12 ChatGPT Image Tricks That Will Help You Become a Pro
By Rick Mulready
Summary
## Key takeaways - **8-Element Prompting Framework**: Include eight elements in prompts: subject, composition, style, lighting, color (with hex codes), mood, details, and context for reliable images. Use negative prompting like 'no people' to exclude unwanted elements. [01:16], [03:14] - **Brand Board via Memory**: Create a brand board using the 8-element prompt with your brand variables, then save it in ChatGPT's memory to reference for consistent future images. [04:57], [05:37] - **Image-to-Prompt Reverse Engineering**: Upload a liked image and ask ChatGPT to 'describe the prompt that could recreate this image' to capture its aesthetic for your own recreations. [06:00], [07:03] - **Iterative Chat Loop Editing**: Treat ChatGPT as a live art director by continuously giving feedback in a loop, like 'raise the camera, punch up the teal' or edit specific parts until satisfied. [15:37], [16:30] - **Multi-Scene Storytelling Variations**: Prompt for multi-scene narratives like 'frustrated entrepreneur, then organizing, then confident presentation'; use OpenAI Playground slider for up to 10 variations at once. [18:12], [19:09] - **Organize in ChatGPT Projects**: Use ChatGPT projects for image workflows like Instagram ads to save prompts, remember hex codes and brand colors without retyping. [21:18], [21:48]
Topics Covered
- Eight Elements Systematize Brand Images
- ChatGPT Memory Replaces Design Team
- Reverse Engineer Prompts from Images
- Iterative Editing Treats AI as Art Director
- Playground Generates Image Variations
Full Transcript
JT GBT has changed the game when it comes to generating images, but most people are not using it to its fullest potential. Now you can replace a design
potential. Now you can replace a design team worth tens of thousands of dollars a month with ChateBT if you know how to use it right. Not to mention, you can now avoid additional software
subscriptions, all those $20 a month subscriptions as well. That's why in this video, I'm going to share with you how to craft prompts in the Chat GPT40 image model that gets you the kind of
images that you want. examples of
different image use cases and one of the fastest ways of generating multiple variations of an image which you cannot do inside of ChachiPT itself that will
make your workflows so much faster. If
you're new here, my name is Rick Mul already and for the past 11 and a half years, I've run a multisefigure online business helping other online business owners grow and scale their business.
Today, I run a community called the AI playbook where I help online businesses leverage AI so that you can streamline your business and increase profitability in the process. I'll link to the
community in the description below. Now
for tip number one, when it comes to reliably generating the kind of images that you want in Chat GBT, not surprisingly, it comes down to clear specific instructions in your prompt.
And I like to include eight different elements whenever I'm prompting chatbt to create an image. Now actually I shouldn't say that I don't I don't
always include eight elements but if I know that I want to get very specific consistent results I will include these elements. So the first element is going
elements. So the first element is going to be your subject. What do you want the image to be of? So describe the main subject super clearly. What kind of appearance are they wearing? Clothes if
it's a person. Uh what's the pose, the expression? Is it a castle? Is it an F1
expression? Is it a castle? Is it an F1 race car? Whatever it might be. This is
race car? Whatever it might be. This is
going to be the subject of your image.
The next thing being the composition.
This is where you want to specify like a camera angle, framing, arrangement of the elements, uh that sort of thing. The
next one's the style. Do you want the uh image to be any specific kind of style like realistic cartoon, cyberpunk, etc.?
Then if you want lighting to be included, specific kind of lighting to be included in the image, this is where you want to include that like soft morning light or sunset or something like that. The color, this is where if
like that. The color, this is where if you have brand colors, so if you have like hex codes of your brand colors you're going to want to make sure that you include those hex colors in your
prompt. The next thing is the mood. Do
prompt. The next thing is the mood. Do
you want it to be like emotional or energetic, mysterious, that sort of thing. The seventh element is going to
thing. The seventh element is going to be any kind of details that you want included in the images. This is where you want to highlight unique features or elements that you want to included or to avoid, which I'll tell you more about
that in just a second. So like props accessories, a coffee cup, uh something like that, facial marks, whatever. And
then the final thing, super important you want to give it the context. So
describe the setting, describe the time period. Is are you creating a a YouTube
period. Is are you creating a a YouTube video thumbnail, describe the context of why you want the image created? Now
within the structure of these eight elements, like I just mentioned, a good practice is to use negative prompting.
