Exploring Unknown AI Tools (Underground AI #3)

Alrighty, I think I'm live, so welcome to the live stream. This is going to be a fun one. I am going to do my Underground AI live. So if you've watched those last two Underground AI videos, what I did was basically on the Future Tools website, I get something like 100 submissions a day. On a slow day, it's like maybe 50 submissions. On a busy day, it's 100 submissions. And to this day I still click every single link that gets submitted and I go and test it myself and well, I don't test every single one. I open the webpage and make sure it's something that looks interesting, that hasn't been submitted 200 times already, that does something kind of unique over all the other tools that are already on this site. So I still click on every single link and at least look at it and in a lot of cases play around with the tools and, you know, outside of the OpenAI, the Microsoft, the anthropic, Google, all of those big companies, you don't really hear about a lot of the smaller tools very often. So I started this series, Underground AI, where I pool anywhere from seven to 10 tools that I find interesting that nobody else is talking about and I give them a go on recording. But I figured it'd be really fun to do it live that way. You know, I am not necessarily cherry picking any sort of outputs or things like that. You guys can actually give me prompts and say, hey, go and test that. Let's see here. I'm just making sure everybody can hear me real quick. All right, cool. It does say we're live. So cool. So that is the idea with this. I'm going to go ahead and do another one of those Underground AI episodes. I'm just going to do it live. So if a tool isn't working properly when I try to use it, you're going to see that if a tool takes a really long time to process and give me back my response, then you're going to see that as well. This is all live. We have no idea what's going to happen. And I've picked, I think seven or eight tools that I'm going to play around with. And if this doesn't go too long, you can even suggest some tools here inside of the chat and we can test some of those out as well. All right, I'm just going to peek at the comments real quick before we get going. Make sure there is no issues going on. I've got three moderators helping me on this live stream in the Last. The last few live streams I've done, it was either just myself or only one moderator, John, who's here, AKA Craze in the Dark. But today we got Craze in the Dark helping Moderate. We got AP helping moderate. And then we have Willow, aka. Let's see, what was your username here on YouTube? Travel. Travel something Travel well, sorry about that. So we've got AP Travel well and Crazy in the Dark here moderating. So hopefully any sort of spam or bots or just people being rude will get booted pretty quickly. They're also keeping an eye on the comments for me. So if, if there's a chat comment that I need to see, they're going to bring it to my attention. So we're going to get into some tools, play around with some stuff and then, you know, at the end do some Q and A, test some tools that maybe you guys suggest on the stream. And I'm done rambling. I'm already three minutes into this thing. Let's just get into playing around with some tools here. I'm going to go ahead and switch to my desktop now. This live stream is actually a sponsored live stream. Mine Studio, a company called uai, has a tool called Mind Studio. And they reached out and asked if they could sponsor one of the live streams and I said, absolutely. I've played with Mine Studio in the past. I've made videos about Mind Studio in the past. So I'm going to play around, demo what Mind Studio can do, and that's going to be the first tool that we're going to talk about. If Mind Studio is new to you, well then this is an underground AI tool to you as well. So let's just go ahead and get into it. So Mind Studio, what this tool does is it builds these automated workflows using AI large language models. And what's really cool about this is you can build this like visual workflow. Let me scroll down here so you can see the little animation. You can build this little, like drag and drop workflow and use multiple chatbots. So if you wanted to, you can have one step in this sequence use Claude, and then the next step in the sequence use gpt4 and then the next step in the sequence use llama2 if you wanted to, but it allows you to integrate with all sorts of different chatbots. So I'm going to go ahead and log into my Mind Studio account and we're going to build a workflow. So I can give you an idea of one way that you can use it and then I'll show some other examples, spend a few minutes talking about Mind Studio and then move on to some other underground AI tools. So the Mind Studio backend dashboard is pretty basic. I'm going to click on this little plus up in the top left and we're going to create a brand new