ULTIMATE AI Influencer Creation Guide | START HERE

When I first started learning how to create AI influencers, it took me literally months to get to where I am now. And I still have plenty of things

now. And I still have plenty of things to learn myself. But the biggest problem on the beginning was that I was extremely overwhelmed.

>> I didn't know what to do. I didn't know what model to use. I didn't know what was Comfy UI. I didn't know what GPU I need.

>> I didn't know how to train my model. I

didn't know what to do first. And since

I posted my last video, which got quite the success, I've got tons of messages, people asking me these questions like, "How do I start? What is a Laura model?

Do I need a Laura model? Do I need to learn comfy? How to do all of this? I

learn comfy? How to do all of this? I

don't know what to do first." And these problems are literally the same thing I had on the beginning. So, I have decided to put all of my knowledge into this

whiteboard. I have decided to structure

whiteboard. I have decided to structure it step by step so I can tell you everything I know so you get the whole idea and you know exactly what is it.

You know exactly which model to use and you know exactly how to start. So before

we start I want to give a disclaimer that this video will not focus on the marketing side of your AI influencers.

This video will focus on all of these things on the creation of an AI model and marketing video will be created in the future. So why should you even

the future. So why should you even listen to me, right? Uh if we look at these images, they look quite realistic.

If you look at the skin texture, it looks extremely detailed. It looks

realistic. And the reason why I am able to achieve such results and I am not achieving these results which do not look real at all is all of the things

I'm going to teach you. I'm not going to gatekeep anything. I will give you all

gatekeep anything. I will give you all of the knowledge I have so you can achieve these results. So let's get over the structure of this video. First we

are going to learn some basics. What

even is an AI influencer? Why you should create one. Then we are going to go over

create one. Then we are going to go over the requirements. If you need a GPU, you

the requirements. If you need a GPU, you don't need a GPU. What even is an image model? How does it work? Do you need to

model? How does it work? Do you need to understand the math and the process of how to start? Then I'm going to tell you all about the environments, all about the models. Then I'm going to tell you

the models. Then I'm going to tell you about Comfy UI, which is a software that you need to learn to use all of these models. Then I'm going to talk about the

models. Then I'm going to talk about the models itself. And step number eight is

models itself. And step number eight is the biggest problems of most people and it's achieving consistent phase. There

are many methods and we will also cover the best possible method which is a lot of training. I'm going to literally

of training. I'm going to literally explain everything in total detail so you understand it even if you are mentally [ __ ] because that's the type of videos I want to do. We are

going to cover the data set, the training tools, how to use it. Then we

are going to cover some prompting not safe for work images which are also important if you want to use only fans and then we are going to look at video

generation and best realism methods. The

purpose of this video is not to go on extreme detail in every possible category because this video would have to have 50 hours if I wanted to do that.

The purpose of this video is to give you the context between all of these things, between models, trainings, what even is it, how to use it. So when you actually

decide to learn and decide to create your AI influencer, you are not overwhelmed, you know exactly what to learn and you know exactly how to start.

Also, the purpose of this video is when I get a message on Discord of someone asking me, "Okay, Philip, the air table and all of this, it's cool, but how do I start? what do I need to train a model

start? what do I need to train a model or stuff like that? I can send them this video so they know all of the things that are and they can focus on the one

thing they want to learn today. Because

the thing is you cannot learn anything that you don't know that it exists. You

know, you cannot learn data set creation for Laura models if you don't even know that it exists. So in this video I will show you everything that exists and I will give it into context so you can

actually start learning and start practicing. The first part of the video

practicing. The first part of the video might be a bit boring but it is extremely essential because if you want to build a rocket ship you have to know

the basic physics right if you don't know how to calculate a velocity of a vehicle for example you cannot build a rocket ship. And this is the same with

rocket ship. And this is the same with AI models. If you don't know the basics,

AI models. If you don't know the basics, you cannot go into the advanced stuff.

So, this video will not be some hyper stimulating editing retention stuff. It

will give you everything you need so you know exactly what to do and you are not overwhelmed. So, let's get started.

overwhelmed. So, let's get started.

Number one, basics. By the way, if you want to get the most knowledge from your AI influencer journey, join our discord which is in the description which is completely for free and we have our own custom models. We have free resources

custom models. We have free resources such as prompt generators and also there are free workflows that are ready to use, ready to input to confi and you do not have to set up anything yourself.

