AI Image Mastery for Marketers – Create Stunning Visuals Fast

BizatWeb
67 followers
16 Video Views·May 27, 2026
Are your marketing visuals falling flat? It’s time to upgrade your creative game! Discover how to harness the full potential of generative AI to produce hyper-realistic images that captivate your audience.
Want to get clients using cold email?
Start here 👇
https://app.outboundest.com/r3emGnyc
Use code: OUTBOUND (20% OFF)
🔥 Get the exact tools I use:
👉 https://bit.ly/bizattools
🚀 Grow Your Business & Master Digital Skills with BizatWeb
🌐 Website:
https://bizatweb.com
📩 Email:
[email protected]
📱 WhatsApp:
https://wa.me/212637648713
📲 Follow BizatWeb:
Instagram → https://instagram.com/BizatWebHQ
Facebook → https://facebook.com/BizatWebHQ
TikTok → https://tiktok.com/@BizatWeb
🔥 Business • Marketing • AI • Automation • Cybersecurity • Online Growth
AI Generated·May contain errors
Hello and welcome to our course on Generative AI. I'm really happy that you took this step
to join us and I'm sure you're going to be learning a lot throughout these whole
three chapters that we have for you. Basically, before we go even further, if you are someone
that just wants to get feed with knowledge, straightforward, a few minutes and start
generating high-quality, realistic imagery, just go straight to Chapter 2. That's where
we are going to be discussing everything that is hands-on related. However, if you're someone
that wants to learn the foundations of what makes AI able to deliver high-quality imagery
and independently, if it's now or two years from now, still be on top of the market, you
definitely want to stick in this introduction of the course. Before further ado, just some
quick information regarding myself. I have been a filmmaker photographer for 15 years now.
I've been working from many sorts of industries, from restaurants, fashion, you name it, all the
way to my own travel videos. You can see some of that also in my personal YouTube channel. I've
been working as a marketeer for almost 10 years now. AI has completely changed the way I approach
not only my art, but also my work, creating high-quality assets for paid ads and also for my
personal work that has really, really brought my work to the next level. The first concept that
we want to discuss is about you as an art director or a creative director. AI has this vast,
vast knowledge of everything that has already been created, that exists in a world, and you need to
basically guide AI to do and deliver that that you are looking after. Think in terms also as a new
employee is hired at your company, and now you need to bring him your brand book, previous photos,
the sort of work that you're looking at. If you just take all this information, throw at him,
and let him decide, it's probably going to fail, right? That's the same thing with AI. You need
to approach AI from this perspective of, I understand what I want, I understand what I need,
or I have the references of what I'm looking after, and give it to him to then create that
that you already foreseen or that you already conceptualized, and he will be able to achieve
that. If you're just coming for random requests, like with the recent or with this beginner
junior employee that you hired, and you just give whatever and say, just give me something,
you're probably going to get something that you will not enjoy, or you will not like it,
or will not look good. That's the same thing with AI. So remember, first, AI cannot create anything
that doesn't already exist. It's just taking all of the information that exists,
and that is good information, bad information, garbage images, and good images. Now we are going
to be going over a list of things that we need to take into account whenever we are creating images
with AI. Those are the bare minimum information that we need to give AI for him to follow what
we want and deliver something that is realistic. The first of those is lighting. I would say it's
one of the most important aspects that we can feed AI to get the results that we are looking for.
You can just ask Chachibuki, actually, what are all types of lighting that exist in fashion
photography, for example, and he's going to tell you about paramount lighting, butterfly,
Rembrandt, so on and so forth. So I don't need to kind of go over each of these specifically. You can
also just feed AI with a reference and ask him what is this sort of lighting, and then you can
just copy and paste to your prompt. What is important for us to understand is there is a
quality of the lighting that we want to show in the image that we are creating, and some of those
are going to give a feeling of real realism and some of them not. What do I mean by that? There's
a lot of filters now online, especially in Asia, for example, China, that use some filters that make
the skin completely smooth. If you ask AI to recreate something like this, it is not going to
look realistic because the image itself that is being shared online is not realistic by itself.