AI. So I'll just go ahead and click Create AI here. And they've got all sorts of pre built templates for these workflows already. So things like an ad assistant, a blog post generator, internal job postings, HTML banner ads, telegram announcements, all sorts of really cool pre built templates. But what I think is the coolest, easiest way to build an AI automation inside of mine Studio is to let AI generate the initial prompt for you. So I'm going to go ahead and click on Generate prompt here and then click Next down in the bottom. And basically I'm going to describe what I want the AI to do. So before I describe what I want the AI to do, I want to give you a little bit of context. So I've got a site that I've been working on over at News futuretools IO and essentially I built this website so that I can share news on my own domain. So if you visit Future Tools a lot, you've probably seen this AI news page where I curate news from all over the web. When I come across interesting AI news, I make sure it is shared on this page. Every once in a while I come across a news article that is maybe paywalled and I pay for access to it, but it's hard for other people to see it without paying. Or I come across a discord announcement that that is essentially news. So I want to turn that discord announcement into a news post. So I created my own news page. So any news that's hard to share from a different resource, or maybe I haven't seen a different resource share it yet, I've got my own place to share it and then it'll make it on this list here. That's why not every single piece of AI news is here. It's just the stuff that I couldn't find a better place to share it. So I just made my own post about it to share it. So that's the context I want to give. And the reason I'm giving that context is because in Mind Studio, I want to create a bot that will essentially look at another news article that already exists and then reword it for my new site while citing the source and sharing some key takeaways. So Basically I can copy and paste URL to another news article and it will write it for my website. That's one of the workflows that I'd like to try to build right now. So in order to do that I'm going to start by telling the AI exactly what I want to do. If you've used custom GPTs inside of chat GPT, this is kind of similar. You're giving it that initial context, that initial system prompt to run your AI workflow here. So to describe the behavior of your AI, here's how I'm going to explain it. I want to find existing news articles, share the URL to that article, and then have that article rewritten in a new voice along with key takeaways as well as citing its initial source. I know my grammar is horrible. That's a huge run on sentence under generation engine. We really only have the one, so you're just going to leave that and we'll click generate. And now basically what it's doing is it's writing the system prompt. Essentially for me, it wrote this prompt for News Article Assistant. The purpose of it is designed to locate existing news articles, share the URL, link to the article, rewrite the article in a new voice, provide key takeaways from the article, always cite the initial source of the article. So it's, it got a little bit of it wrong here. I don't want to locate existing news articles. I will share the news articles. So I'll just put, see, I will share the URL link to the article and I'll just get rid of this first one so that it's clear. So parameters, the assistant must rewrite the article in a fresh unique voice. Key takeaway should be concise and informative. The inertial source of the article must be properly cited. And then it added some additional traits of the bot here. So now we've got this system prompt. The AI workflow bot, whatever you want to call it, knows what its purpose is, it knows what its role is. Now, now I'm gonna move over to automations and you can see I've got this blank automation here. It's got start and it's got chat. I'm gonna go ahead and delete the chat and now we're gonna build out this workflow of everything we want it to do. So I'm gonna click on this little dot here and I'm going to start by adding a user input and this is where I'm going to essentially paste the URL of the article that I Want to scrape and share a new version of. So I will click on here, over in the right we can see our user inputs. I'll add a little plus here, create a new user input. And then under type we have a bunch of options. Long text, short text, text Choice, Image Choice, etc. I want to scrape a URL. So we're going to call this variable Original URL. And then for label text, we'll put Original URL. We can see our little example over here. Don't really need to do anything else here. So now I can come back to my main workflow. I can either click on this tab up here or click on it on the left. And now we've got a user input with the original URL. All right, so now I want it