What even is an AI influencer? Why you

should care? Why you should create one and how to monetize them. Well, it is pretty simple. It's not going to take a

pretty simple. It's not going to take a lot of time. AI influencer is a set of images or set of videos, an online personality that you can post on social media and then you can monetize them. So

if you create a beautiful girl then you can create an only fans or fan view and then horny people on the internet will pay you money. So if you create a good AI influencer that people won't be able

to determine that it is AI you are going to get a pretty good chance of people paying you money. I also believe that AI influencers are the future of social media Instagram and Tik Tok because they

are going to cooperate with brands like Edidas Nike or stuff like that. So now

we have that out of our way. we can

focus on the important stuff. So you

just discovered AI influencers and you have no knowledge at all. So what do you need? Do you need a good GPU? Do you

need? Do you need a good GPU? Do you

need a do you need to rent a GPU? Can

you run it locally on your computer or what is it? Well, most of the image models that we are going to get into in the part number three are pretty

computational heavy. So, if you don't

computational heavy. So, if you don't have a good graphics card, for example, a good graphics card for 2day video models is going to cost you $10,000,

then it's a pretty good idea to rent a GPU online because you have two options.

Like, if you have a really expensive and good computer, fine, you don't have to rent a GPU. But if you don't want to wait 2 hours to generate one image, it's

a good idea to rent a GPU online. We are

going to go into detail on all of the platforms later. So, don't worry if you

platforms later. So, don't worry if you don't have a GPU. You don't have to pay $10,000. You can just pay like $1 an

$10,000. You can just pay like $1 an hour and you can rent a GPU online and you can host all of the models there.

So, you do not need a good GPU.

Part number three, how does it work?

What even is an image model? Well, if

you want to understand what is an image model, think of it like this. There were

some computer scientists and they took billions and billions of images and they captioned all of these images. So they

took pictures of cats and say this is a cat, this is a cat, this is a cat, this is a dog, this is a tree and then they converted everything into computer language. They used some machine

language. They used some machine learning techniques and they were able to create a model where if you type a image of a cat, you will get something like this. When you type image of a

like this. When you type image of a tree, you will get something like this.

And do you need to understand the machine learning math behind it?

Definitely not. I don't understand it. I

have absolutely no idea how the math works and I'm using it. And if you want to know the process in more detail, it's quite simple. There is a noise image.

quite simple. There is a noise image.

There is a predicted noise based on the prompt you have inputed. and the image models the very smart computational model starts dinoising the images until

you get an image of a cat. So these are image models in a nutshell. It's kind of important that I said that because we are going to need that knowledge later.

So what is the process and how to start?

Well, in ideal world, you would want to create a reference image. Then you are able to create a data set. Then you can train your model. Then you generate images, you generate videos, you

generate not safe for work images and you just start posting. It's seems kind of easy, right? Well, in reality is that you try to set up conf, you get some

errors, you input them to check GPT and after a week of pain of installing custom nodes, of installing the models, you get your first images. But the

problem is that you are a complete beginner and the image looks fake as [ __ ] So you go to our Discord, you ask questions, you learn, try test, and you fail. And you fail over and over and

fail. And you fail over and over and over again until you get the hang of it.

So the goal of this video is to go from this to this. So how do we do that?

Well, I recommend you watch the entire video. You follow all the steps I'm

video. You follow all the steps I'm talking about. You are you are going to

talking about. You are you are going to try everything yourself. So when I'm going to talk about setting a setting up comi, when I'm going to talk about renting a GPU, you are going to try that

yourself because if you just watch the video, you are never going to be able to learn it. And don't expect that it will

learn it. And don't expect that it will be smooth and flawless.

The process will be painful because if you have never done anything like this, you are going to have a lot of errors, lot of problems, a lot of dependency problems. So you try it, you fail it and

you repeat you input everything into chat GPT and after a few weeks you can get a good hang of it. So now we are getting to the start. So step number one

is to decide what environment are you going to use. Do you have an extremely powerful computer and you are going to run everything locally or you have a

[ __ ] computer like me? So you are going to run everything on a rented graphics card also. Then you have to decide if

card also. Then you have to decide if you are going to use local models that you can either use on your own computer on the or on the rented GPU or you are

going to use API models. And now we have to answer all of these questions. What

is an API model? What is a local model?

