We use Aitana López, an AI-generated influencer, as inspiration for our next project. Let's build an AI model using Midjourney, ChatGPT, and Custom GPTs. Try the Image Caption Generator: chat.openai.com/g/g-yCGszMTwJ...
@@TH15N4M315UN4V41L4BLApollo 11 program gave the boost to microprocessors which fast-tracked their evolution. Today would have looked very different otherwise…
@@TH15N4M315UN4V41L4BL Almost 60 years now since the last moon landing... Makes you wonder doesn't it? Especially since NASA has openly stated that the technology for humans to leave low-earth-orbit doesn't exist yet. In fact nothing organic can survive beyond earth's geomagnetic field except on a very small scale right now for the sole reason that radiation shielding that can block radiation from the sun is simply far to heavy to build a compartment that would be able to shield a human. Surely science is very close to solving this problem though or else Elon Musk wouldn't be moving forward with a plan to send humans deep into space on a trip to Mars within the next 30 years. In theory individuals, entire vessels, and eventually even entire settlements will have to be protected by their own manufactured magnetic "force fields". That was nature's natural solution for Earth and is probably our best bet as well. That or be surrounded by A LOT water or lead, which is today's tech, and is very primitive compared to what the laws of physics would suggest is actually possible with more advanced knowledge.
@@TH15N4M315UN4V41L4BL Indeed, artificial intelligence has greatly shortened the years of science and study. Anyone can become rich in a very short period of time. But I swear to you, by the One who created you, me, and everything else, if I prove that they went to the moon, I will give you whatever money you want,
Hey Wes, i’ve been into AI for about the past six months now. I wanted to throw a compliment your way - out of the hundreds of videos. I’ve watched on AI, your presentation skills are the best. I love the way you don’t talk 90 mph and at the end, you recap the way you created the AI model. Super impressed with you, and you have a new subscriber. Good luck, and see you online!
This is so nice of you to say and I really appreciate the comment. I've been creating a document of all the best, feel-good comments and have added yours to the list - just so I can take a look back every once in a while to remind myself why I'm creating videos!
I watch a lot of YT vids and by far yours are the most in depth thus far. Knowledge, voice, skill......all at a followable speed. You deserve a YT content award. Thanks! Your hard work is appreciated.
Thank you so much for your kind words! I appreciate you 🙏 I wish I used the Face Swap in the video JUST to show you guys how it worked, but completely forgot. My bad.
Wes.....you rock. I was looking for some information on this and boom.....a video with everything I wantedto get started. Thank you for the information. You did the video perfectly. No boredom, straight to the point and a step by step to boot. Again, thanks a million!
00:01 AI model Itana Lopez earns up to $11,000 per month 02:08 AI models generate income and have increasing popularity 04:00 Mid Journey is an AI image generator with realistic outputs. 05:50 Using AI to generate photo-realistic model shots 07:35 Generating new images using AI with seed link and prompts. 09:16 Creating an AI model to generate Instagram images 11:00 Creating AI influencer profile on Instagram 12:48 Using chat GPT to generate photo ideas for an Instagram model
@@iosushi This is mentioned at the start of the video (in the Business Insider article). They suggest: 1. Sponsorships 2. Advertisements 3. Subscription to Fanvue
How do we know these earnings are real? Everything can be faked....and sooner we will see "gurus" teaching us how to make "zillions" with AI models...is kinda obvious ! Ofc we can make money but not that easy...
@@MauiBodyworker Yes, two places: 1. Within ChatGPT: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-tktRnxszaFg.html 2. Turning it into a paid web app: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-rd4AeG1wBRI.html
omg, this guy is the best I have ever heard. He speaks and explains everything clearly, and in order. I am new to AI and want to learn more about AI stuff, and how this guy teaches and explains it is so good. I actually feel like I am learning from someone smart and patient. Keep up the good work.
I know, so sorry. I instantly regretted NOT filming this after the video was done. I made an old video about that FaceSwap tool if you're interested (it's not my best work though so no judgement haha) - ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-D1D4_9aqc2Y.html
Face swapping works like charm untill you want to add glasses or something obscuring face. You can all that stuff using free stable diffusion, it's much better then midjirney at that point and is free. You can generate random people then use that to creat Lora of said character giving you great flexibility while retaining coherence and same face without need of face swapping. That's before you start using controlnet etc giving you much much much more control then all the paid generators do together.
Part of the problem comes from before AI due to the air-brushing of model photographs. I personally don't think we can differentiate now between real and AI unless there are gross errors in the AI model like warped hands. The "too perfect" rule can't be relied on anymore because we have had "too perfect" models for decades now due to air-brushing so is it AI or merely air-brushed?. I guess you could say that the demise of truth and reality really did begin with Photoshop and now it's just way more efficient.