to basically summarize that URL. So what I'm going to do is I'm actually going to click Send a message, but it's not going to send a message back to me. What it's going to do is it's going to create a message as a variable. So over here it says Response behavior. And under this I'm going to select Assigned to variable instead of Display to user. If I did display to user, it would basically act like Chat GPT and just give a response back to me. But I want it to actually save this scraped information here. So the variable that I'm going to assign is, let's call it Revised Article. And then we can choose our model down here. And you can see we have all of these models available. And this is one of the really cool things about Mind Studio is that I can have each step in my workflow use a different model. So if I want this first step to use Claude 3 Opus, I can do that. I want the second step to use Gemini Pro and then the third step to use GPT4 Turbo, and do that. All right, so now under send a message, I'm gonna basically write a prompt like I would do for Chat GPT. So we've got our scraped input here. If you remember, we've got the variable is called Original URL. So I'm gonna copy this variable, jump back to our flow, and then under message, I'm just gonna put these little brackets here and then paste in our variable that says Original URL here. And this is going to basically be the scraped content. So original URL is going to be the sum of all the scraped content that it did in this first step. So I'm going to put here's the original article, and then I'M going to give it a prompt. Here's the original article. And then we've got the original article essentially pasted in there. And then the prompt we're going to give it is rewrite this article in a new voice, use simple language, keep it around 500 words. I don't want it to be crazy long articles. I just want it to kind of quickly summarize the main point of the article. It's going to name it Revised article. I'll show you what we're going to do with that in a second. And then under model, lately I've been using Claude 3 Opus. I think it's the best one there is. And then for max token response, I'm just going to max it out here. And now we've got that step. So if you're following along, you're going to input a URL, it's going to scrape that URL, it is then going to rewrite the article in a new voice using simple language. And this rewritten article, it's going to save it in a variable called Revised article. I'm going to go ahead and copy our variable name because now I want it to give me a bulleted list of key takeaways. So I'm going to send the message here, I'm going to put new article. And then in these little brackets here, I'm going to put our revised article. And then for my prompt, I'm going to say give me between three to five bullet points of the key takeaways from this article. Make them brief and to the point. And once again, instead of display to user, I'm going to assign it as a variable. We'll call this one Key takeaways. And then if I wanted, I could change the model. But I like OPUS again. So I'm going to go ahead and just use OPUS for each of these steps. Max out the response tokens here. So now scrapes the URL, rewrites it, saves the rewritten version as a variable, reads the revised version and then writes three to five bullet points. And oh, sorry, I've moved my face. So hopefully you can see it now. So there you go. So you can see over on the right here, I'm assigning the revised article to this key takeaways here I'm going to use Claude 3 Opus. And then let's see here. So now we've got our revised article, our key takeaways. Now I want it to do one last step. I want it to generate a title for this news article. So I'm going to do one more send a message here and this time I'm going to grab the original. Actually, no, the revised article. I want to grab the. I'm getting myself confused here. All right, I want to grab the revised article here. So I'll copy that variable and then I'll do. Here is the revised article. Actually I need to put it in these little brackets here. There's the revised article. Please generate a simple to the point title for this article that includes the main point. All right, Once again assign a variable. We'll call this simply title. And then model I'm going to use opus. And now the final step is we can have it display everything it did back to us. So I'm going to go ahead and add another send message here. This time it is going to display to the user. And the sender is going to be the system. And in this message I'm going to have it send us the title. So I'm going to use a little bit of markdown here. If I put these two little hashtags here and then put title, it will use the markdown language for the title. So I'll put that. Our variable for title is just called title. So I'm going to put our little brackets new title. Then I want to do key takeaways. So I'll put some