It lacks character, it lacks texture. There are other things that make us identify that this might be a
real-life image and not an AI-generated image, but that lighting doesn't help at all because that
whole effect is created digitally. So, for example, a flash photography. Everyone knows
flash photography when they see one because of the high marks that you have, like the spots of light,
harsh spots of light, especially like nightclubs and things like this. And when we see that sort of
imagery, we know that there is someone behind the person taking the photo. That's why also when we
use AI to create this sort of lighting, it looks really realistic to us, almost indistinguishable
of a real-life image or not, because that sort of light makes us or tricks our brain to see as
someone was behind that person and took that photo. That's a really important concept for us.
So, whenever we are trying to create something realistic, we need to identify does this image,
this reference image that I want to use to create this, looks realistic or this is like a filter
that already makes the image itself not realistic by itself. And which sort of lighting I can use
then to achieve this goal of creating something that looks more realistic. The reason why I would
say in the beginning that Midjourney now is not possible to create the same type of imagery as
Chachapiti is, is because Chachapiti at this moment listens to your prompt way better than
Midjourney. So, if you ask flash photography like this, this, this, this, this, you get an image that
resonates with the prompt, while Midjourney still doesn't do that as perfectly. There are ways to
achieve that on Midjourney, putting some image references, creating some sort of character, some
sort of style that is possible, but it takes a lot of time. So, that's why for now I would say stick
with the generations from Chachapiti. All right, so another important concept that we need to understand
is the concept of perspective and angle. When it comes to perspective, not only we are talking about
where you as the art director, creative director wants to position your camera, so angle, perspective,
they work with each other to create some sort of feeling, some sort of composition and the way that
image is perceived, but also which sort of lenses you are using, which sort of camera you are using.
I think camera we are going to go a bit further into the texture part, but we'll focus a bit more
on the lenses and also the position of the camera now. So, there's a difference, for example, I'm
using a 20mm lens right now and that's a really wide angle that allows this camera to capture
not only my computer here, part of my bed here on the left, myself, so from head all the way to
torso, a little bit of the background, nothing is blurry because it's really wide lens. You don't
need to know all these technicalities, okay? Like I said before, also for the lighting, you can ask
what are the sorts of lens that we have or I want this sort of style in my photo, I want someone
that looks, if they had bigger, what sort of lens I would need to achieve that. You can ask all these
questions to AI, of course, or also just use a reference and ask, but that will affect, for
example, if I'm in focus, my background is still in focus, if I'm in focus and the background is
out of focus, all of this will be influenced by the choice of lens, how we take the photo,
the settings on the camera, yada, yada, yada, but we don't need to know that, at least for
the purposes that we have here. What we want to understand is that where we are positioning the
camera, is it from the bottom, is it from the top, is it someone holding a cell phone, is it
someone having their photos taken, is it someone that is using a professional camera with a
telephoto lens, like taking a photo of an animal or is it a portrait of someone that is super
close to the camera, all of this is important. So, remember, when you are generating your prompt or
thinking about the prompt that you are creating, think about also the lenses that you're using,
is it a wide angle, is it a zoom lenses, are you positioning on the bottom, are you positioning up,
are you positioning straight on the eye level. We already talked about the lenses and now we are
going to be talking about the camera. We don't need to fit AI into a specific camera that we
want to use, that's not the goal. What we need is to understand that the camera itself is going to
give some sort of character to the image that is being created because each camera will have some
sort of texture that is added to the final image and without adding these imperfections that each
camera brings to the final image, it is going to look really digital or it's going to be really
AI generated because it lacks character. So, those things, there's a bunch of things, again, you can
ask AI but the most important ones are grain, chromatic aberration, vignetting, there's also some things
that happen, for example, depending on where the light is placed or what is the FPS that you're
using, like scan lines and other things like this, everything that is related to imperfections. This
is also important for you to take into account, especially grain. Grain tends to give like this
extra flavor of something looking a bit less digital in case of a digital camera against
an analog camera. There's also this part, but in general, what we want to take into consideration
is this sort of imperfections in our image that comes mainly by the camera and lenses that we
are using to take the photo in specific times of the day, specific conditions, and specific brand
and things like this. So, really important to also talk or add this sort of flavor into your prompt.
Let's talk a little bit about post-processing. This is also an important part of the image itself.