Good thoughts. And you're right... looking at the photos I generated and would honestly not know they were AI unless someone told me. Because our TikTok and Instagram feeds have been showing these types of photos (but of real people) for ages now.
I think the important part is that if you are willing and openly state your model is an AI and do not hide that, people are often willing to ignore this. And not in the sense of denial, they simply do not care. At the end of the day, fake or not, you are getting to see pictures of a beautiful model that suits your fancy in some way. And most who are seeking to drop money on that sort of thing don't tend to care about how real the individual is. If the content is provided, the customer is happy. It's been used as a marketing ploy for generations. People do not often care how they get what they want as long as they get it. Consider that for generations, no one cared about the deaths of animals to put food at our tables until someone decided to yell about it on social media. And even then, still don't hear people complaining about it often.
@@christopherpoet458 "And most who are seeking to drop money on that sort of thing don't tend to care about how real the individual is." I definitely have to disagree with this. I think knowing if the girl you are extremely attracted to in this instagram is real or not, not only matters but would be the smoking gun/dealbreaker for anyone to pay money for it. Just think back to your celebrity crush when you were a wee naive little shit, wasn't half the turn on that 1 in a milllion famous Jim Carry "Yes I've got a chance" quote? Wasn't it understanding the Many Worlds Interpretation at adolesence and jerking off to science knowing that there is a universe out there where you and her are in love and it's the same universe they cure all aging and disease so you literally get to spend eternity with the woman of your dreams? However, it is highly likely the outcome I just proposed would be just 1 of a quadrillion simulations anyway so perhaps you were right all along and it won't matter because real will have lost any validity to it's meaning that it's true source would be so far away from us that the only way you could reach it is for a whole society to all agree that what they are seeing is exactly the same as the person next to them?
Clothing and fashion, alot of old skilled patternmakers and technicans are being laid off now. Its one of the reasons why quality and fitting sucks. It is because a large portion of the design and pattern making are now being handled by AI. Most young people don't even know what quality fabrics or fitting is like so they buy all the crap in stores. You only see remnant of quality in overpriced high end brands nowadays.
Wes, this is awesome man! In-depth, explained in detail, and where you can follow along. Subbed! I think I should create an AI and...manage my model. 🤣
The better way to do this is to skip midjourney and just go free with something like SDXL/ComfyUI. You have the benefit of working off of community based checkpoints, creating your own, or using a base model with a custom lora. The trick is to generate a face that you like and then do a face replace on the images you create. This way, the face will remain consistent across generations and you still retain the ability to be creative with hair color, style, eye color, makeup, accessories, etc.
If we have a weak PC, can we instead buy a cloud computing service to emulate a powerful computer? Or the cost per time will be totally ridiculous for imagine generation?
@@WesGPT I'm curious about one thing. Only the Pro and Mega allow you to use Stealth Mode, where you can only see the prompts you write. Am I wrong? When I watched the video in your example, I saw, "Only you can see this." But you mentioned that you only signed up for the basic.
@@Physics_Dude You're correct. People CAN see your prompts and images on the Midjourney home page. I'm using it in a private server though, so that's why it probably says that.
Thanks for sharing. I have used a couple GPTs with good results, but just began working with images the other day. Disappointing, but I suspected that had more to do with user errors by me than the GPT. Still, I had to laugh when you asked for long nose, and Dall-E just completely ignored you, and when you requested specific improvements, it just got worse. Now, I wonder if the my problems had more to do with the GPTs I used. I appreciate that you went from Dall-E straight to Midjourney. I'm about to plunk down the money. It seems that it is just more simple to go with the one that works most easily rather that burn a lot of time experimenting with the others.
Yeah, I'm with you. DALL-E 3 has it's place in image generation (primarily for logos, text, etc.). But for people, nothing beats Midjourney. Some viewers also told me to try Stable Diffusion but haven't given it a shot yet.
@@WesGPT You did half the job, which was an excellent exposition of the technology. The other half, making money requires an agency or knowledge of how to advertise your AI model.
Some people treat this AI influencing as brand new. It's very much like people having a favorite author of fiction creating a fictional character, feeding his/her fictions and fantacies into your mind, maybe including product placement, and then it becomes real and influencial on you.