Who is an API provider and what is an opensource model? Well, let's get

opensource model? Well, let's get started with the API models. You see

that there are two types of models. Open

source model and API model. Also, there

are the useless models like Picasso and open art that host some of the models but we are not going to use them. So,

think of it like this. API model is a model that is running on someone's server. Imagine there is a Google server

server. Imagine there is a Google server and you send a message from your computer to the Google server and you tell him I want to generate an image of

a cat and then the Google server processes it and sends it back to your computer and you get the image. It is

not computational heavy for your computer at all. All of the heavy lifting happens on the Google server

and the best providers of these API models is full.AI or wave speed. The

reason why you should use an API model is that it's not computational heavy at all. And also there are some image

all. And also there are some image models that are good for our use case but they can only be used through API.

They cannot be used on your own computer. So it's good to learn these

computer. So it's good to learn these providers. So to recap that an API model

providers. So to recap that an API model is a model that is running on a server and you just send a prompt and the server returns an image. And then there

are opensource models. Open source means that all of the backend code that you do not understand can be downloaded into

your own computer and it can be ran on your own computer. So you do not need to pay anyone. You do not need an internet

pay anyone. You do not need an internet connection and you can just run it on your own computer. And as I said before, if you have a good computer, you can use

it on your own computer. And if you don't have a good computer, you have to use a cloud and you have to rent a good computer. You have to rent a good GPU.

computer. You have to rent a good GPU.

And there are two main providers of such GPUs. There is a runpot which in my

GPUs. There is a runpot which in my opinion is easier to set up but is a bit more expensive. And then there is

more expensive. And then there is vast.ai which is less expensive but it's more complicated to set up. If you want to learn how to use Ranpot, watch the video

in the description because I have a 20inut tutorial covering everything on RunPOT so you can set it up yourself.

Also, if you use my referral link on RunPOT, I'm going to get more money and you are going to get more credits. So,

it is a win-win situation. So now when we have covered the difference between an API model and a local model, I have some action steps for you because as I

said the best way to learn all of this is to set it up yourself and to try it yourself. So if you want to follow this

yourself. So if you want to follow this video, go to fall.ai or wavepeed.ai, put some credits into there and try to use some image models. It is the easiest

thing you can do. You do not have to set up any dependencies. You don't have to rent a GPU. You will just go to that website. You will choose an image model.

website. You will choose an image model.

You will input a prompt and the image will get generated. And step number two you can do is to try to set up a comfy UI either locally on your own computer

or on a cloud GPU. So part number six is a confi. What even is confi?

a confi. What even is confi?

Comfy is a software that you can use to run local opensource image models. It is

a software that looks somehow like this and there you can input the prompt, you can input the negative prompt and you can load the models that are open source

that you have downloaded. And look, I'm going to be honest with you, Confy UI is quite a complicated stuff. like there is a lot and a lot of things this image

that I have here that is a screenshot from confi is like I don't know 2% of what there actually is to learn so you

do not have to learn everything in confi to use confi like you have to learn the basics like you have to know like there is a model you have to choose the model there is a lura model that you can

choose there is like the prompt that you can type but how all of these things work like what is a sampler, what is a latent image like it's good to know but it's not essential. The most practical

thing that you can do is to set up comi is to download some workflows from our discord and try to use them. Try to set them up, try to play around with that stuff and you see what happens. So what

can be the practical use of confi? Well,

you have found some models on the internet that are open source for example one 2.2. So you will go to confi and you can either create your own workflow which I don't recommend because

there are smarter people than you or than me and they have created the workflows already. So you learn how to

workflows already. So you learn how to input workflows into config UI. You

learn how to download custom nodes. You

learn how to download the workflows, how to use them, how to download models and you are going to run it. That's it. You

do not have to learn the details on how to connect everything because it is absolutely not necessary. And as I said, you can either run it locally on your

own computer or on a cloud. And if you want to learn Comfy UI in more detail in the practical way that I have just described, there is a video on my

channel covering just that. Okay. So now

you know what is an image model, how to start, what is the process, what is config, what is local model, what is API model. So now you have to decide what

model. So now you have to decide what model are you going to use. There are

tons and tons of models. These are the most used in the community and the evolution of the models is quite simple.

Back in 2022 or I'm not sure what year was that, there was stable diffusion 1.5. Then there was better stable

1.5. Then there was better stable diffusion. Then there was flux and now

diffusion. Then there was flux and now the main models we are using to generate AI influencers is one Gwen or Cedream.