mark down here. Key takeaways. Our variable for that is just key takeaways. I'll add that there. And then we've got our revised article which I could just copy this here. Go to send message and put revised article. Paste that. So now if you're following along to the workflow, put a URL in here. It's going to scrape the URL. It's going to rewrite that same article as a new article. It's going to make bullet points with key takeaways. It's going to then pick a title for it and then it's going to send me a message back with title, key takeaways and revised article all in this one workflow. All I have to do now is find an article to test it with. So in order to do that, let me just go to Feedly here, see if there's a news article I can share. Let's just do this one here. Microsoft, who invests 2.9 billion in AI cloud infrastructure in Japan, I can grab this URL here and we'll use this as our test article. So let's publish this one and then I'll open the published app and you can see now here's our published app that we just created. Original URL. I'm going to paste in this news article that we just grabbed and I'm going to click next and I'm crossing my fingers that I did everything right because now it should scrape the article, write a new version, give me some key takeaways, give me a title, and then display it all on the screen for me all at once. That was the end goal that I was trying to accomplish here and hopefully that's what I get. And while this is happening, I'm just kind of taking a peek over at the comments here. Can you feed in multiple URLs and aggregate the URL? You have to do it one at a time. You can't, you know, list like a whole bunch of URLs. There may be a workflow that you can set up and get creative with that will let you do that. But right now it's one URL at a time and let's see what else you got. Somebody is saying is scraping considered fair use? Basically, if you think of like the news, right, One website will report on the news and then pretty much every other news website out there will, well then also report on the same news. I'm kind of doing the same thing. I'm citing my source. You can even see, here's what it created for me. The title. Microsoft invests 2.9 billion in Japan's AI future boosting skills and infrastructure. You can see it's not the same title, a brand new title that it just created. Here's the key takeaways. Microsoft is investing 2.9 billion in Japan. Investment focuses on enhancing cloud computing. Microsoft is collaborating with Japan's Cabinet Secretariat, et cetera. And then it shares the source here and then it actually duplicated the key takeaways. I'm not quite sure why it did that. But then we have the revised article down here and this is a completely brand new version of the article that it just wrote for me. So I can literally copy this article, copy the key takeaways, and then grab the title that it generated, jump over to my WordPress blog here and create a brand new article for it. All these images, if you can't tell, have all either been generated in Dall E or Mid Journey as well. So even more AI to kind of keep this AI news website up to date. Very, very meta. I know. So that was my AI voice refiner. Now there are other cool things that you can do inside of Mind Studio. I actually was playing around with this one called Meeting Insight Summarizer. Let me click into this workflow real quick and just show it off to you. This one is designed for if you have a zoom meeting, you can actually download the transcripts from the zoom meeting, upload them into this tool here, and then it will actually send the takeaways that you need to know from that zoom meeting over into Slack for you. So if we look at the automation here, user input, it's asking for the transcript from the meeting. So you upload a file here and then it sends a message to generate the meeting summary from a transcript. And then what it does is from that it takes that summary that it generates, sends it over to Zapier through a web hook, and then this zapier webhook actually sends it off to Slack for you. So the message shows up in Slack and then it also sends the user a message on the front end user interface to, to show you what message was also just sent to Slack. So that's a really cool workflow. If you're not familiar with Zapier, really easy to create, you come in here, you click create a new zap for your trigger. You would essentially just set up your trigger as a web hook and catch a hook. And then we, we skip the this step and then under our, and then it gives us this URL here. So we'd copy this URL and then you can see under our zapier webhook we would plug that URL in right here. And then this is the response that we got from this last step in the automation. It's passing that into zapier and then under Action, you can have the action, set it to wherever you want, send it to a Google Sheet and Evernote Slack anywhere Zapier will send a