So, for example, don't include or no use the word no. So, if you're drawing a picnic scene, for example, you can put in like no people, no birds, uh no
blanket or whatever it might be. So be
really really clear on what it is that you want in the image as well as what you don't want in the image. And here's
where it gets really interesting. Once
you understand this framework, you can totally systematize your entire visual brand because you just keep including these elements within all the prompts
where you want to create images and then you just start saving these prompts.
Even better, and I talked about it in a recent video here on the channel, a pro tip would be to use ChachiBT's extended memory to your advantage. So, for
example, you could say uh you could give ChachBT your hex codes for your brand colors and you could tell Chachbt to remember your hex codes as your brand colors. So, the next time you want to
colors. So, the next time you want to create an image, you could say using my brand colors. It's all you'd have to do.
brand colors. It's all you'd have to do.
You don't have to retype in the hex codes. Or if you have a hero image that
codes. Or if you have a hero image that you like to include in your brand elements for example or creating images you could tell chatbt to remember that
as your hero image and then you can just reference that whenever you want to create a new image. All right, the next thing I want to show you now that we have our prompting techniques for creating consistently reliable images with ChachiBT, let's create a brand
board because a brand board can drive the future look and feel of all your brand images. So this is the prompt that
brand images. So this is the prompt that I like to use and it simply includes each of the eight elements that we just talked about. So again, it's got your
talked about. So again, it's got your subject composition style etc. and the input variables for each one of these sections of your prompt. Okay, so
now let's actually create a brand board from that prompt. And I'm going to put the variables from my own brand as input variables into that prompt, which is this prompt right here. I'm going to click on go, and we'll see what comes
up. All right, here's the finished brand
up. All right, here's the finished brand board that it just created for me based on the prompt I just gave it. It's
actually pretty good. I mean, it's got the colors, the fonts, examples of images that we could use. I could
totally download this and I could then go and upload this into Canva, for example, and use this as a brand kit in Canva if I wanted to. Or better yet, I
could actually ask ChattBT to save this brand board in its memory for me so that I can just reference it whenever I want to be creating images. is I could just say based on my brand board do this and
it'll remember it. Tip number three this is what I like to call image to prompt reverse engineering. So this is where you upload an image and then you ask chat to describe the prompt that
could recreate that image. This is great for when you know you're on the internet and you come across images that you love the aesthetic, you love the look of the image and you want to be able to
recreate that type of aesthetic or look in your own images. use this tip right here. So, I'm going to paste in an image
here. So, I'm going to paste in an image here. It's just a screenshot of an image
here. It's just a screenshot of an image that I found on Instagram. It's actually
an Instagram ad. And by the way, if you run ads at all, this is a great way if you consistently see the same ad over and over and over, chances are that ad is doing well and the person or company
is continuing to run that ad because it's doing well. Well, you can learn a lot from that. take a screenshot of the image itself, stick it in the chatbt here, and you can simply say something
like, "Describe the prompt that could recreate this image." And with that simple of a request, describe the prompt that could recreate this image. I could
be a little bit more specific there, but I wanted to see what it would do. Here
I have this detailed prompt that describes that image. So, I can use this prompt to recreate that image. So, let's
actually use the prompt and see what it does. All right. And here's the image
does. All right. And here's the image that it just recreated using the prompt that we asked Chat GBT for to describe that original image. Here's the one that
I just created. Here is the original. I mean, pretty darn close
original. I mean, pretty darn close right? Um, makes me really hungry right
right? Um, makes me really hungry right now. So, this is a great way as you get
now. So, this is a great way as you get inspiration when you're online to prompt reverse engineer a specific image or images and have it write the prompt for you and then you go and customize that
prompt for whatever it is that you want to do. Now, speaking of capturing things
to do. Now, speaking of capturing things that I come across online, before I share the next tip for you, one tool that I use all the time in my workflows to capture the inspiration I come across
online or content that I want captured and and summarized is recall. recall
captures and summarizes anything that I watch, listen to, read, or open online.
And it saves it in one AI powered knowledge base. It instantly summarizes
knowledge base. It instantly summarizes YouTube videos, PDFs, articles podcasts, and more. I can also chat with all the content that I'm capturing inside of Recall as it saves it and
organizes it in its own smart AI library. I don't have to lift a finger.
library. I don't have to lift a finger.
