Groundbreaking AI Uses You Can’t Miss

Artificial intelligence keeps moving at the speed of light and therefore also the applications of this technology keep changing and evolving. And that's why today I will be giving you a grand overview of everything that came out throughout the last week that you can start using today. Now, I want to be very clear, we're not going to be covering all the most recent AI news here. We're going to be talking about tools that came out that you can put to work today or new ways to use existing applications. Moving forward, you can expect this update once a week. And with that being said, let's dive right into it because the last week was very generous. And in terms of what we can use AI for today. And we're going to start out by talking about probably the craziest use of AI that I've seen recently, and that is this new GitHub repository. And before I lose you, no code required, I'm going to show you app how to do it in your browser. Okay, but basically what this repo does is it takes images of humans and it animates them. And look, this is a little scary, but I think you should know about it. Okay, so these are fully motion controllable, deep fakes. And as you can see in this little demo, you can take any character and make it do all these tick tock dances and so much more. The thing about this is that it caught people off guard and everybody's stunned by how good this actually is. Look at that. Now this has actually been released by China's Alibaba Group. And look at that. This is seriously impressive. But in the spirit of this new weekly show, let's get into using this because I wouldn't be showing you if we couldn't. Right, so if you head on over to this hugging face space, you can find a link in the description. So what I'll be doing here is I'll drag and drop this image of James Bond. And in my testing I found drag and drop in to work actually better. Just wait for this orange outline here, then you can let it go. And then you can use one of these motion sequences that down here. So I'll pair it with the running. Oh, I'll have to drag and drop again, no problem. And now you can just hit animate. And here's the thing, this thing is so overloaded that four or five times I tried this, it didn't work. But with a little bit of patience and just keeping this tab open in the background, I could make it work. So you could see the result in this video Here's a little hint. Avoid the U.S. waking times. And here you go. That's the result again. Just remember, this is the worst this tool will ever be. And some of these others, especially these TikTok dances, are looking really solid, especially considering that this just came out and it's the first version of it, right? Think iPhone One versus 15 Pro Max. Next up, I want to show you a brand new ChatGPT plugin that released and this is only for ChatGPT plus users. Also, I should note this is absolutely not sponsored, but I've been a user of Whimsical since a while. I really like them for various diagrams. They have a free plan that just works for everything I do. And now they also introduced a ChatGPT plugin. So all you need to do is you go over to ChatGPT plugins and in the plugin store you can search for Whimsical. And look, this is just my personal opinion, but this is the best way to generate various diagrams and visualizations, including flowcharts, mind maps, and so much more inside of ChatGPT. So as I have this installed, I'll make sure it's checked here. Perfect. And I'll just give it some random prompt like visualize an org chat of me, igor as the CEO and 15 cats as my employees. And Whimsical is going to go in and without you having to log in or anything, it's just going to get this done. And I know there's other plugins that do this, it's just my personal preference always was Whimsical. That's why I didn't use the other plugins. And that's now it's here. So I would just say give it a try. Isn't this perfect? Cat Employee 1 to 15. Again, it can do many different things. You can explore the website to see what's possible. And as you can see right here, all of this is editable. Very nice. And as I mentioned, my favorite thing here is that this is completely free at the time of this recording. Perfect. So moving on into the world of AI art, this came out recently and anybody who has not given this a try yet needs to go in and test this because it's so much fun. Up until now, when you created AI art, it took a while, right? You wrote your prompt, you hit enter, and then 15 to 16, 60 seconds later you got a result. Well, with stable diffusion, Excel's new model, the Turbo model, this changed in a big way because if you head on over to their tool Clipdraw Co Stable diffusion, Turbo. You're gonna find that once you log in, I usually just use Google for all of these because it's so simple. You're gonna be able to start typing and as you type, it develops. Check this out. Now let me tell you, this completely changes the user experience because I can iterate in almost real time. And if something doesn't work, you could just play with it and it updates instantly. So if I just add something like an anime cut, it should change it right away. There you go. So I think you get the point. But before we move on to the next tool, I want to talk about this briefly because this really changes the implications of AI art. As we're able to do this instantly, it will open up a whole new set of applications where you need the result instantly. First thing that comes to mind for me is the retail experience in clothing stores, right? Very soon they might just have their own custom models. And as you're kind of sitting there describing what you would like to try out, the speech recognition could transcribe it and a model like this could visualize it right away on their very own model. Plus a million other use cases where instant feedback is important. Because think about this little example that I just outlined. This way wouldn't work if there's a 45 second delay between you speaking something out and then an image actually appearing. That would ruin the retail experience. So this is actually a massive step for society, even though you might not be using this on a daily basis. Just something to think about. All right, moving on. In the theme of AI Arts, there is actually this incredibly entertaining app by Leonardo AI. So you can just sign up and start using this for free. Then they have paid plans, but I just went in, logged in with my Google account and went to app Leonardo AI Live Canvas. And what this allows you to do is you can just start drawing and it updates as you go through this. Check it out. So what I'll do here is I'll maybe draw a tree. And before I proceed, I'm going to change this to environment. And as I draw my little tree over here, I'll switch on over to blue, increase the brush size, paint a little lake here. How about this? Let's do like a mountain in the background, maybe add some snowy peaks to that. And then also in gray, I'll add a kind of background here. And let's see if we can put a sun up in the sky. There you go. As you can see, this is interpreting it completely differently from what I'm intending. But let's see if we can get it there. Okay. This is getting closer and closer. Maybe another tree is what we need. Right? So this is our tree over here. Don't judge me on my drawing skills. That's not why you're here, right? Are you sure about that? And I just realized that the tree actually looks more like this. It's not all green. Right. So how about that? Maybe a little, little more gray to delineate this. Okay. Okay. Not bad. So it got the water right. It didn't pick up on the trees, but that might be due to my drawing. And it went crazy on the hills. I actually kind of like that. And then I could just switch this out to a painting. Alright. Oh, that's weird. Then what about photography? So it seems like environment works the best, but maybe I could just make it a little less creative. And there you go. Here we really get closer to what we imagined. Right? And if we go fully creative. Wow. That's wild. Very nice. So is this incredible? This took me a minute to create and you could do that yourself. This is one of the funnest tools to show your friends when they ask you. So, Igor, what's up with all the AI stuff you're dealing with? Is there anything that is actually interesting? If you whip out this one on them, you might just win them over and convince them that playing with these tools is not just a smart move for your very own future, but also a lot of fun. And again, this is freely available right now, so go ahead and try it out. Okay, so next up we have Meta's brand new AI language translation models. And if you've been following this closely, you'll know they're releasing a lot on this front. And this one they made actually available, so you can use it to today just like everything else in this video. So there's various links down here to try these out, but I found that the hugging face one works particularly well. So once this loads, you can drop any audio in here or I'll just use my microphone to record this. In Heuten, ich ha feinen gefeldiss video. It's just German for have a nice day and I hope you're enjoying this video. And we're gonna go source language German and target language English. Hit translate. Voila. Five seconds later. Let's check this out.