Most professionals, they will not just take a photo and use that as is, right? They will bring
to some sort of software where they're going to add the post-processing on top of that.
Usually, some sort of color grading, corrections in the skin, and things like this. So, it's also
important that we feed AI with the sort of post-processing quality that we want to be added.
In this case, it can be removing imperfections on the skin or adding imperfections in the skin,
making the image sharper, creating some sort of blurriness in the background, adding a specific
color or tone to the image. This is also important for us with AI. And we can separate, for example,
the color with this post-processing part, or you can combine them together. This is up to you,
but just remember that we also want to give AI information about the post-processing part.
What sort of style we want to add on the image that we already captured in real life. Okay, so color,
like I said, can also be part of the post-processing part of your prompt. But in this case, I want to
give a bit more emphasis on just on the color part, because we want to feed AI with the sort of
mood and art direction that we want for the image, not only in terms of the color that you have in
set. So, always remember your art director, creative director. So, you are taking a photo of something,
and that something looks and has the colors in a specific way. So, for example, my scenario here,
we have this, we have this wardrobe, and the white part is here. So, you can see that I didn't put
an effort on making a whole set here. I'm trying to make really like a raw sort of style video
that is being recorded in the same place where I work, so there is no art direction. But if you're
looking for something that is a bit more artsy, like some sort of shooting for your fashion brand,
then you definitely want to add these props and colors. We're going to talk about props separately,
but we want to think about the colors in the scenario. Is it warm? Everything in the image
is going to be warm and give more like this Latin vibe, for example, or it's going to be colder.
A lot of blues also in the colors, or more contrast between warm with yellow and blue.
All of this is important. So, remember, we need to be thinking about the color in the set and also
the color of post-processing. Another really important thing, and this comes more from the
3D world than from photography itself, because when you are taking photos of a person or a specific
scenario, everything is already there, right? It's already created, and you have certain things,
or you as a photographer, have certain things that you are not usually thinking about,
that when you are creating something on 3D, you need to think way more often,
and that also adds a lot to the realism of your 3D renders, for example, which is the textures.
There is a lot of textures, a lot of textures in everything that we have, that we are in contact
with. For example, this notebook, there's like this texture here, and this texture here also on
the side, that it's more like a leather type of texture. The walls also have some texture,
has also some imperfections, metal, so on and so forth. So, if you were a 3D artist,
whenever you were creating objects and you were texturing the objects, you were giving a specific
texture to the image, and without a high quality, good texture, those 3D renders will look really
non-realistic, and you see that a lot, for example, in real estate. You see a lot of images of places
and spaces that lack character, because they make this really perfect texture that doesn't
look realistic, and that's the same thing with AI. We need to feed AI with the sort of textures that
we want, especially if we want to create realism. So, if we are doing a generation of a toilet,
for example, without the cracks, the graffiti, and things like this, it is going to look like
a 3D render. It is not going to look realistic. So, texture is really, really, really important,
and this is something that most photographers that are doing AI-generated image forget to add
to the prompt. Now, another really important part of our prompt needs to be props and accessories.
If we are talking about fashion sort of look or sort of generation, we want to say,
does a person wear earrings, a person, a necklace? What is in the scenario itself? Is there like a
red cup somewhere? Is there like a headphone in the table? All of this needs to be fitted to AI.
If you don't fit the sort of information, you're going to get something really generic,
no art direction. So, the same way as in movies, you have a prop designer, you also need to think
about that when generating AI. And I know it's a lot of information, it's a lot of things for us to
remember and take into consideration, but the good thing is that we can really optimize this
whole process with AI. For example, I have an assistant that I created on Chachapiti that
remembers all this information for me, and if I don't fit him with this, he's going to ask me,
Leo, what sort of lighting do you want? What sort of props should be in this scene? What sort of
color you want? And if I fit an image that lacks any information of that, it's also going to ask
me that. So, just remember that you, whenever you are doing the prompt, you need to fit the sort of
information. And you'll see that even if when I give to my assistant a specific image,
he's going to ask me about the texture, what props, and things like this. If I give the same
information that I give to the assistant to generate the prompt, I give as a prompt itself. So, without
using, for example, this assistant that is going to list all these important things that we need to my
image, he's just going to be lazy and not give to me at all. So, remember, all of the scenes that we're
listing here, they are important for us to add in the generation, even if we have an AI step before
the prompt itself. And finally, to summarize the whole process, there are also the more technical
things that we need to input towards our prompt regarding what is the ratio that we want. Do we
want landscape? Do we want vertical? So, also really important, don't forget that's nothing worse than
generating the perfect image and having it in square while you want it in vertical format.