It does it automatically for me. And a
surprisingly cool feature has been the quiz feature. So when I'm on my phone
quiz feature. So when I'm on my phone and I have a few minutes to spare, I just run through my personalized learning schedule that I've created within Recall so that I can actually
take in and learn from the content that I'm capturing. So if you want to check
I'm capturing. So if you want to check out recall, I put a link in the description below. Use the promo code
description below. Use the promo code recall Rick and you'll get 25% off Recall. So for image tip number three
Recall. So for image tip number three we're going to use what we've learned so far, especially the memory feature of ChatBT, and we're going to use the image model to create a logo. And I'm going to give it very minimal information, but
I'm also going to ask it to leverage the memory that it has of me. So I'm going to type in generate a minimalist owl logo in my hex colors. I'm going to click on create an image. And again
we're using the 40 model here. And let's
see what it comes up with. All right.
And here's my minimalist owl logo in my hex colors. This is very minimal. And I
hex colors. This is very minimal. And I
would say that it could be a little bit more colorful with my hex colors, but not too bad. So, you can ask ChatBD for any kind of file format that you want.
So, for example, if you want it as an SVG extension or a PNG file, you can just simply ask it for it, and it will give you instructions on what to do with that file once it gives it to you. All
right, for this next image tip with ChatBT, you can use the image model to create a brand avatar. I'm sure you've seen these brand avatars going around online. Well, we're going to create one
online. Well, we're going to create one right now. I've written a prompt to
right now. I've written a prompt to create a brand avatar in a specific style. So, now all I want to do is
style. So, now all I want to do is upload an image of me and it'll give me a brand avatar that I can use online maybe in social media or what have you.
All right. And let's see what it comes up with. All right. And here is the
up with. All right. And here is the brand avatar that it came up with. It's
pretty good, right? But just keep in mind that all the elements that are included in here, including the shirt.
So, like this is the image that I gave it. I can give it any additional
it. I can give it any additional instructions in the prompt to make any kind of changes that I want to create this brand avatar. But as you can see here, it did that in a matter of 90
seconds. Not too bad. All right. So for
seconds. Not too bad. All right. So for
image trick number five, now that you can create consistent branding in your images, let's talk about some advanced techniques for image generation with the 40 model. And the first thing we're
40 model. And the first thing we're going to do is use the image model to add elements or an element to an existing picture or image. All right. So
what I've done is I've just uploaded the exact same image that I used previously for the brand avatar and I just said edit this image such that the person is holding a coffee cup. So just to recap
this is the image that we are using and we are hoping that it's going to add a coffee cup to the to my hands. All right
and the image is generating the face looks a little wonky but it's not too bad. I think it said a good job in
too bad. I think it said a good job in the And there's the coffee cup. I mean
that is really cool, right? It looks
really natural in how I'm holding the coffee cup with a simple prompt. Taking
an image, edit this image that the person is holding a coffee cup. And here
it is. I would say the only thing that's a little bit questionable is the face on how it's done it, but it's not too bad if this is the original. So, what I could do on this, and I'll talk more
about this coming up also, is I could click on the image here and go to the little select tool and I could select the the face and then I can describe
what I want to change or replace or edit or or update. I could highlight the coffee cup and I could say make the coffee cup black or whatever it is. So
there's a lot of different ways that you can uh edit the images in addition to just giving it simple prompts which we'll do in the next one because with
the next advanced technique it's what I like to call multi-perspective generation. So this is essentially like
generation. So this is essentially like you're prompting the for the same scene to be from different viewpoints. So, I'm
going to put in a prompt here for a castle and a dragon. And so, I'm going to put this right in here. Actually, I'm
going to start a new chat and just going to put that right in here. I'm going to click on image and we'll see what comes up with because then we can use a little bit of iterative editing here, which
we'll talk again more about here in a second on getting different angles and multiple perspectives of this image. All
right. And here's the image that it created. It's pretty good, right? Like I
created. It's pretty good, right? Like I
sun's really cool. Detail is pretty cool. Love the dragon there. Okay. So
cool. Love the dragon there. Okay. So
now let's ask the image model here to make a couple changes. So I'm going to ask it to revise the image so that the castle is seen from above. All right.
