Using ChatGPT to Create Images ... Even if You are not Sure What Image You Need.

Simplify AI Image Creation by using ChatGPT

Write a short nonfiction eBook in thirty days or less. Our course shows you how for just $35. Click here to learn more.

---

Writers, publishers, and course creators frequently need images to illustrate blog posts, social media content, or slides for video. However, finding the right image for your content can be time-consuming. Sometimes you aren’t even sure what you are looking for. You go to the stock photo site and run all sorts of keywords and nothing shows up. 

Even if you go to a generative AI art program like Dall-E, Midjourney, Canva, or Firefly, you have to describe specifically what you want in the picture. However, sometimes if you are dealing with an abstraction like anger management, productivity, or evocative writing, it is hard even to think what that image should be.

This is where ChatGPT can help. First, this only applies to ChatGPTPlus, the premium version. It has DALL-E installed, but since Chat is a large language model it is also skilled in writing prompts for an image-generating program. So, even if you don’t know what you want specifically, you can give it a general idea of what you want.

For example, you can say, “I am writing an article for my blog about time management for writers. Could you create an image to illustrate the article?”

Of course, the more information you give the better the results. You might, for instance, designate that the image must be photorealistic, be 1600px wide by 900px tall, and a clock be the dominant image in the final result.

You might have to regenerate the image a few times to find one you like, but it’s faster than looking through a ton of stock photos.

A second way is to upload the content you want an illustration for. For short pieces, you can just copy and paste. You can click on the paper clip and upload the file for longer content. Then create a prompt like this, “Chat, I uploaded a file called (name of file). This is a blog article on the subject of ____. Please read the file and create an image to illustrate it.”

I’ve done that often and got good results.

Finally, if you have a good idea of what you want to see in the image, you can give chat a general description and let it create an image for you. The difference between this and using Dall-E, Firefly, or one of the others is that you can be more conversational and a bit more generalized.

For instance, working directly with something like Dall-E, you need to say something like “A humanoid robot standing in front of a big flat video monitor. It has a paintbrush in hand painting an image on the screen. The background is a painter’s studio.”

With Chat you can be more general and conversational. For instance, “Chat, I need an image to illustrate an article about using ChatGPT to generate images. Here’s the idea I have. I would like to have a humanoid robot dressed like an artist and painting a picture on a large video screen instead of a canvas.”

In fact, that is the prompt I used for the image for this post.

Using any good generative AI image creation program will produce great results, but using a large language model like Chat can help take a general or even abstract idea and turn it into something more concrete and ready to use.