It is pretty much the three most used models. If you want to generate videos,

models. If you want to generate videos, the best models to try is either Clink, one or C dance. If you want to do a

video to video to copy to copy some dancing Tik Tok that you have found, you use one animate. As I said, the purpose of this video is not to show you how to

do everything because it would take 50 hours, but it is to give you the entire overview and the context in between things so you know what to do and what

to learn. If you want to imagine like

to learn. If you want to imagine like what even is an image model, just think of it like this. Like there is an image model called Gwen, right? And you can

think of it as a collection of millions or billions of images that are puted into computer language and that they are somehow trained like that is all you

need to know. So now when you know the differences you decide to generate some images but you cannot get a consistent face. Well, there are many methods that

face. Well, there are many methods that you can do to achieve a consistent phase and we will go from the easiest to the hardest. The easiest thing that you can

hardest. The easiest thing that you can do is to use image edit models aka Cream. What you can do is that you input

Cream. What you can do is that you input a reference image and you type a prompt like image of this girl sitting on a chair and then you either get the same

person or a very very similar person and then you can use that to create a data set but we will talk about that later.

Step number two is that you can use face swap but most of the online face swaps are not actually good. Method number

three that you can use is a good prompting. You can create a prompt which

prompting. You can create a prompt which will describe the girl itself. So if you type a prompt an image of a 20-year-old Eastern European blonde woman with high

cheekbones, sharp jawline and stuff like that, you can get a quite similar faces.

And the last easiest method that you can do is an image to video. If you have an image of your girl and you want to have more images that you can use to your

training data set, you can put it into a video model and you say like that the girl is turning around. Then she will actually start turning around and you

will get multiple images of her. But of

course, the best and real method for the best results is training your own Laura model and that is the thing we are going to be covering right now. So, what the

[ __ ] even is aora? It is the demon that everyone is scared of. The acronym means low rank adaptation. And if you want to understand the math behind it, you can

look at these pictures. But I have absolutely no idea how the math works.

So, I'm going to explain it to you in [ __ ] terms. So imagine that there is an image model and the image model is

trained on 10 million images, right? And

the image model contains also blonde girls, but the model contains this picture and it is an image of a blonde girl. Then there is a different woman.

girl. Then there is a different woman.

Then there is a different woman and there is this different woman as well.

And all of these images are images of a blonde girl. So Laura training is

blonde girl. So Laura training is basically that you take the entire image model then you add like 20 or 50 images

depending on what model are you using.

You add the images of your girl that you want to train the model on and then you give her a keyword that probably wasn't used in the model. Right? So you do not

caption these images like an image of a blonde girl because it already exists in the model. You caption it by Elizabeth

the model. You caption it by Elizabeth HH, which is most likely not used in any of the 10 million images that the model is trained on. And if you do the

training successfully, then every time you type an image of Elizabeth sitting on a chair, you will get an image of this girl. Sounds kind of easy, right?

this girl. Sounds kind of easy, right?

Well, it's not because there are the four main problems. If we are using a Laura to generate images of your AI influencer, well, how can you even get

those first 20 images in the first place, right? It is the creation of a

place, right? It is the creation of a data set. We are going to cover that in

data set. We are going to cover that in the next steps.

How to do that actual training process.

Do you need to understand the math or how does it work? How can you use your newly trained model? And how to know if your training has been successful? And

before we get into the data set, remember that I said you have to caption these images that you input into the training model. Well, there are two

training model. Well, there are two types of captions. You can either use straightforward captions which looks like this. Elizabeth blue dress outside

like this. Elizabeth blue dress outside sitting on a chair or you can use descriptive captions and it is an Elizabeth and amateur photograph capturing a joyful moment blah blah blah

blah blah. When you are choosing which

blah blah. When you are choosing which captioning style are you going to use just go to chat GPT Reddit our doesn't matter and depending on the model that

you are training on you choose which captioning style are you going to use because for example if you are training on stable diffusion image model you want to use straightforward captions and if

you are training a one 2.2 two model you want to use a descriptive captions and there are two ways that you can caption your data set either you do it manually

so you take each photograph and you create a text file saying an image of this girl blah blah blah blah blah and you describe the image right or in today's world you can just use AI you

can use joy caption or other AIS like chat GPT gemini cloth and it will create the captions for you so now we are getting into the data set. Problem

number one, getting those first 20 images. Well, there are two methods.

images. Well, there are two methods.

Method number one, which is illegal and you should never do that, is to train on a real woman. So, you go to Instagram, you take some images and you use them as

your training data. So, every time you then generate an image with your model, it will look just like her. This is

illegal. This is identity theft. And if

you do not have an explicit consent of this person, never do that. So the only option left is to create an AI data set.