Also, the mood. So, this is where you put the more generic things, what most people nowadays
just give straight, or the only information that most people will give to AI nowadays, which is
like landscape, a surreal fantasy style, and a person in the middle, for example. These are
information that is like the basics that people use. You still want to add in your prompt, just
don't forget everything that we've discussed in the previous videos. Before we start this chapter,
here is just a few of the results that we are going to be getting, and actually the ones that
are quite like your average prompt or your average generation, which is this tree that
we are seeing here, they are alright. They don't look good. So, if I'm being honest,
I will not use any of these. I think that they are just not even close to what we can create
nowadays. And then here are the final generation that we are going to be getting. This one is a
bit more stylized, so you see that we will create this one together in the last video of this
chapter, because it's a bit more modern, and it's also a bit more, looks more realistic for the
time being that we are, but this one looks really realistic if we are creating something for the
1950s, for example. Then, same thing, this one is basically a more advanced method without using
our assistant, and that we are going to be learning in chapter three. But this one, you see,
still we get some nice results, some nice texture, some realistic mood. Of course, the whole lighting,
the setting, the clothing, the whole thing makes it look a bit more like cinematic, so not as
real life as, for example, an iPhone photo, but it's still something that would pass as realistic
if we are trying to generate a cinematic image. So, that's basically it. So, this is what you
will learn how to avoid after finishing this chapter, and this is actually the type of imagery
that you'll be able to start creating. Let's start here by looking at some references. I also
created a few questions here that you see in the description of this specific video of
important things that we need to be taking into account, and you will also see how by just feeding
this information to Chachapiti, now, not always, is going to give you the best and most realistic
results, but we will discuss that in a few seconds. So, let's just choose one image here that we find
interesting. For example, this one from La La Land. Let's take this. Let's copy this image. Let's add
here, and let's use our questions here to analyze as Chachapiti analyzed for us, so we don't have to waste
time. If you don't want to use a reference, then, of course, just take all of this information here and
try to define yourself. So, for example, define like a photographer. If you're not a photographer
and you have no idea, just ask Chachapiti what are the types of lighting that exist, answer as
a photographer, or what are the type of textures that exist on the objects, answer as a 3D artist.
Okay, so it should be straightforward. So, let's do that now. Let's run it over, and we will get our
answer in a few seconds. So, here we already see the lighting, natural ambient lighting at golden hour,
mix of practical, warm strength with lamps, soft feel skylight, gives smooth skin tones, white
lampposts, and warm lights and contrast. He described the scene. That's not what you're looking for.
So, remember, that's why, if you want to learn more about this, go back to the introduction
chapter, but we will try to tackle all of this here. Texture, skin, smooth, slightly matte,
true but not true, it's not smooth, fabric, dress and shirt, cotton, light, polyester, soft molds,
wall fans, object color, this tends to be quite good answer, so woman dress, saturated color,
pink, off-white, okay, that's good, we can use this, post-processing, I'll explain why it's good and
why it's not good, post-processing, warm and nostalgic grading, that's also good,
visible imperfections, very subtle grain, no obvious connected relation, blah, blah, blah, blah.
Okay, but just because we don't see this here doesn't mean that it's not present. Like I said
in the introduction part, there are certain imperfections that the camera has, that when
we put to post-processing, we remove them, but for AI, we want to make it less digital
and more like real life sort of imagery, so we want to add things that were removed by the
colorist and the editor. Mood, intimate, romantic, this is fine, tends to also be good,
this angle and lens also tends to be quite good, so I'll keep that, clothing style also good,
image size is around this 2.3. Okay, this we can also change, because again, it's about what we
want, so really important here, like you were seeing, there are certain things that it's better
that we input, that we give, and that is going to allow us to generate a better image, so let's just
run it over here, create image, let's use the same information that was collected from this reference
and let's see what we get. All right, we received our image here, by itself is already quite good,
but take into consideration that he's using the history of the image that we put to generate
this, right, you see that it's almost mimicking the face information which was not added in the
prompt. If we create a new tab and we put the same prompt, we're going to get something completely
different. There's a few things here that we need to input to make this one even better, and we're
going to talk about that in the next video. All right, so here's the generation that we got on the last image.