Good morning. How are you today? I hope you like this video.

Not bad, right? So you could just do this with any file. And again, this is free. Isn't this just insane? I mean, what kind of age do we Live in that. This is freely available. Now, again, this is just a preview, but you can use it. So go ahead and do that. As you can see, there's a variety of languages in here, and if you care about the technical details of how it works, you can find that in this article that they published. Well, Meta has two more announcements that just came out and I need you to know about both of these, because these are not just fun tools for content creators. These might actually change the world. So look at that. The first one is a translation that follows your speech style. That is surprisingly good. Now, I've seen others, but Meta's execution in the AI space has been excellent. One could say a lot about the company, but what they're doing on the AI research front is incredible. Let's have a look at some examples here.

So glad you are here.

I am so happy to see you. Fantastic results. And what this is doing is similar to what 11 Labs launched a few weeks ago, where the AI would then mimic your intonation. This is really interesting. I absolutely love it.

This is really interesting. I absolutely love it.

This was a real unlock because this allowed you to fix some of the mistakes that the voice synthesizer made. By default, you could just override that with your very own tone and effectively let it re speak the word in a way that might sound more natural. Here it does it automatically, no two steps required. So they have a little demo of this. So you can just click demo. And let's just say this. Let me start in German and output it in English. Gives me some examples here, but I'll just freestyle this. Let's explore the beautiful language that is German here. And in this case we'll do like a grumpy German tone. Wage Strasse. Oh, God, that language is just the best, isn't it? So that should have said, what is this ambulance full of butterflies doing on the main street?

What's that ambulance full of butterflies doing on the highway?

So that's a non expressive. That's not great. Right. But this expressive translation. What's that ambulance full of butterflies doing on the high? So I don't think it captured the anger quite well. Right, so let's give it one more shot and this time let's throw a curved ball with a bit of a Viennese accent. I don't think it's gonna be able to do this, but let's try. Okay. So I wouldn't even know how to translate that. So let's see what it does here.

Egyptianism is typical again.

Egyptianism. Nah. Egyptianism is typical. Again, no. So that didn't work at all. So you can't do accents as expected. Not gonna hate on it for that. One more shot here. Okay.

Have a nice day, lady. What can I get you today?

