See all Snowball AI transcripts on Youtube

Welcome to Snowball AI. Today, I'm excited to share with you my process for creating unique and consistent characters for children's books using some amazing tools. In this video, I'll be showing you how to use Mid Journey, an AI image generator, and ChatGPT, a language model, to bring your characters to life. I'll also be sharing a few essential tricks that I've picked up along the way that will help you create characters that are both memorable and consistent throughout your book. And for the finishing touches, I'll be using Photoshop and Canva to make your characters pop off the page.

So if you're ready to take your children's book illustrations to the next level, stick around and let's get started.

Now that we've talked about what we're going to cover in this video, let's dive into the

first part is called

the plan. The plan consists of 4 parts starting with the book, chat GPT, mid journey and Photoshop and Canva. For my children's book I've chosen to focus on the adventures of a little girl named Lila and her trusty sidekick, a lovable corgi. They'll explore the world and encounter different animals along the way, but to make the book even more engaging I've decided to include a nursery rhyme for each animal they encounter. And of course, Lila and her corgi will be present on every page to keep young readers interested.

I've also decided on a style for the illustrations. Colorful and whimsical, with a touch of realism to make the animals come alive. The theme of the book will be adventure with a focus on exploration and learning about different animals. For the next step we'll be using ChatGPT. ChatGPT is a groundbreaking language model created by OpenAI.

Imagine having an AI buddy that can chat with you, help you with homework or even brainstorm ideas for your next project. This super smart AI understands human language like never before. It's been trained on a massive amount of text, making it knowledgeable in a wide range of topics. And don't worry, you don't need to be a tech genius to use ChatGPT. It's user-friendly and designed to make your life easier and more enjoyable.

Stay tuned for more and discover the endless possibilities with ChatGPT. So without any further ado, let's go. I usually start by telling ChatGPT what I am trying to do. In this case, I write, I'm writing a book called Lila and corgi exploring animals. It's about a little girl called Lila and her corgi dog.

They love to explore and with every page they discover a new animal. Don't write anything yet if you understand answer with yes. It replied with yes I understand. Amazing! For the next prompt I wrote, Now I will write the animals depicted 1 by 1 for every page and I want you to write a short 4 sentence rhyme about that animal.

The rhyme should be intended for children or parents reading out loud for their children since it's a children's book. If you understand reply by writing I understand and I will give you the first animal. ChatGPT replied with I understand please provide the first animal. For the next step I asked it to give me a list of 25 animals that is interesting to include in the illustrations of a book for children from 1 to 4 years old. It did exactly that and I now have 25 animals that I can feed back to chat GPT so that it can generate the rhymes for me.

Look at the deer, so graceful and free, in the forest, running wild and carefree, with antlers tall and a coat of brown, a gentle creature seldom seen in town. Now I tried with another animal from our list. See the octopus in the sea, with 8 long arms, it glides so free, it squirts ink when it's scared or shy and changes colors Before your eye just amazing. This is as exciting as getting a new subscriber. So please click all the buttons down below Now that we have our rhymes it's time to start creating the illustrations.

And that's where Mid Journey comes in. What is this music?

SMASH IT!

Mid Journey is an AI image generator that can create stunning illustrations based on your specifications. I used Mid Journey to create the initial sketches for Lila and her corgi as well as the different animals they encounter. For high quality content creation I recommend using Mid Journey to create different parts of a scene instead of the entire scene at once. The pros with that strategy is that you can be in more control and be more consistent. The cons is that it takes a little more time.

To be fair mid-journey makes the illustration and prototyping process insanely fast, so 1 must be lazy to not put a little more effort into it and easily create high content. Let's move on to something important, and that is the prompts we use. Okay Billy, here we go. For Lila I wrote, a girl with short curly hair, different angles, in the style of children book character, character sheet, white background. With the important keywords being character sheet, different angles, children book character.