Yep, exactly! I guess it's just weird for people because of the beauty/sex angle. But another viewer said they've been doing stuff like this in Japan for almost a decade now (but anime characters)
@@WesGPT A certain amount of attention from mainstream media makes it cool and hype and places it in the public consciousness. They are often late to the party :)
"If you really look at it, you can see it is a AI model" I agree with that, but at the same time, since real models are edited so much, they also look just as fake as the AI model, so it is understandable that at least people who do not use AI them self will not see the difference. I love the motivation for creating this model without a huge ego.. lol Imagine all the struggle he had with some of them before he did this.. lol
From now on to make a totally honest and natural fashion add you have to have real models without photoshopping and even give the model's name to make her real and personal...
Your teaching style feels so authentic in helping others. I appreciate the break-down and the speed of info given. I'm just starting out with AI stuff and have wondered how to do something like this. I was a little unclear as to what the correlation is between Midjourney and Discord?? Also, is the Image Caption Generator in ChatGPT something only you have access to? I am not too keen on using my own face for privacy reasons, but I am curious if this kind of thing works just as well based off an existing personal photo, that can use the same face to regenerate various images and settings. Apologies if this is a stupid question lol.
Not a stupid question, I appreciate you commenting! Midjourney is a Discord bot. That's the relationship. Soon, you'll be able to generate images on Midjourney.com so Discord will no longer be needed. As for Image Caption Generator, you can use it here: chat.openai.com/g/g-yCGszMTwJ-image-caption-generator (you do need the ChatGPT Plus plan though)
You should utilize Midjourney's permutation prompt tool. Basically, it allows you to put variables into the prompt. This allows you to change the subject or any other part of the prompt. For instance, "wide shot of a brunette woman laughing {in a mall food court, on a yacht, in the forest} shot on kodak gold 400" This will generate 3 sets of 4 photo blocks. Each one will be like you created 3 different prompts, but you did it in one single prompt. I think this would have been helpful for you in your video. Although the video was really great and separating them out may have helped me more in understanding the material better than if you did one single prompt. Hope that helps.
This is amazing, great video. Makes you really want to give it a go. What other shot types would you recommend, as opposed to Kodak Gold 400? For someone who doesn't know much a bout photography? Thanks
So with the newest version of Midjourney (v6), you don't need to throw in the camera type to get realistic photos. Just makes sure the setting is on RAW. The next video in this series goes further into human photo-realistic generations: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-hAyCRJKBzXA.html
I love how they're not actively trying to deceive everyone and still get that much followers lol Well, it's not like they can, anyone who have some experience working with Stable Diffusion or Mid Journey would know it's AI generated. Of course, it would look real for those who hasn't dealt with a lot of AI generated contents, specially those coomers lmao
I sent this profile to my cousin, who doesn't have a lot of experience with AI, and he was completely fooled. Like you mentioned, I'm going to try sending to a friend who IS experienced and see if he can tell that it's AI-generated.
I do AI model as well, I use Stable Diffusion XL with custom trained lora model and use fine tuned checkpoint from civit, it works extremely well, although it requires local hardware like RTX 3090 or rent a server to process this. My bigger question about Aitana Lopez is what's the percentage of her followers that are even aware that she's an AI model? What a way to take advantage of low IQ simps, lol.
I've been hearing that Lora model floating around but haven't looked into it. I'll check it out! As for the people who don't know she's an AI model - a lot of the comments were saying how much they love her. Yikes 😬
From what I see in the comments most of them seem to know. The comments section is filled with people saying 'this is AI' and fans replying 'I know'. Seems a lot don't care.
After a few trials, I found that the AI tends to replicate not only the face, but also the elements (including clothes) of the original picture. In my case, i couldn't get rid of some gym walking machine. In your video, the windows seem to be always around. How can you make the AI only takes the face of the original drawing?
A better original seed image would be just a picture of her face (like a close-up portrait). In the video, I use one of her on a boat which is probably not the best idea. You want as little in the seed image as possible (in my experience)
Awesome tutorial. Say we want to use the AI model for a fashion brand. How do we get the model wearing the brand’s clothes? Can we upload a photo of the product and ask MidJourney to combine those? Any ideas are welcome. Thanks!
GREAT question. I actually didn't know the answer to this but then @LiebsterFeind reminded me. There was a demo of Adobe Firefly recently where they changed the clothing of a person live, in video. So this would be the way to do it.
Currently, the closest option available involves utilizing Adobe Generative Fill or Adobe Firefly. You can alter the attire of the model by specifying what the AI model should wear. However, it's a hit or miss when it comes to using a particular brand.
