Here's a simple way to create consistent characters with Dall-E 3.
If you need to create a series of images with a consistent look, you might be pleased to know that this is possible in Dall-E 3 in ChatGPT.
In my previous post, I explained a cool and unique feature of Dall-E 3 that supports seed numbers. Seed numbers allow you to recreate the exact same image later. Dall-E 3 normally generates images randomly each time you provide a prompt. But with a seed number, you can force it to generate the same image again.
When GPT Store was announced, creators were quick to put together GPT applications that leverage the seed numbers to create consistent characters.
In this article, I will show you the step-by-step process of creating a consistent character using a GPT app called “Consistent Character GPT.”
Let’s get started.
The first thing I need to do is generate a character. Later on, we’ll make variations of that character while keeping the looks and aesthetics consistent.
Head over to the GPT store and under the Dall-E 3 tab, click on the “Consistent Character GPT” app.
On the app’s dashboard, click on the “Click here to start creating character designs” button.
The first thing the AI asked me is to select whether my character is male or female.
Prompt: Great, let’s start creating some amazing character designs! First, could you please tell me if the character should be a male or female?
Prompt: Please provide a name for the character.
I named my character “Jim” because I plan to create an avatar of myself!
Next, it asked me to give more details about the character.
Prompt: Could you share details about the character Jim, such as his age, country of origin, hair style, outfit, and any other characteristics you’d like to include?
I set the following details of my character:
Make sure to add the “Full body” in your prompt. The AI will only generate a face if you don’t specify.
The final question it asked is the style or theme of the image. I selected Pixar Animation because I wanted to see a 3D render of myself.
Here’s the final result:
Awesome! Now, it’s time to make variations of me.
Right now, you can instruct the AI to change the image however you want. For example, if I wanted my character to walk, this is what the result looks like:
All the attributes from the original image look almost exactly the same. Here’s another jumping pose:
It looks super cute but the hands are always inside his hoodie. So I asked the AI to create an image of me in waving pose.
Overall, I had so much fun creating these consistent characters with Dall-E 3 and ChatGPT. Major thanks to ailemonacademy.com for creating the Consistent Character GPT app that made this process smooth and user-friendly.
I can see several use cases for this technology, like quickly mocking up character designs for games, animations, or comics. The consistency would allow asset creators to easily generate different poses and expressions without having to redraw the character each time.
Looking to the future, I’m excited by the possibility of sketching my own character designs and having an AI generate poses based on my drawings. Rather than relying completely on AI-generated images, I could leverage the technology to expand on my own artistic visions.
This is only the beginning as generative models keep advancing. I am excited to see what other creative use-cases will be unlocked in the coming months. For now, I highly recommend experimenting with the consistent character GPT app yourself!
‍
Software engineer, writer, solopreneur