Just for you to see the difference that we would have, we'll create a new chat here, and let's first
just go with, I want a couple walking down the street during sunset, hyper-realistic, let's do
two by three ratio, something like this. This is what the EU, this is what the regular,
this is what the regular user would input whenever they are creating an image, just so we see how
it looks without even starting with our nice little questions. Not bad, right, it's not following
because we didn't give any directions in terms of color and colors and other things like this,
but we can see it's AI-generated, right, it lacks some of the characters that would make us
believe that this is a hyper-realistic imagery. The background is quite good, the sun is also okay,
just the face scheme, the structure of the face is not there yet. So, although the ChatterPT will
become better and better, or any sort of image generation tool will become better and better to
deliver the final result as fast as possible, there are still things that he needs to decide
by himself that he most of the time forgets to add. Okay, so I'm going to copy here the
sponsor that we used before, and let's add on a new chat, and now let's generate. I'm going to
quickly add here that I want a hyper-realistic, hyper-realistic, realistic image, and let's see
what we get. All right, so here's the image that we got, it's a bit different than the one that has
more information from the previous image that we added. It still looks a bit weird, the face is not
super realistic, so we might want to give specific instructions if we want this to be
a bit different. It's not bad, I think a lot of users would already be satisfied with this,
and that's fine if you think this is more than enough to push whatever you want, but there are
still things here that we can improve to make it even more realistic, so bring it to the next level.
A few things, I'm going to copy this prompt here, and we are going to think together of what we
believe lacks realism. So here, hyper-realistic image, then the lighting. This lighting here, I think
it's fine to have the golden hour sunset lighting, but for me, this should be added at the mood part.
So let's see here, mode, like this, and you see also the twilight setting and the body language,
there's a bunch of things, so let's just leave it there. And the lighting here, what I actually want
there is a softbox outside the scene that brings a really strong lighting on the subjects
with Rembrandt lighting. Sorry, I wrote it wrong, let's put here Rembrandt lighting.
Basically, this is the thinking as a photographer that we discussed before. This could be butterfly
lighting, paramount lighting, it doesn't matter. You can just ask, it's going to give you a list,
or you can ask again, in this case, because it gives the overall lighting of the scene, but we
want the lighting as a photographer. The photographer is lighting the subject. The lighting of the scene
is more like the mood, the scene. So here, we want to input our manual, our way of thinking in this case.
Texture as a 3ds artist, we got nothing here actually, which is quite bad, but then let's look
again at our reference. If we were a 3d artist, what we need is some nice linen fabric.
Let's just add linen fabric on clothing, imperfections on the fabric, for example, and then
we have a lot of metal here and concrete, so concrete on the floor and background scene,
background scene, shiny metal, shiny metal with imperfection. This should already help us to give
some extra light. Then let's come back here, so skin, never smooth, we don't want smooth, we want
the skin to have texture, strong, strong texture on the skin, couple is around 40 years old, don't
make the skin smooth. We want to give this extra imperfections to make it even more real.
Fabric, that's also fine, but we kind of already talked about this on the texture,
wall, fence, fantasy, that's fine. This is also fine, pavement, we have then a little bit of
contrast to the artist, pavement, object colors, woman dress, actually color pink, this is fine,
this is fine. Actually, this is the 3d texture itself, so we see that some of the things
here are repeating, I just removed a few things, especially the skin, so that's the most important
part. Background, this is fine, lamp coating, glow, post-processing, warm, nostalgic, that's fine.
Let's add here, actually, subtle grain, photo taken by a canon 5d, or an iphone, let's go
taken by an iphone 14. That's it, that's fine, mood, that's fine, angle, that's fine, coating style,
that's all good. Let's take this again, push it here, and let's see what we get.