Not bad. Another expressive translation. Have a nice day, lady. What can I get you today? Didn't work too well, so that was not fantastic. Right. So it's a bit hit or miss, but I really enjoy this tool. I think this built into smartphones could be game changing if it just works live as you speak. Interesting. And one last tool here yet again from Meta. And yes, this is another image generator and I would love to show it to you, but as I'm in Europe right now, it's not available, so I'll have to get a little tricky here. So what I'll do is I'll use my VPN to just get around these limitations. And honestly, this just has been the go to way to use some of these tools. And yep, as I change my location to New York, this just became available. It would be great if somebody in the comments could clarify if this is legal. I'm honestly not sure. This is just a freely available research preview. Right. So this shouldn't be a problem. And absolutely everybody's doing this. I don't know though. Leave a comment below if you do so, yes, this is another image generator, but one thing is exceptional about this one and that's the fact that this is actually built into or soon will be available to billions of users between Facebook, Instagram and WhatsApp. This is the image generator that most of humanity will experience first. So let's see what happens when we use the legendary prompt cat with a hat. Yep, I can live with this. This should make for a good experience for most humans on planet Earth. Oh, look at this one. I approve. Please, Meta, roll this out to everybody as soon as possible. All right, next up we're going to talk about something very exciting because this, this could bring some of your old photos or childhood memories back to life. So as you can see in some of these examples, it's not just enhancing the image by adding sharpness or increasing the pixel density. No, no, no. This is actually using an AI image generator to enhance the original. So for example, if you keep your eye on this little part of this futuristic motor, trike, whatever this is, you will see that it adds brand new components. But they perfectly fit the original. But they're new. Right? So if you're looking to maintain 100% of the detail in the original, this is not Going to be your tool of choice. But in most cases you might want to add new features like this. Like let's take a hyper realistic photo like this one and look at that. I really don't mind this enhancements. So I think one could debate the ethics of this. Just like one could debate the ethics of using Photoshop on images. But I personally am very happy about the fact that we have a powerful tool like this. Now look at this waterfall image. This is absolutely incredible. Okay, but there's a downside to this because the two main services that offer this right now are both paid. Okay, so both Magnific and Crea AI ask for a payment. So the free version you can get on the wait list and when they pick you can get free access. And for $30 a month you'll be able to use this right away. So in this case we'll just use Magnific because this is actually a creator made tool. Javi Lopez on Twitter made this. And there we go. For $39 we have access to this. So long term fans of the channel can at this point have a guess at what I will upscale here. Little tip, it's my all time favorite game. Absolute timeless gem, Gothic one. And if you didn't know, which is very probable as this game was mostly popular in Germany and Poland 20 years ago. This is the main character, the nameless hero. And as you can see, the graphics of this games are not exactly at a 2023 level. Meanwhile, everything else about the game, including music and world building is better than anything I've seen ever since. Anyway, enough nerding out over a 20 year old game here. Let's drag and drop this image in there. Let's say two times upscaled. We're going to optimize for video game assets. Perfect. And I'll just use the default settings here and say upscale. Let's see what we get. Okay, so initially we have a 15 minute upstart time and then every image is going to take 60 to 90 seconds. And that turned out to be a lie because in real time this took less than 15 seconds. And here we go. This is the upscaled image. Before, after, before, after. This is absolutely incredible. So let's just have a quick look at these settings, maybe make it a little less creative and turn down the hdr. And yeah, I really want the resemblance to be high. So let's upscale this once more and see what we get. But this is already really impressive. I mean just look at his vest. Wow. I wonder how long it will Take until we will be able to upscale 30 images per second and keep object and character consistency in video. Then we'll just be able to remaster all the old games graphics while to think about that. Either way, here's the second result. Yeah, I actually like this a little more. It resembles the original. And look at that. It just works. Okay, wait, this is too much fun. I'll just drag and drop this other Dall? E image that I generated into here and now. Let's go a little crazy with it. Okay, let's say 3, 3D render. Let's say hyper realistic, godlike entity, Mother Nature. And we'll make this very creative turn up. HDR all the way. And the resemblance to -5. All right. Super curious to see what happens with this one. There we go. Whoa, look at that. So it seems like the prompt didn't have too much of effect, but I absolutely love this. It turned this Dall? E image into something that you could put on a billboard, I guess. I mean, look at the level of detail before. It looks like there's a layer of fog in front of it. And that's the case with a lot of Dali images. Right. But in many other aspects, it really excels. So with this workflow of Magnific plus Dall? E, you could achieve hyper realistic, high resolution pictures. I mean, look what it did to the lip and the chin. This is absolutely insane. Kind of scary, to be honest. But I love how it maintained the line and actually made it even sharper. Wow. I'm seriously impressed. I'm saving this one. Let me run these settings on one more dial image just to confirm. Yep, same thing. Incredible. This really brings it to the next level. But in this case, it makes the human a little too realistic. Right. I like what it did with the cyborg on the left, so yeah, maybe I would Photoshop those two together. Either way, extremely