message, it can pass that message over. So pretty, pretty cool stuff. I really, really dig mine Studio for creating these workflows and automations. One other thing that Mind Studio showed off to me was this new profiler feature. This is really cool because it allows you to essentially test different large language models against each other. So if I wanted to here you can see I've got GPT3 Turbo, but I can change this to Claude Opus. And then over here I can change this one to GPT4 Turbo and I can add a third profile if I want. And let's say this one we want, let's do Gemini Pro as this third one here. And then let's see, done, done, done. And then if I give it a message here, like write me a poem about AI and YouTube and hit enter. You'll see that opus, GPT4 Turbo, and Gemini Pro will all respond to that prompt. We can see which one's faster, we can see which one generates our a better response. So just another kind of cool, interesting feature inside of Mine Studio if you want to kind of compare different large language models against each other and actually see them play out in real time. So that is Mind Studio. Again, I want to thank Mine Studio for sponsoring this live stream. I did put a link down in the description. I believe there's a coupon code on there. I think it's friends 100 gets you your first month free if you want to jump into Mine Studio and play around with it. So thank you so much to Mind Studio for sponsoring this and letting me play around with the tool. They did provide me with $50 in credits to play around with, show it off, share it for you, and you can create some pretty cool workflows. I mean, I'm kind of thinking of fairly linear workflows here, but there's some really interesting, like outside the box workflows that you could probably think of, especially when you start connecting it up with Zapier and things like that. All right, so that's Mind Studio. I've got about seven or eight other tools I want to play around with. But really quickly, before I do, let me just kind of check in with the chat over here, make sure there's nothing I'm meeting I'm missing here. All right, fix the face cam issue. Thanks, John, for pointing that out. Couple comments asking blocking settings in the bottom right. Yep. My Studio says I. I believe you set opus to temp 1. That's too high for Claude 3 opus. It should be 0.2 to 0.4 for this. Okay, so cool. Somebody in Mind Studio is actually in the comments helping me. There we go. So let me get back into. Get out of my workflow. It's up here. So yeah, what he's saying is under the temperature, I should set it at 0.2 to 0.4. So you could set your temperature there. 0.2, 0.4. And then let's see, I publish that. I would go ahead and do that with all of my Opus integrations here. So that would be the proper way to set up Opus. Make sure your temperature is set to. And that is straight from the team at Mine Studio who's hanging out in the chat with us right now. Mind Studio also says it's one month free for the pro plan. There are usage fees associated with the workflow, but you get $5 in usage as well. So thank you again to Mind Studio. Let's see here. What else should I show off in Mind Studio before playing around with some other tools, kind of clicking around, seeing if there's anything that I forgot to talk about that I wanted to talk about here. You may have watched my old video where I played around and made a Mr. Beast bot, so you can check that video out as well where I deep dive into Mind Studio. Cool. Well, thank you for. Thank you again to Mind Studio for sponsoring it. Thanks to Mind Studio for hanging out on this live stream and actually providing some additional help and constant text and tips for people. So really, really appreciate that. And you'll definitely be hearing more about MindStudio for me. I play with it a lot and I will be showing off more workflows in the future, in future videos as well. Alrighty, let me close out of some stuff here. And I just realized I wasn't showing my screen again. I'm really bad at doing that on live streams where I start talking about something and I forget that I'm not sharing my screen. And thank you to Jungle Inc. Crypto for the donation. I really appreciate that. All right, I'm going to leave it on my screen for a little while instead of going to my face cam because I have a tendency of not switching it back properly. So the next tool that I want to talk about on Underground AI is one that I personally haven't heard of very much of, but I was talking to Kip Bodnar, who is the CMO of HubSpot, and he asked me if I've played around with this tool. And I said, well, I've heard of it, I've listed it on future tools, but I've never actually clicked into it and played around with it yet. And that tool is called Recraft AI. So let me pull Recraft AI up again. This is one I haven't really played with, but it's definitely popular because it's got 400,000 plus users and over 