All right, so we have our image again, maybe I pushed a little bit too hard on the edge here,
in comparison to our reference, but you see how it feels a bit more natural, I still feel a bit
too artsy, too generated, right, and that it's mainly because this lighting on them is still
not the way we want to become super realistic. So you see that even by giving AI all of this
information, being really precise, we still are unable to get something that looks
extremely realistic, and that's actually when we are going to be checking out the assistant
that we created, that is able to help us to create something that is even more realistic.
All right, so here is the assistant, on chapter 3 we are going to have a really small chapter that
goes about creating these ourselves, so what we are going to do is take our inspiration image,
we're going to be adding it here, given to the assistant, and he wants some extra information,
they want to preserve this romantic vibe, yes, what color palette, pastel, that's fine,
3, editorial, then 4, same style. Now the AI is going to be generating the prompt for us,
don't know why he added an emoji this time, but that's fine, shouldn't change anything.
We cannot create inside the assistant itself, because the assistant still uses the old version,
however if there is a new update, and sometime soon they might have the new one,
whenever they launch the API, you can just use on the same assistant that you created,
but in this case we created a new chat, where we asked for this image to be created.
Now let's check a few things here, that might be different, so we have a harsh red flash at
twilight, so it's a good edit part, creating sculptural shadows on faces and clothes with
glowing vintage street lamps in the bulky background, so you see it's a bit more
experimental. Perspective and angle, frontal angle, slightly low to elevate the subject,
capturing the mid-step, confident posture and engaged expression. Frame style,
draw a robot, well you can stop the video and read the rest, but you see that there is a bit
more of explanation, and also trying to give a bit more imperfections, fine-tuning and other
things like this, so subtones, skin lines, lightness, flare from the lamps, soft pastel
gradient, yada, yada, yada, this is a bit more complete, right, and it was basically
without any effort from our part to generate the prompt itself. There, here is the final result,
something that is way closer to a hyper-realistic image, there is one small trick that we can use
one step further, if the images that we are getting here is not there yet, which I'm gonna
show in the next chapter. Alright, so after we generated, we are still not satisfied with the
final result, which was this that we got, and it still wants to add a bit extra details, a bit
extra layer to make it even more realistic, that's when we are going to be using this tool called
Magnific, it's quite expensive, so I don't think we have to use this, I would be satisfied with
either something like this, I would bring it to Photoshop actually to tweak a little bit more,
or try to generate even more variations until I get something that I'm satisfied with. You see that
here has a lot of these stylistic 1950s, and that's also why it looks a bit more fake than the
regular one, we will generate a more modern example again, but what we do here is upscale, we upscale
by four, at the current version of Magnific, I would go for optimize for portrait soft, otherwise
if we put hard, it's going to give too much detail in the face and it doesn't look good, also the
same for standard, so I'll just go portrait soft, maintain everything on the default and upscale,
and then this is before, this is after, so it does give a bit extra realist in the face, but for
example the dress here, you know, the fabric doesn't look too good to be fair, so I'll be satisfied with
the before and just play a bit on Photoshop. Now, if we look here at everything that we generated, we
have firstly the one that is like the lazy prompt, which it's okay, we didn't add all the information
that we needed in terms of colors, we just asked you to create something for us, we did, but you see
the face just looks like old-school, mid-journey style, just looks weird, like it doesn't look good.
