Hello everyone. This is Professor

Patterns and in this video we're going

to be going over the new GPT image one

model that's now available through the

OpenAI's API. Now this is a really

powerful model and I've been using it

for so many different applications. Now

one way I've been using this is I've

actually created a pipe function on Open

Web UI meaning that I can now use this

model directly on here. Sure, I can use

it to generate images of an apple. But

here's another image of this person

who's staring off into the distance. And

you can see it's incredibly realistic.

Um, I also have another image over here

of a warrior priestess. And now I can

even compare this to what I have on chat

GBT. And chat GBT actually doesn't come

close. Even though it is the same model,

I do have a lot more insight into the

parameters and I can tune a lot of those

parameters as well depending on the type

of uh output that I would like. So

that's one way I've been using this

model. Now another way that I've used

this API is I've simply created my own

sort of web app that's running locally

on my computer. It's an AI photo booth.

I'm going to share the entire codebase

with you. So you can even try it out for

yourself. Uh maybe you can create your

own sort of photo photo booth and

variations of it. Now it's a very basic

application. All it does is that I can

either take a picture or I can upload an

existing picture. Let's say that I take

a

picture. Now I can say well turn this

into a Disney character. And then I'm

going to hit transform image. So it's

going to transform this existing image

into a Disney character variation. Now

over here I also have another variant

where I actually uploaded an image. So

this is the same exact image that I've

been using for um every single one of my

YouTube thumbnails. I uploaded the image

and I said turn this image into an anime

warrior character. And then this is what

it ended up getting me. So I can

download this image now. And then if I

take a look, this is what the output

looks like

currently. So you can even create your

own ideas. For example, where users can

go in and say it's like a variation of

this app. A user goes in, takes um a

couple of pictures, maybe some of their

front face, some of them looking at the

side and the other side, and then your

app essentially generates like a um

LinkedIn headsh shot or actor type head

shot or something for them. So those are

a couple of different variations of this

app that you can kind of turn this into.

Um, if you take a look over here at the

Disney character, this is what it

actually generated. Uh, this was the

base image and then this is what the

output looks like. Now, the main thing

that's powering this is the OpenAI API.

Now, to actually get this OpenAI API

key, you will have to create an account

on OpenAI. And then once you created

this account, you will also have to

verify your organization. Meaning that

you have to upload maybe like a driver's

license or something and then upload a

picture of your face and it verifies

that you are that person. Once you do

that, it it is pretty much

instantaneous. It's not like you have to

wait or anything like that. But once you

do that, you will then be able to use

that API. So then all you would do is

just go over here to API keys and then

ask it to generate a new API key for

you.

Now, yeah, sure. Sure. Okay. Well, at

least it's not showing all of my API

keys. But this is what you would do is

just create a new secret key and then um

you can copy and paste that key put it

into whatever application that you

wanted to whether it's open web UI or

your own apps. Now if you go over here

to the image generation API sort of

overview it gives you a full idea on

like all the things that this sort of

API is capable of. So first we can use

it for generating images. So this is a

very basic one just use it for image

generation. And this over here we can

see the code in either Python or curl or

JavaScript. So say for example Python

all you would do is you can simply copy

this code put into a Python script

execute it and it's going to run it's

going to help it's actually going to

create uh these images for you. The

other thing that you can also do is edit

images. So over here it's showing us an

example where we have 1 2 3 four

different images and the prompt is you

know collect or put all of these items

into one basket and then that's this is

the output. So these are four input

images and this is the final output

image. It also gave us the code that we

would use to do something like this. I

actually used parts of this code to

generate the app over that I had shown

before uh where it generates variants of

an existing image. There's also other

things like edit an image using a mask.

So like in painting for example where

you have this as the uh base image and

you can uh say okay well this part of

the image maybe I want to add a flamingo

or something. So you can specify what

part of the image and then based on that

it will add in the appropriate location.

So there are so many different things so

many different um characteristics that

you can take a look at. Um the other

place that I also recommend go uh taking

a look into is the image API. So this is

a little bit more of a in-depth

documentation. So it shows you all of

the APIs that you'll need for create

image uh which is going to be this one.