50 million AI generated graphics. So a lot of people know about it. It's just one I haven't really played with yet. So I want to jump in, test this one out and see how it does. I'm going to go ahead and log in here. I'm going to actually pull it off my screen real quick while I log in. Once I'm logged in, I will pull it back. Ready? All right, so it's going through onboarding. I'll use it for personal. What do you plan to create all of it? I don't know. Vector image vectorization? Yes, all of the above. Have you ever used AI design tools before? I'll go ahead and say yes. Quite often. How did you hear about Recraft? I heard it from a podcast. I was on the Marketing against the Grain podcast when Kip Bodnar shared it with me. All right, so we've got a little tutorial. Describe what you expect to see, generate and browse. All right, let's Recraft by Cloud says. Does Matt play Fortnite? I have played Fortnite. My kids play Fortnite more than I do. Though I will not be playing Fortnite on this stream. I'm curious, have you guys played with Recraft before? Let me know in the comments if you have used Recraft. Also, if you have any prompts you want me to test, feel free to type them in and we will test those prompts. So one of the things about Recraft and the thing that Kip brought it up to me for was Recraft will actually generate vectors. So it'll create vector illustrations, which, if you're not familiar with vectors, you typically open them in a tool like Adobe Illustrator, but it makes the image, like, infinitely scalable. So if you're trying to blow something up and put it on a giant sign, a vector is what you're gonna want because you can scale it to whatever size you need to. A lot of people haven't used Recraft yet, so cool. All right, so we've got some different styles over here. We've got some vector options. I'm really interested in the vectors because vectors are great for logos and again, things like signs and blowing stuff up really big. So I'm going to go ahead and do a. Let's just do a vector art. Just start here. I'm going to do my. Go to a wolf howling at the moon, not howling at your mom. Let's see. Modify current image. Let's just go ahead and click Recraft. Oh, let's look at our settings here. Level of detail. We've got primitive all the way up to extreme. Let's kind of put it midway. Then we've got some negative prompts that we can add here. Let's go ahead and click Recraft. And that should generate a vector for us here. Let's see. What are you guys saying here? I do the channel to keep the basilisk away. Vector AI are not good enough. Let's see Pickle Rick in a Superman costume with super Saiyan hair. I don't know if I said that word right. Well, there's my wolf howling at the moon. That's pretty good, actually. Let's see. Okay, so this generated two versions, and this almost feels like a Figma board. And so far, I haven't paid for anything. This is. I don't know if it's a free trial or if this tool is just free to use. Well, we'll explore that here in a second. But I could scale this up and notice because it's a vector, it maintains its quality as I scale it up. That's pretty cool. All right, so I want to try this other prompt that was just given, mostly because Rick and Morty is, you know, it's trademark, and I want to know if it's going to actually pull in Pickle Rick. Let's go ahead and do cartoon. Let's see if that makes a difference here. And let's do recraft. If it actually knows what Pickle Rick is, I will be kind of surprised. Let's see. Brian Mooney says a pug dressed as a doctor. Okay. Okay, I'll try that one next. AP says does chat thing. Actually, that's better than I thought it would be. I don't think it got the Rick and Morty Pickle Rick reference, but it got a sort of humanized humanoid pickle with a Superman costume. And then I don't think it quite got the hair right. But here's the other variation, actually better than I was expecting. The S is a little bit better on this one than this one. All right, so what was the pug prompt here? A pug drop dressed as a doctor. I'll go ahead and plug this in here then. I don't know what style to do. Let's do Vector Kawaii. Try that. Let's see. Somebody wants me to generate a burger. Pickle Rick. It's Pickle Rick. Let's see. Okay, there's your pug dressed as a doctor. A little cartoony. Let's see if we do a different style here. Maybe a non vector. Or do they all. Are they all vectors? Not quite sure. Let's try psychedelic. Let's try a psychedelic pug dressed as a doctor. It's not Pickle Rick. It's Larry from VeggieTales. Yeah, that's probably closer, actually. Alrighty, so let's see a person meditating in the jungle. We'll try that. All right, so this one is the doctor Pug here. A really, really fancy doctor Pug. But yeah, it's don't believe this one's a vector. Let me see. Paste it in here. Doesn't tell me the file type export. Oh, I can export as a PNG or jpeg, but it's not a vector. All right, let's