Then here we have the one with just the questions that we asked, it kind of gives us a good
result, but still it looks weird, right? The face, the structure, the lighting, it's just not good, and
you see also that there's no shaping of the face, so it's kind of like if you were a lazy photographer,
didn't bring your lighting to the set and just took a photo like this. Same thing for here, this one,
the face still looks a bit strange-ish, but it looks a bit more like the actors, it lacks some
texture, it lacks some extra character or extra realism, especially the lighting again, and the
texture here on the sides, but the image reference was also lacking quite a lot of texture. Then we have
here the one that we optimized based on writing a few extra things, and you see that compare the
scheme of this here, this one here, with this one here, this could pass as a believable edited
photo, so something that we took in real life, and then just edit a lot in Photoshop. And this is the
last one, still looks a bit edited, like an advertorial from 1950, but if you put this on a journal, like a
mock-up of a journal, I'm pretty sure that this would pass as a real-life photo from that specific
time. So now let's do a bit less of a stylized sort of image, because when we're using a movie
we are going to get the sort of time period or mood that the director wanted, or the cinematographer
wanted to achieve, so let's find something that looks a bit like a regular photo, right? We could
use Instagram, in this case I'm just searching on Google, see if I can find something that
looks a bit more like a regular day in a life, for example this photo here, right? We know that
this was not a professional, just a regular photo. So let's take this one, let's
push to our assistant, let's edit here, let's wait for the prompt to be generated, he's probably
going to ask a few extra questions for us. Okay, so we uploaded the image here, color, accent,
keep the same as reference, mood theme,
real life, and we want two subjects, two subjects, same thing. Now let's wait for our prompt to be
generated, there we go, I'm going to copy it here, and I'm going to create a new chat. Again,
this version of Chatchabitty doesn't allow us to create directly with the assistant,
with the new image generation version, so let's just wait a little bit now.
All right, and this is the image that we get as a result. You see, it's extremely realistic,
it looks nice. Here, of course, we're using the flash, this is something that in the introduction
we talked about, that gives like this feeling that there is an actual photographer behind the
camera taking the photo, but we can change that, we can ask for a more soft light, Rembrandt style,
fashion style, and we're still going to get something that is quite nice and realistic.
So there you go, you see how you can actually leverage this assistant that we are going to
be creating on the next chapter to create something even more realistic. All right,
so this is the final chapter of our little course, and we are going to be learning how we can create
this really nice assistant that is going to allow us to be way faster whenever we are creating
some hyper-realistic images. So the way to do that is, of course, you need the plus version.
If you don't have the plus version and you don't want to pay for the plus version, you can use
the reference for the assistant that is here, that I'm going to go over in a few seconds,
to basically tell behave as an assistant, and then you push all this information, and then you start
as you were starting with the assistant like we saw in the past few videos. So here what we're
going to be doing is actually you're going to come here in my GPTs, you're going to create a GPT, and
then we are going to configure, you're going to name it anything you want, so realistic image
generation. You're going to give a description here, so you an artist art director creating
high quality real life images, something like this. You can put whatever you want here to be fair.
Then we are basically going to be copying and pasting these instructions, okay, and I'm going
to go over this. Here is everything we actually already discussed about with a few predefined
things that allows us to always go towards the more realistic version, and then we can tone down
and change as we start getting generations. So your visual prompt assistant trained to turn user
inputs including scene description or reference images into cinematic hyper-realistic vertical
flash photography prompts. Here we can also just change to realistic vertical Rembrandt lighting
photography prompt. It's up to you. I like the flash ones. Like I said to you, it gives like
this feeling that there's a photographer behind the camera, suitable for AI image generation tools.
Your output always follows a specific structure that includes camera format lighting. Then here
I'm putting always, but you can change that also if you want to do more like vertical. It's up to
you or you can just remove and always put a new prompt. I just like to get what I'm looking after
the most as easy as possible. Harsh flash, sculptural shadows, then perspective angles,
framing style, surface, post-processing, color logic, accessories and props, overall mood and
narrative. Here you could add as many as you want, but in this case there is a few things or a few
settings that tends to work quite well, and just by providing this first input whenever you show a
image or something like this, the AI is already going to give you most of the information that
you need. So my recommendation is that you just copy and paste and keep it like this.
Then your job is to ask clarifying questions if needed, color scheme, mood, visual reference,
subject, gender, clothing, settings, time update, style, all of these. So you see every time I add
the prompt or add the reference, it does ask me some questions. Then never invent randomly,
it's really important. Always build from what the user gives you or asks for. And then lastly,
example user says, I want a woman in a weird surrealist scene with red and green stuff and
a plastic duck. You return, which build prompt that includes all visual elements, angle, light,
composition, materials, humor, blah, blah, blah, blah, blah. Basically, this is the summary of
everything that we want, and it's now automated that we can just use whenever we click here now,
create, we have our assistant created and ready to be used. So I hope you enjoyed the course.
Don't forget, you can still join our discord or paid discord where you're going to be getting
a lot of updates, where you're going to be getting a lot of updates every month. And happy generations.