How can you do that? Well, the best way that is used today is to use image edit models aka Cream. So what you do first

is you create a reference image and then you use an image edit model like Cream and you prompt image of this girl sitting on a chair wearing stuff like that, right? and you get another image

that, right? and you get another image of the same woman. You see they are quite the similar and if you are able to get 20 images like that then you have

the data set complete. If you want to speed up that process then watch my previous video where I have created an air table where you can do just that

using 100 prompts at once. So let's say the data set is created and now you want to get to the actual training where there are many tools that you can use.

The easiest one is fall.ai and the hardest one to use is a diffusion pipe.

So the fall.ai trainer is pretty straightforward. You go to this website

straightforward. You go to this website and then you just input your training data. You go to additional settings. You

data. You go to additional settings. You

use synthetic captions and then you hit train. and then you get your first Laura

train. and then you get your first Laura model. It is the easiest method, but you

model. It is the easiest method, but you have the least amount of control over these results. So, I recommend doing

these results. So, I recommend doing that if you are a complete beginner and you want to train your first model. But

if you want to get more advanced, I strongly recommend that you use all of these other tools. The next tools that you can use is either an AI toolkit from

Ostris or Musubi Trainer. These tools

have much more control over the results because you can choose custom settings, but they still have a graphics interface. So you can upload those

interface. So you can upload those images here, you can name it here, you can put the trigger word here. But if

you want to get the best possible results, then I strongly recommend you try to set up a diffusion pipe and you train your model here. It does not have

a graphics interface.

So it means that if we go to diffusion pipe GitHub, if you install it, you have to type these commands into the command line so you can actually run it. I'm

going to be honest with you, I have an IT background and when I first started to learn how to use diffusion pipe and when I wanted to train my own model

using diffusion pipe, it took me three [ __ ] days before I got my first successful training run. And it was three days of pain of putting errors

into JG GPT and it just didn't work until it worked. The reason why you should use diffusion pipe is this thing.

You have a complete control over the training process so you know which checkpoint to use. Actually, I'm going to show you the real results I had. And

as you see that there are some steps during the training process, right? The

only thing that a step means is how many times did the Laura training tool looked over on the data set images. The reason

why is it important is that if it does not look enough times on your data set images, the model will be undertrained.

So even if you use your keyword of Elizabeth, it's it sometimes may generate your girl, but many times it going to it's going to generate something completely else, right? But if

it looks too many times on your images, the model is going to be overtrained and it's going to look [ __ ] up. So this is the reason why you can use an eval tool

like this. And you can see exactly when

like this. And you can see exactly when the training process has reached the bottom. And if we look here and you see

bottom. And if we look here and you see that around 250 steps, it is where it is in the bottom. And it is the checkpoint

that you should use. In my next video, I will try to show you the entire setup of diffusion pipe with the evaluation tool.

So you can do that training yourself. So

let's assume that you have successfully trained your Laura model and now you want to use it. Well, the only thing that you are going to do is that you are going to download your newly trained

model. You are going to either download

model. You are going to either download a workflow that supports Laura models.

For example, if we look here at the confi screenshot, you see that there is a load Laura model. So you will just choose the model that you have downloaded and you are going to use it

with your keyboard that you have chosen.

So, Elizabeth HH and then you just generate an image. It's that easy. If it

sounds kind of abstract to you, don't worry because soon I'm going to create the most comprehensive Laura training video that there is. Now, we are getting

to part number 13 and it is prompt engineering. Even if you have the best

engineering. Even if you have the best data set and even if you have the best trained Laura model, if you don't use good prompts, you are not going to get

good images. The difference between

good images. The difference between those three images is in the prompts.

This is the first prompt. This is the second prompt. And this is the third

second prompt. And this is the third prompt. And as you can see, it is a much

prompt. And as you can see, it is a much better realistic looking image. And

luckily, you don't have to create such prompts yourself because there are AI.