It's always good to include adjectives in the prompt to help describe your object. We can imagine something easily, but might get disappointed if we don't describe it. For example, instead of writing a girl, Write a cute girl with short curly hair and be more specific for the best results. I will probably make an in-depth guide in the future focusing on prompts for consistent character poses, colors, expressions and what they wear. If you would like that, write a comment down below and support this channel.

I usually go with an aspect ratio of 3 and 2, but there is a unique technique for generating character sheets using a 316 aspect ratio. This technique creates a long sheet with many more outputs, allowing you to generate a variety of different angles and perspectives for your characters. There's 2 main methods to achieve consistent characters and generations on Mid Journey. The first way is to upload a few images of the character or object that you like. Crop the character and save them as separate files.

You can use Photoshop, Canva, Paint or whatever program you like to achieve this. The next step is to paste the images into Discord and write the same prompt that you used to generate the image that you liked. This method simply reinforces the output and uses the images that you pasted into your prompt as reference. By using the same prompt again, you align the final result even more with the desired character. The second method is using something called seeds.

Seeds is a parameter that can be used by writing –seed and then the seed number of the image you're trying to use as a reference. Here is a quick example of how it works and the end results, demonstrated by MidJourney. In the first example, there is a prompt being used several times, without the use of seeds. If no seed is specified, Mid Journey will use a randomly generated seed number, producing a wide variety of options each time a prompt is used. In the second example, the same prompt is being used 2 times, but now with the added seed parameter.

We see that by using the same seed number, Mid Journey has been able to generate the exact same results. That is the power of seeds. It makes an image generator like MidJourney to give the exact same results, even though its main purpose is to generate new results and be somewhat random. The secret to character consistency in MidJourney is to combine both methods. This is what I used to achieve the following results.

In this prompt I specified that she is wearing scientist clothes. You can make some changes in the prompt and still get stable and consistent results using this method. Here I changed it and wrote wearing safari clothes instead. As you can see there is a difference in the close but the rest of the output is still the same except for some minor details. These images here is more different than the previous ones because I used Mid Journey version 5 instead.

Many people say that version 4 is better for digital art since version 5 tends to be more realistic in its outputs. But I believe that experimenting and using both versions will give you the best results depending on the project. Here we wrote wearing a beautiful dress And we got this cute image. Here's another 1 of our main character, Lila, holding a basketball. I used version 5 to get this result.

And the last example that I want to show is this 1, where I wrote, wearing winter jacket. I love how the nose turned red and is fitting for a cold climate. As you see here it went perfectly with the environment and setting. Now that we have enough illustrations of our main character, we can move on to generate the animals that we want to include in our book. We will start with the sidekick Corgi and move on to the other animals 1 by 1.

I went with a pretty simple prompt to achieve the rest of the results. I wrote beautiful Corgi, different angles, in the style of children book character, character sheet, white background. I included the aspect ratio of 3, 2 and BAM. Look at this adorable Corgi, makes me almost want to invest my entire life savings in Dogecoin. Almost.

I used the same prompt for all the other animals as well. I even kept the same seed number and tried to get similar style and theme. Here I changed corgi and wrote deer instead. After that, I put all the other animals from the list 1 by 1 and Midjourney did the rest. Bless you Midjourney.

For the backgrounds I used different prompts and the main focus here was to describe what I wanted. I wrote the following prompt African desert and nature with a small water pond in the middle, in the style of children book illustration. The essential part here is to include in the style of children book illustration, to get similar results as the rest of the images mid-journey generated. Finally, let me show you how I used Photoshop and Canva to add the finishing touches to the illustrations. With these tools I was able to add texture, color and depth to the illustrations to make them really pop off the page.

I also used Canva to create the layout for the book, arranging the illustrations and rhymes in a way that's both visually pleasing and easy to read. It's amazing how much you can do with just a few simple tools. I want to share 2 simple tricks in this video for adding a character to a scene. The first is the use of shadows. By placing a simple shadow beneath your desired object you add another dimension to your character and the scene.