Ok. So roughly $50 a month subscription fee for all the AI tools? How much for the software to generate videos with the same model? Also for the chatbot, how to trade it? Thanks for the knowledge
I wouldn't say $50/month for all the AI tools. Can definitely get away with $10/$15 per month. For the software that generates videos using the AI models - I'll have to dig deeper. And I assume you mean TRAIN the chatbot (not trade). You can just use GPT-4-Turbo in the OpenAI API.
I am offering this type of AI model for my clients who can't afford to hire real models. Example coffee shop. I created an AI model drinking coffee and stuff that is related to their business. To have a consistent face I just use a face swap tool.
Yes, I create images for their daily posting on their Facebook page and Instagram, and it's not limited to AI models that I create. Sometimes, I generate coffee mugs images using AI with quotes and sayings related to coffee.@@Jcs187-rr7yt
@@Jcs187-rr7yt Sometimes I offer them a bundle of 30 images that they can use for daily posting or 60 images for two posts per day in a month, so they have 30 days of content by using scheduled posts. Sometimes I offer free captions and hashtags using ChatGPT for the images.
This is a very good tutorial and I'm subscribing to your channel so maybe, I have a starting point to learn about this topic (AI in general) and then more as this video has done. This topic seems so big and complex that I can't figure out where to start and a path to advance.
Appreciate the kind words and welcome to the channel! It's definitely a big topic. I've only made one more video in this "AI Influencer" series and it's this one: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-hAyCRJKBzXA.html Check it out and let me know what you think.
just a reality check though... have you seen anyone coming up with I earned real money with it? the model is generated, doesn't wear the clothes they suppose to promote, no jewelries, they can't say anything, can have no reels (unless you rip a vid of someone else and swap a fake face on it). She can never come in public etc. This is just an unrealistic bubble which has thrown in the air where nobody actually has seen any monetized results from it.
I suppose. If you've been living in that bubble. AI can talk, move and wear anything the generator can put them in. Tupac performed onstage over a decade ago (2012), AI can definitely make that leap. The technology was not only available this entire year, it's consumer-level which means you and me can use it, do it and profit from it.
It's more about the AUDIENCE and the REACH that these AI models can get on a platform like TikTok and Instagram. The fact that so many comments on her page revealed they didn't know she was AI, shows its potential.
If you can use it to build followers you can then do whatever you want with all those views. Sell branded merch, make money off of affiliate marketing links. I have actually thought about trying to make an AI onlyfans page. You might be able to dupe a lot of people into subscribing Edit: I hadn’t yet watched the video and now I see that’s basically what these guys are doing lol
Hello. First of all thanks for the detailed guide. IMHO the final images look a bit blurry. Is it intentionally or is it the result of the RU-vid compression?
This is mentioned at the start of the video (in the Business Insider article). They suggest: 1. Sponsorships 2. Advertisements 3. Subscription to Fanvue
When you add a Only fan link after a few months you make 100.000 when the video generation also works and the image/video generators get good even when not restricted. As I heard the most money is payed by people looking to chat with these models so if you have a chat bot which talks to people this industry is dead for normal only fans models and their e pimps. I think the New York time made a podcast about e pimps making money with only fans where the e pimps hire writers for the models and only ask the models for images and videos all the texting is done by other people. The e pimp do the whole marketing and money grabbing the girls make the pictures/videos and get a share of the revenue. Also are you sure with some more Prompt engineering Dalle 3 would not be able to deliver the same realism? I think there is a lot of content on how to write better prompts for Dalle 3 the fact that it is free is a big advantage and most people would just try it out if it is free. I also follow your AI Model curious how this evolves :)
Right, and the chat-bot will probably be way more attentive and interesting, a digital Cyrano de Bergerac on steroids available 24/7 that never tires and never has an off day.
So intriguing. I've never heard of the word "e-pimp" but it makes sense. Adding a chatbot is a fantastic idea for an add-on to this method. I might make a video about it because the video here looks to be pretty popular.
Wow, I don’t really use any AI and didn’t know what it can do. Now I’m amazed by AI. Thanks for the great video. I’m going to start learning AI from now on.
Awesome presentation. Thanks for breaking it down then doing a final wrap with the steps. Really well laid out! Here's a question. If the created AI actually starts generating AD revenue, who is able to collect on the account? You the creator or the AI. It would seem a bit of a legal grey area?? It would suck to see your creation generate revenue you're not able to access. Probably a lame question as I'm not up to speed on AD revenue collection. Never had more than 9 views on a vide
You realize the AI is a program and not an entity with a bank account and legal rights, yeah? This is like saying "Hey, I created something in Photoshop; it's making money now, do I get the money? Or do I pay the photoshop program?" lol
As someone who works with influencers, I get the frustration sometimes 😅 I have started to work on my own AI influencer just to see if she can blow up, like the others, but I have started with a more complicated tool and video. So I am happy to have stumbled upon this video, that is super easy to follow! Thanks ❤
A most inspiring video!!!! I think Ive already seen a number of female and male AI models on Facebook, Instagram and other social media. Some look rather cartoonish but others are so realistic I cant spot them. Is it the eyes that are the giveaway or other facets of the image.??