You can look at the Python code or curl

or NodeJS code for example. Um you can

see that there are a bunch of different

parameters. We have things like uh the

prompt. We have things like the

background the model. So moderation. So

any content moderation that you want

either you can have it as low or you can

just have it as the auto. and auto is

usually going to be the default value.

Um, you can specify the output format if

you want it as PNG, JPEG or WEBP format

or or anything like that. The quality

now well this is going to be really

important because this directly ties to

the cost of the model. So sure we can

use something like uh medium, high or

low or the GPT image one model but then

the pricing actually for these models

does end up being a little bit more on

the costly side. So, I've actually been

using the these models for a while. I've

actually generated like so many

different images so far and total cost

$8. So, it's not like it's breaking the

bank or anything like that. But, um you

you still want to be careful. Imagine

that you create this app and you just

say, "Okay, you know what? I just want

to have everyone around the world uh go

ahead and start using this app of mine

for free." Well, uh, this is me

generating I think about, um, no, it's

not 82, but it's, uh, close to about 30

or so images when I was actually testing

everything out, and that costed me about

$8. And these are actually going to be

the high quality images. So, that's the

main sort of difference over here. So,

the entire API reference for all of

these things are available here. The

only thing that you'll have to do is

just verify your organization um

generate a new API key and then you can

use this as a reference for whatever app

that you wanted to create by yourself.

So in my case the what I did was I

simply u copied this entire

documentation and I put it onto VS code

client and this is what it gave me. I

just simply went over here and I said,

"Hey, you know, can you convert this

into a web application and it did it did

all of those things for me." So, if you

haven't watched my series before on um

you know, vibe coding or AI based coding

or anything like that, I'm going to link

that in the description. Um that will

show you exactly how I'm not really a

coder, but I've coded all of these

things from scratch. Sure, there there

could be some flaws and security

vulnerabilities, but I've made sure not

to include any of my API keys in the end

variable. And just for safety, I will

still purge all of my API keys after

this video. Now, within Open Web UI,

that's the other place that I've

actually put this into as a pipe

function. So, if I go over here to my

admin panel and then select my functions

over here, the image gen, that's the

function that I created. I also have one

for variations of an existing image. Um

but the image gen this takes in a list

of different things as the input. So for

example over here I can provide it um a

lot of these different factors and I'm

also going to upload this as well. Um

the idea here is that if I go to my

valves I can okay now I definitely have

to purge this API key but I provided the

API base URL the model the size like

what if I want 1024 um or you know for

example 1024* 1536 and then the quality

for example if I want high quality low

or medium quality uh the output format

what I want that as and the moderation

whether I want auto or low for example

and then you can hit save And remember

that this is a pipe function. So if you

want haven't watched my video on pipe

function before, I'm going to link that

in the description. But basically a pipe

function is a model that you can choose

on the open web UI interface. So you can

simply choose this model and then ask it

to generate an image of a cyber punk

car, cool ghoul, weapons, guns, etc. And

then this is now going to be sending

this to the OpenAI API for image

generation. And then it's going to be

returning that particular image back.

And there we go. So, uh, created this

image of Cyberpunk, cool car, weapons,

and guns. So, it's interesting that it

selected this car. That's nice. Well,

either way, you get you get the idea.

So, what I'm going to do is um, as soon

as this video ends, I'll make sure that

I push all of my changes to the GitHub

repository. So, you can go in and you

can start using the app that I've also

created. Maybe if you wanted to like

play around with it, I'll make sure to

include a readme file as well. Um, you

can go in create like different variants

of this app. U, the only thing that

you'll need to do is add the API key in

the environment variable and then

everything else will just run um fine.

So, if you do create a multi-million

dollar SAS business, then please

remember to tip Professor Patterns. Uh,

but that's it for this video. Thank you

for tuning in. I will make a more

in-depth video as well on some other

things that I am thinking about doing

with this API. But uh for now, that's

it. Thanks for tuning in. I'll see you

in the next one. Goodbye.