The second trick is the placement of the characters, scene or objects relating to each other. For example, by putting a character behind an object you can create a sense of a living environment and an illusion of depth of field. Here I use Photoshop to erase the palm of the panda. This simple trick added the illusion of the panda holding the bamboo tree. Try to think more creatively and you will not have to relay on generating the exact position or angle of a certain object to create something truly amazing.

If you're interested in more in-depth content about scene perception and understanding the basic psychology of a scene, let me know in the comments. I personally love that technology has gotten so far that a person alone or a few people together can create something incredible in a few hours which otherwise would have taken weeks, months or years only 1 or 2 years ago. As the guy from two-minute papers say, what a time to be alive. Here is another example of adapting the character to the scene. I have put the hippo tuto, hippop, hippie popo, hippocampus, hippopotamus in the water, but his feet isn't wet.

I mark the legs and use the gradient tool and choose blue and darker gray. This creates the illusion of the hippo's legs being underwater. The gradient tool creates the effect that I want. That is, the further down the legs are in the water, the darker the shading. Now the only thing left to do is to add our main character, Lila.

I use 1 of the character sheets I generated using Mid Journey. It's a rear view of our character. Use that word in your prompt to get your character facing away from you. Remember to combine it with words like character sheet or character concept design to get various consistent results fast. Here I try to create a shadow for Laila.

I use a method where I mirror a copy of the character and flip it horizontally. The idea is good but it's better suited for when trying to reflect the character in a river or nearby water. I realize this here and change it to the simple eclipse version with Gaussian blur. I create 2 more copies of shadow and put it underneath the legs of the hippo. There is many more details a perfectionist can work with in this scene but for now I am proud of it and we'll save it there.

If you have any suggestions or creative ideas on what could be added or removed from the scene make sure to write your suggestions in the comments below. It's for free, only for you. Please come again! Here I tried something at first and then changed my idea. Happy little accidents.

I like to add the important parts of the image separately. The little variations in style can sometimes add character to the key objects in a scene. I realized that the image looked plain and the penguin was kind of melting with the background. I needed a better background and the penguin to stick out. Photoshop is an amazing tool and I recommend you to get familiar with it if you don't use it already.

Adobe is implementing AI in almost all of its products and with communities already implementing stable diffusion as an add-on inside of Photoshop, things will only get crazier and I don't see Adobe or Photoshop getting left behind in this race. Getting back to the image I ended up replacing the first penguin with another 1 and chose a different style. I created a shadow for the penguin and aligned the position to be outside of the cave. I put Lila inside the cave looking at the same direction as our penguin. This draws the focus of the viewer to the direction of the character's gaze.

We finish it off by placing our dear Corgi and put 2 more penguins at the peak where the attention of the viewer is drawn. I make the third penguin a little smaller and put him behind his friend. I finish our masterpiece by placing a shadow under Lila's feet and change her color temperature. In this example here I generated an image where we can see below the water as well as the land and sky. I try to depict the octopus underwater and have the environment feel alive.

I proceed by cutting out the octopus and put 1 of them behind the aquatic plants. It creates the effect I want. For Lila I switch the head part with another pose that I like better. It shouldn't be harder than that. Now that I have completed all the illustrations for my book, I switch over to Canva to put the entire book together.

The layout of the book will be an illustration on the first page and a nursery rhyme on the second for every animal from our list. This can help the reader view the illustration and read the rhyme at the same time without having to turn the page. I copied the rhymes from chat GPT 1 by 1, pasted it into Canva, adjusted the text size and aligned it to the center. I could have chosen a better background for the nursery rhyme pages or changed the colors for every page to make it more attractive, but it does the trick for now. Remember to create a beautiful and simple page for the ending and another 1 for the cover of the book.

Some good tips here is to include sneak peeks of some of the illustrations inside the book, title and a short description of the book. Use at least 2 different fonts in the cover if you're creating a children's book to make it more playful. Okay join my discord, subscribe, bye!