Understanding Ghibli Images.

If you’ve been hanging around social media, it’s impossible to miss the new “Ghibli Images”. In today’s shots, we’ll see what it is and what’s driving the sudden activity.

Before we begin, our aim at AI Shots is to get your acquainted with developments in AI in a simple and easy to understand manner. If you like what you read, consider subscribing.

Ghibli is an animation studio based out of Japan. They create beautiful films with their characteristic look. They’re known for their hand drawn looks, beautiful backgrounds, expressive characters, focus on nature and a magical and cozy feeling.

You can just Google to see some of their creation.

Now, the reason they are popping up everywhere is not because of Ghibli itself but a new launch from OpenAI. OpenAI had this Dall-E model that’s used for generating images. Dall-E is decent to play around with but we’ve never been able to create images that we can use. (Some people do post these on linkedin and it just shows ☹️ ).

About a week ago, Open AI launched a new model for image generation. And it’s very, very good. They’ve made it available in their free plan also. You should go try it out.

Let’s understand what “good” means here. The output image is something you can actually put out there. The banner image for this post is created through that model. When this launched, some people tried it for a “Ghibli style image”. The beauty of Ghibli quickly caught people’s attention and became a default prompt that people try with this new model.

So it just blew up. Nothing more.

How can you use this?

Now, let’s talk about how can you use this model in your work. There are a couple of ways to start with

• Quick Image generations:
  Say you want to create an image for a blog post like this or for your Instagram post. You can use an existing photo of yours or create something completely new with this model and use it. The fact that these images are of really high quality means your content will look premium and yield better results (whatever that may be).

• Concept Iterations:
  One of the things I use it for is to iterate on vague concepts in my mind. As creators, we have a root but there are many ways to represent that. You may not be able to think of all of those. What we do is to drop the concept into chatgpt and ask it to give me a few ideas. You pick one thread and make further changes to refine the concept to arrive at the result you want. It’s like creating a mock up for your designer. Only that it’s become much easier.

Potential issues

Some of the potential issues that come with this is that now it’s way more easier to create fake documents. Here are some examples.

This means that, it’s much harder to verify documents in most places. Companies are already investing more and more into verifying identities digitally instead of physical documents. This new development will only exacerbate that trend.

Financial applications which predominantly rely on documents, ID cards will have massive levels of fraud. This also creates markets for companies that can effectively verify this.

There are a lot more 2nd and 3rd order effects that will only become clear as we experience this new change.

Stay alert, stay informed.