take a look at this. I want to. Oh, wait, there was one more I'm going to try here. Person meditating over a jungle. That's the last one I'm going to try and then I'm going to look into it a little bit more, figure out price, all that kind of stuff. I'm gonna do vector again because I really think the vector is where this one shines. If we're making a JPEG or png. I mean, everything else does that really well. Also, not many of these tools do a really good job at vectors. Let's go ahead and do a person meditating over a jungle and let's do it in a flat 2.0 style. See what that generates. It's really interesting because it does just kind of create this giant Figma feeling board where I can just add more images. I kept on recrafting the same image and then I realized I can just create more images next to it instead of recrafting the same image. So yeah, there's a slight learning curve, but not too bad. Let's see. Not a svg, you can click and make it a vector. Somebody said. Oh, vectorize. Yeah. So even if you do generate as a JPEG or png, you can turn it into a vector. So thanks for pointing that out, mindful mini pods. All right, so ap, let's see. Somebody is asking, will it actually make vector graphics or just a vector looking graphic? Will it scale? So this one supposedly is a vector. So export as. Yeah, we actually get vector. We can actually download it as an SVG file if we want. This one, it made it as a JPEG or png, but we can actually click vectorize and convert it into a vector. Let's see. So I don't know how long it takes, but it's working on making this one a vector. So let's just. I'll just wait for that for a second. It does people well in Photorealism like Donald Trump or Biden, and it actually does it instead of saying no. Well, that's good to know. All right, so just turn this into a vector and now I can export it as a vector file. Very cool. All right, I gotta try one more image. Now I gotta try like Trump or Biden. Just because it was suggested and it said it was really Good with people. So photorealism. How do I get to photo realism? Photorealism. All right, so we'll select that and let's just do Joe Biden on stage with an American flag. Sure. I want to see its realism. And then we'll actually dive in and look at some of the other stuff about this tool. Yeah, I mean, that's not bad. It's not the best Joe Biden I've ever seen, but it's not horrible. I mean, it's okay. It's okay. I feel like Mid Journey might still do a better job with it, but, yeah, not too bad. All right, now I really want to sort of dig in a little bit here. All right, last one, last one. John is telling me to do Joe Biden and Donald Trump hugging. Let's see what we get. All right, that's it. That really is the last one. That is the last one. You know what? You know what? I don't know what's going on with Biden's face there. And he's also got an extra finger, but that's pretty common with these AI image generators. Still, that one is actually not bad, that one. You know, you could tell who those are supposed to be. All right, all right, okay, so let's look at the subscription real quick here. I'm just curious. So for free, you can get commercial license, but the images are public for 20 bucks a month. Commercial license images are private and will not appear in the community gallery. You get faster image generation and a limited time offer, or that's if you pay yearly or if you pay monthly, it's 25 bucks a month. So it looks like there's no real limits and they're available for commercial use. So pretty, pretty, pretty, pretty cool. All right, so let me just look at my notes here. Make sure I covered everything I want to cover. Covered how much it costs. I think so far what the value in this tool is is really for making the vectors both in generating a vector straight up or actually vectorizing a non vector image. So that seems to be the best use case for this if you're looking for realism. I feel like midjourney is probably still a little bit better at realism, but this one's, this one's really, really good at the vector stuff. So just turned that into a vector for me. If you guys want to see, you know, Trump and Biden much larger on the screen. There you go. I know you all wanted that. Yeah, that one's pretty cool. Especially since you can create vectors seemingly completely for free. I'm not sure if there's any limits. I don't see any sort of credit system or anything like that. I like this sort of Figma board style, this giant, I guess, like infinite whiteboard where you can just keep on generating images. That's really, really cool. So give this tool a thumbs up. I like it. Let's see what do we got here? All right, so I'm going to move on to the next tool here. Otherwise I can definitely spend 30 minutes on every tool and we'll be going until the middle of the night. So I'm going to move on to the next one that I wanted to play around with. And this one's actually open