm not going to cherry pick. I'm going to show you exactly what I get out of it. So hopefully you found this helpful. That's Hotshot. As always, if you want to discover Even more cool AI tools, check out FutureTools IO. This is where I curate all the coolest AI tools that I come across. I've also got a free newsletter, so I'll send you some of the cooler tools straight to your inbox. And if you like videos like this, you want to see AI tutorials, AI news, you want to see me test out cool AI products like this video and subscribe to this channel and I'll make sure more videos like this show up in your YouTube feed. Thank you so much for watching. Thank you so much for nerding out with me. I really appreciate you. I'll see you in the next video. Bye bye.