With that simple tweak, this is the image that we just got. So as you can see here, now we're looking down on the castle and the dragon rather than up at
the castle and the dragon. Now, if you wanted to, you could put that multi-perspective generation request right within the prompt. So, maybe
you're getting back like three different images, the castle seen through a forest or from a moat of the castle, and then from above like we just did. So, you can put that into your prompt and Chhat GPT
will give you that same image from different perspectives. For tip number
different perspectives. For tip number seven, we're going to use the 40 image model to transfer a style that we give Chad GPT to an image. So, this is what I've written in this prompt here. Create
a handdrawn digital illustration of the attached image. I'm going to use the
attached image. I'm going to use the exact same image of me before in a bold minimal, energetic sketch style, like a high performance concept, blah blah blah. I've given it the hex code colors
blah. I've given it the hex code colors and let's see what it gives us. All
right. And here's the output that it just gave me. I think it's actually really, really good, especially considering it really did a good job following the prompt. So, handdrawn
digital illustration of the attached image bold minimal energetic etc. etc. I think it did a really good job with this. So, this is a great use case
with this. So, this is a great use case for you for any kind of brand photos that you have in your business, for example, and you want to turn them into fun sketches like this. Easy way to do it. Did it in about 90 seconds. All
it. Did it in about 90 seconds. All
right, for tip number eight, this is all about image expansion or outpainting.
So, this is essentially where you take an existing image and you ask Chat GBT to expand the image in some way. So, for
example, expand the border into some scene that you want. So, what we're going to do here is we're going to take the same image that I've been using here. So, I'm going to tell Chachi BT to
here. So, I'm going to tell Chachi BT to expand the attached picture into a landscape view inside a coffee shop. So
the goal here is to put me inside a coffee shop of the image that we've given it. All right. And here's the
given it. All right. And here's the image that it just created. It's
actually pretty good. Uh, it's a little I think I have like demon eyes, but it's pretty good. Like, I would totally work
pretty good. Like, I would totally work in a coffee shop like this. And a simple prompt. take an original image, expand
prompt. take an original image, expand the attach picture into a fill-in- thelank, whatever it is that you want to do, and chatb takes care of it in like again 90 seconds. So that is image
expansion or outpainting. Now, the next tip I want to talk to you about is editing your images. We talked a little bit about it already, but you do not have to start from scratch necessarily
when you don't like an image. TATBT is
really good now at taking your feedback and making edits to the image. This is
what I like to call iterative chat loop editing. So, you're basically
editing. So, you're basically continuously giving it feedback in a loop to get it to an image quality or something that you're really really happy with. Basically, you're treating
happy with. Basically, you're treating the image model as you're a live art director. So, you could say something
director. So, you could say something like, "Love it. Raise the camera. Punch
up the teal." Um, I could say with this image right here, for example, I could say add coffee to the uh bar back here
or add a barista behind the counter or something like that. I could fix the I don't know if this is like an Italian spelling of cappuccino or not. I don't
think it is, but I could uh I could definitely select it. I could tell ChatPD to correct the spelling of cappuccino, for example. So, you become
the art director of your images. just
simply giving feedback to chatbt on whatever it's kicking out to you. And
you just repeat that process until you're really happy with the image. For
this next tip, we're going to continue to take everything we've learned so far in this video and we're going to create assets for our business in terms of advertising. So, for example, you can
advertising. So, for example, you can use the image model to create ad creative. And I'm going to show you a
creative. And I'm going to show you a prompt here that very simple where we're going to create an Instagram ad creative. You could do this with
creative. You could do this with Facebook ads or uh Tik Tok ads or LinkedIn ads, whatever kind of ads that you want to be running, but I'm just going to say create an image for my Instagram ad creative in my branding.
Again, so I'm using the memory feature of ChatBT that aligns with the following ad copy. I'm purposely keeping it broad
ad copy. I'm purposely keeping it broad in the exact prompt because I want ChatBT to really get its inspiration from the ad copy rather than me telling
it something, not giving it the ad copy.
Then there's misalignment within the creative itself. So, let's go ahead and
creative itself. So, let's go ahead and see what this comes up with. All right.
And here's the image that it just created. Not too bad. I would totally
created. Not too bad. I would totally use this for an Instagram ad creative.
And and by the way, this is actually true. If you want to check out recall
true. If you want to check out recall which I talked about earlier in the video, and you want to get 25% off, just uh use the code recall Rick. It's linked
to in the description below. But pretty
cool, right? All I did was a simple prompt. I included ad copy and I told it
prompt. I included ad copy and I told it to create an image that complements the ad copy. Don't just replicate the ad
ad copy. Don't just replicate the ad copy in an image, but give me a an interesting image that I can use as ad creative. And then I can create
creative. And then I can create different variations of this if I want.