source. This one was by request on Twitter. Somebody asked if there was a open source tool that will take a still image of a person and make the lips move. So there is this tool did. I've played around with it in past videos. If you've been around the AI space for a little bit, you've probably already seen it. It's a tool where you can upload a, you know, a single 2D image and it will actually make the mouth move to an audio. So you can type in some text, give it a sign of voice to it, and that person will speak in that voice. Or you can upload an audio file and it will sync up the voice to the audio file that you gave it. And somebody was saying, hey, is there a free open source alternative to did? And there is. It's called Sad Talker and you can actually install this locally on your computer. Run it on your own GPU if you want. But I did find a hugging face space that also does this. So this is kind of the same idea as did. I feel like DID does it a little bit better. There's a little bit more motion in the head and a little bit more blinking. It feels a little bit better. But this essentially does the same thing. There's a handful of examples down here. This one's my favorite because this is the face they gave, but this is the audio input. The holidays are a time of celebrating, with loved ones enjoying the festivities and muddling through Wisconsin's cold. And so when you put the two together, you get this video. The holidays are a time of celebrating, with loved ones enjoying the festivities and muddling through Wisconsin's cold. But for many, this year has been. And so the voice and the face don't really match up that well. But you get the idea of what it can do. There's some other examples here. Let's try this One, here's the input audio, And then this is the image they gave and then this is the output. So if I wanted, I can actually remove this image. Let me see if I have an image on my hard drive somewhere that I can just toss in here real quick. What do I got? What do I got? I won't use that one. I was going to use an image of someone else, but I don't know if they would appreciate it. So let me go to my images here. Headshots. I'll go ahead and use. I'll use this image, which is like my profile picture for everything, and I'll just use this same audio. Actually, you know what, can I do a type? Where do I type? Thought this one had an option to type in some text and it will actually do the text, but I'm not seeing it here. So maybe I do have to just select some existing reference audio. I'm just going to pick some existing audio just to keep it cool quick and replace it with my head here. And then I'll click generate. And it's too busy right now. Keep trying. All right, so there you go. That's. Yeah, it won't generate right now because the hugging face space is too busy. I could duplicate the space, but if you've ever duplicated a space on hugging face, sometimes it takes a good five minutes just to get the space duplicated. It's kind of a bummer. It's too busy. Of course, the moment I'm doing a live stream, this one's too busy for me to use. But I guess, you know, the idea was to show you what it does. And if you know what DID does, which is short for DE identification, by the way, in case you're wondering why they call it did, then. Oh, wait, hold on, hold on. Let's see. What is AP saying over here? All right, so he's telling me about one called Any Portrait Official, which is actually, he says, more state of the art than even sad talker. All right, so this is definitely the first time I've seen this one because AP is pointed out to me right now in the inside of the discord. I've got a little discord going with the other moderators here. Here, I'll type it in. It's-id.com for anybody that's looking for the one that I was talking about. So let's look at some of the examples here. Pull up this one. So here's the reference audio, Here's the reference image. Let's go ahead and generate. This is going to take about 28 seconds here. What we get? Let's see. Please make a video to favorite AI tools learnx AI romo AI. Cool. Just kind of keeping an eye on the comments. All right, so here's what it generated. It's only one second. Why it only changed one second. Did I skip a step here? Length. I got to crank the length up here. Try this again. And I should probably turn off my Slack notifications while I'm on a live stream. All right, so now we're trying to generate. Hopefully I can get it to generate a little bit longer of a video. And then next I'll try uploading my face, See what we get. Then I'll move on to the next tool that I had on the list after this one. Anytime now. So the next one. Oh, there we go. It is going to be. So this now generated 3 seconds border. So if you're looking for a way to take a 2D image and add audio over it, probably the simplest way to do it if you're willing to pay or try the free trial, is this did. Com. It does exactly what I'm doing, but.