And if you go to our discord, there is a channel called free resources. You can

go there and you can choose any prompt generator that you like. The Google AI studio opens and you see that you can

just I don't know post this start typing and the AI model will generate the best prompts possible because if you look at

the conversation history it is pretty heavily trained and as you see it fought for a second and now we are getting these excellent prompts. So once you

learn the prompt engineering and you learn all of the previous stuff that we have covered then you are probably able to generate good images right but now

you want to create a not safe for work images because you have to put something on your only fence right and since now you exactly know what an image model is it is very important that the image

model that you are going to be using for not safe forward images is trained on porn if it isn't it is not going to generate consistent high quality

not safe for work images. So who would create such model right like there is not a commercial company that is going to create such highquality poor models

but luckily there is a community of gooners and community of gooners is on civitai.com if you go to that website you see that

there are many models and there are tons and tons and tons of not safe for work models so you can choose anything that you like. Okay, now we have covered all

you like. Okay, now we have covered all of the image stuff and now we are getting to the video stuff. There are

three main methods that you can use to generate videos of your AI influencers.

You can either use text to video but there is literally no reason to use this one because you do not have such control over the video. It is much better to use

image to video because once you learn to create a good quality images then the video is easy because the only thing that you are going to do is to create an

image that is good-looking and you just prompt it. You say this girl starts

prompt it. You say this girl starts swaying her hips while brushing her hand through her hair and that's it. You have

a video generated. There is nothing else to do. If you want to use local models

to do. If you want to use local models to generate imagetovideo videos, you can use these models. And if you want to use API models, you can use these models.

These are the most used models in our community. Also, there is a videotovideo

community. Also, there is a videotovideo method where you can literally download a dancing video from Tik Tok. You can

replace the character in that environment. They input your Laura model

environment. They input your Laura model for better quality and then you get the same dancing video but with your AI influencers. The model that is used for

influencers. The model that is used for that is called wall animate. And one of the last things that we are going to cover is how can you achieve realism. If

you look at my picture, you see that the skin is quite detailed and it looks very realistic. If you look on the other AI

realistic. If you look on the other AI influencers like this, you see that the skin is not detailed and it just does not look good, right? Well, what are the

methods that you can do to achieve these results? Well, step number one is don't

results? Well, step number one is don't aim for perfection because AI influencers exist already for a while.

So, don't prompt it like 8K, 4K, ultra realistic, hyper realistic, and stuff like that.

Instead, use these prompts. Do amateur

photo, heavy HDR glow, deeply crushed shadows, imperfect skin, so you get more realistic looking images. Method number

two is a brute force. Look, in the end, it is just a numbers game. So if you have a [ __ ] settings, if you have a [ __ ] model, if you have a [ __ ] prompt, even

then you can get a good image, but you are going to have to generate 1,000 images and one of them will be good. But

if you got good settings, good prompt engineering, very well trained Laura model, then if you generate 10,000 images, probably 500 of them will be good. So method number two is a brute

good. So method number two is a brute force. And method number three that

force. And method number three that every AI influencer on Instagram that is looking very realistic is using is just a post-processing. You will take your

a post-processing. You will take your image, you will put it into Photoshop or your favorite image editor and you will like add imperfect skin. You will

distort the contrast. You will just [ __ ] up the image a little bit so it looks more realistic. Method number four is

more realistic. Method number four is skin detailers. There are many confi

skin detailers. There are many confi workflows that focus just on that that they will add the blemishes, they will add the imperfection on the face of your AI influencer. And method number five is

AI influencer. And method number five is that you have to realize that garbage in equals garbage out. If you have a [ __ ] data set and [ __ ] prompt, you will get

[ __ ] results. If you create a high quality data set, you train it very well and you have very well- created prompts, then you will get good results. It's

simple as that. So how do you start?

Well, I recommend you join our Discord, watch the tutorials. Every time you watch something, try it yourself because it's not going to be as easy as it looks. You have to fail many times. You

looks. You have to fail many times. You

have to repeat and every time you get some error message, it doesn't work.

Just use AI. We have chat GPT, we have Gemini, we have cloud code, it doesn't matter. Input your error messages into

matter. Input your error messages into AI and I promise you that you will learn much more. So to recap this video, this

much more. So to recap this video, this is all of the things that we have gone through and now you have the context of the AI influencer creation. You know the

basics. You know what to do, what do you

basics. You know what to do, what do you need, how does it work, what is the process, how to start, how to set up confi, what models to use, how to achieve consistent phase, how to do

Laura training from start to end, and how to use prompt engineering, generate not safe for work images, generate videos and stuff like that. If you have any specific topics that you would like

to see covered in great detail, for example, the diffusion pipe or Laura training, write me a comment under this video and I will create that video. It's

that simple. Don't forget to join our discord which is completely for free.

There are free resources, for example, the prompt generator. There are many confui workflows that you can use completely for free.

If you have any questions, either write a comment below this video or shoot me a DM on Discourse.