Wow Wes this was so cool! I love how you explained and press click by click so important! New subscriber keep at it!!! She reminds me of an actress let me see if I can think of her name.
Great video! just a quick question, any idea why the pics generated after the seed pic are all in black and white? also when i add the image link and add the new prompt it regenerates the original picture so i change the the original prompt description and it says it cannot validate.
Hmm, there might be something wrong with your seed image that is glitching the system. Try starting with a brand new one. Definitely should not come out black & white consistently unless you've asked for that.
@@WesGPT thanks for the reply, I kinda solved it just by adding "in colour" with in my prompt lol, I'm slowly understanding what I'm doing. Thanks again 👍
Awesome video, great recap. My only bugbear is you didn't demo how that face image thingy worked. I know you didn't need to use it but a demo would have been nice. Still though, good enough to get me subbed, so thank you.
I know, my bad! Definitely forgot while filming. I do demo it in this video (if interested) - ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-c9TjVE0sUR8.html
I guess i will never do this, or at least, not for now. The hardest part is to buy the Midjourney basic plan. But as i can see, it worth the price, as it's very realistic and very impressive!! Maybe your way to show things is what make this also attractive.
Great info! One question, You changed from half body to full body in your Midjourney prompt, but it still gave you all half body shot. Can it create a full body, legs, arms etc?
Generating full-body shots is tricky as it defaults to faces and heads. Try describing features of the legs/feet and it may trigger a full-body portrait.
Wes, been following your channel! great content, especially like your experimental stuff. Question- You think it is possible for mid journey to create an AI influencer that can feature products that you define in the image itself? Example - AI influencer you build, featuring product xyz for example (an image you would upload)? What you think? That could take product placement to the next level.
Hey Wes, thanks for all the continuing awesome content. Any more further improvement on the ability to build your own AI influencer showing your own products or any products that you ‘ask’ them to? 😂
How would you add accessories / change outfits. If you're looking to have your model advertise things, the model needs to wear those specific things. ie - a specific pair of sunglasses. Any ideas on how to put a specific pair of sunglasses on your model?
@WesGPT Could you make a follow up video to explain the business, legal steps and profit side of AI modeling? Including ownership, copyrights and marketing.
@@WesGPT I think that we would all be interested in you "doing a full tutorial/deep dive into making an AI influencer business". You certainly know your "s***" better than most. They contain so much information that it hards to stay above water, before the water level rises. In fact, here's an idea for you. I, and I'm sure a lot of others, get some of my most useful information, from the comments section. I normally cut and paste, select and useful comments that often involve info about the use or effectiveness of various apps, gpts, methods, ideas, etc. It's horribly time consuming! What about creating an agent that "strips" the most relevant comments and separates them from the whole? You could even have interactive filters that allow the content stripping, based on: Instructional, software mentioned, advancement of an idea, etc.
@@PhilipCardwell I will definitely make a full AI influencer tutorial/deep dive soon (it's on my list). On the other point, let me think about this AI agent idea!
Well done. Great concept. I am going to do this for my instagram channel in the make money online niche. I am an old dude and nobody wants to look at me. Thank you.
Great Video Thank you! When I try to save my image at around 7:12 in your video, I have no "save image" or "copy image" button that pops up. All of the others do but that one. Any suggestions?
This is wild. 😲 I think the second two look a little younger than the first (especially the smiling one), but it's crazy how you can do this so easily. Can you make videos of the same model also?
@WesGPT No, I started with just the images. After the 3rd prompt, they started looking less and less like the saved id image. The face swap just made them look smoother with less detail and more like edited ai images than a real person. I'm gonna keep working on it to see if I can make it look better.
I made a video about this tool a long time ago but might update it! You can watch it here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-D1D4_9aqc2Y.html
This is really cool and helpful, but just like you said the ethics of AI and how some people didn’t realize she was fake. I find it weird that you were talking about this AI basically and gendering it. But now I’m doing it too lol.
There's an obviously deformity and remarkable discrepancy in the symmetry of the arms, biceps and triceps, in the photo with her arms raised. AI still has issues with human anatomy generation in dynamic kinesiology.