powerful tool. Again, Kriya is the cheaper alternative. And with that being said, let's move on to improvements in one of my favorite current tools, Runway ML. And what they do really well is they turn images to short video clips. As you can see in this launch trailer, it adds a little bit of slight variation, but up until now, we didn't have control of this. Okay. It was just a bit random. And recently they added camera moves. So let's just add the image we upscaled a few seconds ago. And as mentioned before, you only had this one slider here. Okay. It was general motion and you could pick if you wanted a lot or little anything beyond five Usually didn't work too well unless you had something very abstract or psychedelic that works. But with realistic subjects, you wanted to stick somewhere around 2 to 3 for best results. But now we get more granular control. And these are camera motion settings. And I briefly want to give you an overview because. Because there's really three categories here. They don't say this. So the first two sliders here, horizontal or vertical, are when you move the camera horizontally or vertically, which is very different from pan and tilt. Because pan and tilt assumes that the camera is fixed on a tripod. And then panning is this horizontal axis and tilting is the vertical axis again, but you're not moving the camera. It's static. Okay. And then the third category here, roll and zoom, are all about specialty effects. So roll would be a movement like this where you're kind of messing up the horizon. This is a special specialty effect that they use in music videos or movies like Inception when they want to portray surreal effects. In practice, you won't be using this too much. And last but not least, the most intuitive one, zoom is simply when you zoom in on the image or zoom out. I think everybody should get that one. So we get to control all this now. So let's say we add a little bit of zoom onto this image, and then I'm going to do a little bit of horizontal movement. But here's a little trick. Whenever you're doing these movements with an actual camera, and let's say you're doing a product shoot, there's one technique that is kind of a best practice for doing these shots, and that is when you're moving the camera around something, you want to keep the object in the same place, not to make the movement too dramatic. Otherwise, the background and the subject is moving, and it just becomes a little confusing. So with product shots and humans, too, it just works really well when the camera is moving around something, but you keep the subject in one spot. And that's why they actually added this little crosshair here on the right. See that? So when I'll be panning around the scene like so, I'll make sure not to go too crazy on the horizontal. Otherwise the background and the subject will be shifting all over the place, and we will get wonky movements. Plus, the same thing as with the motion slider before goes here. Anything beyond a three, maybe a four, is really pushing it. So I'll make sure to make these movements rather subtle in order to get quality results. So I actually experimented a bunch with this. Anything beyond four really messes it up regularly, with the exception of the zoom. On the zoom, you can actually go a little harder because all that is is a slight scale up and then it has to move the proportions of the objects a little bit. That it can handle surprisingly well. So let's just say zoom to 10 and let's line up these crosshairs a little bit. And one last little tip is when you double click these sliders, they're reset to zero. Plus you could be using this to go incrementally and then double click to reset. Nice. So I'll just say save. And you could add the motion brush here too. And in the Runway Academy, and this is where I stop exploring this. Otherwise it's going to turn into a dedicated Runway tutorial. But here I actually have an article about motion brush and camera control and so much more. So if you want to dive deeper, this is the best place to do it. And here they recommend you start with the motion brush, not with the the camera motion. So we're just going to leave the motion brush alone in this case and say generate while it does that. I should note that a few generations are free here and afterwards you're going to be paying $15 per month as of now. But again, at the time of this recording, they do have a free plan, so you can try it out. And there we go, the result. Look at that. I don't even think it's that bad, but as you can clearly see, we went overboard on the zoom here. So actually what I said didn't hold up with this example. So let's tone it back. Let's go back to zoom of 4. Say save and generate one more time and see, see what we get. But honestly, the first two seconds are kind of cool here. Okay, so this is the second generation with less zoom, and I actually kind of like the first one more. I think the first two seconds here are actually usable in some music video or something like that maybe. Either way, it's really great to have these more granular controls as you now have six sliders instead of one. Alright, so very briefly, you might have seen this already. I covered this in another video. This is Stabiliti's AI Stable video diffusion. You drop an image in here and it generates a video from it. This is very limited. It's only two seconds long, but completely free. So same thing here. You just drag and drop an image into here and here you get it to move. Now you might be already picking up on the fact that we could combo some of these tools, right? So we could generate something with Stable Diffusion Turbo in milliseconds and then we upscale it with Magnific and use something like One Way or Stable Diffusion Video to make it move. Guys, this is getting insane. You can just stack some of these tools on top of each other and achieve things that were impossible a few weeks ago. And with all that being said, these were all the tools that were released recently. There was a lot to cover here, and you can look forward to another video just like this in seven days. Subscribe and hit the notification bell not to miss that. And if some of the video tools sparked your interest here, I created a video dedicated to AI video beginners covering the entire landscape and how to get started with some of the most powerful tools that are available now. I'll see you there.