And speaking of variations, this is the next tip and the final tip that I want to share with you in this video. This is
what I like to call multi- scene storytelling. And what I mean by
storytelling. And what I mean by multi-seen storytelling, ChachiBT the 40 image model is really good at creating visual narratives obviously if you prompt it correctly showing like a
specific transformation. So for example
specific transformation. So for example in scene one you can say frustrated entrepreneur scene two organizing with tools or scene three then giving confident presentation. So it's almost
confident presentation. So it's almost showing the story arc this you could actually use this for an ad creative as well. And to do this, I'm going to show
well. And to do this, I'm going to show you in the OpenAI playground. And the
reason I want to do this is because I'm going to click on images. By the way this again, this is the uh the playground platform.openai.com.
playground platform.openai.com.
And I'm going to come down here to the bottom. I'm going to put in my prompt
bottom. I'm going to put in my prompt that I've written for this. And you'll
notice here that it says 1x. Well, I can choose that. And I can run the slider
choose that. And I can run the slider along to tell it how many images, how many variations of this image that I want it to create. This is great because
I can't create more than one image at a time in the normal chatbt tool. But if I want to create different variations of it up to 10, I'd have to use the or I
can use the playground. You just simply click on images in the lefth hand column here. I'm going to click on three. I'm
here. I'm going to click on three. I'm
going to give it the prompt. And again
we're going to create our multi-seene story telling a visual narrative. All
right. And here's the result. I think
that took like 26 seconds to generate these images here. So, you can click on expand and look at the image. Now, it
cut it off. I think I could have been Now, I was pretty specific in the size but uh 3280 x 1080 panels. Okay, I could probably be more specific in the sizing
but you can tell here of the storyboard that it's created for me here. And I can go back because I looked I because I asked it for three different variations.
So, here's the next one.
Here's the sad, organizing, and confident. So, again, it's telling a
confident. So, again, it's telling a visual story all based on the prompt here. And I can create, if I wanted to
here. And I can create, if I wanted to I can create again up to 10 different variations of this. And I'll often come in here and use the playground to
generate images because I can do this.
Just a quick side note here on when you're in the playground, you can click on this little toy box if you will or I don't even know what that is. A toolbox
maybe. And you can use all of these different sort of presets. Logo, 3D
city, birthday card. They have a whole bunch. And I think they've added to this
bunch. And I think they've added to this actually in the past little while. But
especially if you're creating like product designs or anything like that, I would be doing that over here in the playground. So they because they give
playground. So they because they give you they give you some more tools. The
other thing here is you can go in here if you're not setting in your prompt which I still recommend that you do uh is to go ahead and choose the uh sizing.
I would probably click on landscape. If
I'm going to do this same prompt again I would click on landscape. You can do quality and then click on advanced. And
you can do the different output uh format, PNG, JPEG, WEBP. Now, one final tip. What I would recommend that you do
tip. What I would recommend that you do if you're going to be creating images on a regular basis, especially brand images for your business, I would use chatbs and then you can just use the 40 model
within the project. So you can come in here, give your project a name. So like
maybe this is uh Instagram ad creative and this is your Instagram ad project.
So you have the prompt, you don't have to keep retyping it. It remembers all of your hex codes and brand colors, etc. etc. just do it all within the same project. You can create different
project. You can create different projects for all the different image projects that you're doing in your business. Now, a note of caution when
business. Now, a note of caution when you're generating images within chatp40.
This doesn't absolve your need for creativity. Quite the contrary. A lot of
creativity. Quite the contrary. A lot of people have talking about that was like oh, I don't have to be creative or it's taking my creativity away. You still
have to design the image and have taste to know what's good and also give it feedback as to what you don't like or the types of things that you want to change, which is going to take a
creative eye now. You don't have to hire a design team anymore to accomplish these things. Right? But you can also
these things. Right? But you can also use the type of work that you're doing or that we talked about here today in the video. you can take down these now
the video. you can take down these now fleshed out ideas of a design that you're thinking about and then if you do want to work with a design team then give them the things that you're creating in chatbt now they have a
stronger starting point. All right, I hope you got a ton out of this video here today. Again, if you're an online
here today. Again, if you're an online entrepreneur and you're really looking to leverage AI in your business so that you can scale your business, streamline it, and increase profitability in the process. I want to invite you to join us
process. I want to invite you to join us inside my AI playbook community. I'll
link to it in the description below.
Thanks so much for watching and we'll see you in the next
Loading video analysis...