pro member exclusive

how to use chatgpt to create images

if you can dream it, chatgpt can build it.

by sandi leyva

chatgpt 4.0 is enabled with computer vision, which means it can derive information from images that you input. it can also create images that you can use in your business.

check out our courses on chatgpt,
offered as a deep discount bundle or separately.
webinar schedule:
more chaptgpt here

here are just a few things i have found when using chatgpt 4.0 for image creation:

chatgpt 4.0 is the version of chatgpt you have access to when you have a paid account. chatgpt, the free version, does not read and output files like chatgpt 4.0 can. chatgpt 4.0 is multimodal, meaning it has multiple capabilities, including those of dall-e, a text-to-image model. access chatgpt 4.0 via https://openai.com/ .

you can probably think of several use cases, especially in marketing:

  • images to illustrate your website
  • social media graphics
  • article and blog illustrations
  • graphics for paid ads
  • illustrations for presentations
  • images for marketing materials, such as prospect kits and proposals

when inputting an image file and a prompt to describe the image, chatgpt can do so competently.

when asking chatgpt to create an image, the way you word the prompt or input request has everything to do with how well your image will come out. chatgpt has a particular default style that is best overridden with a few tips.

first, chatgpt tends to overcomplicate the graphics it creates, making them very busy. include in your prompt a request to have chatgpt simplify the design by using just a few design elements (it will still create too many, but at least it will be simpler than if you haven’t included this in the prompt).

second, make sure to ask chatgpt to not include any text in your images. it tends to misspell words in graphics. you can always easily add text later using free programs. here is an example of both an overdesigned graphic and a graphic with misspelled words:

 

third, describe the type of images you want, such as photograph-like, drawing, cartoon or the like.

and last, include as many details in your prompt as possible. i have seen phenomenal graphics come out of chatgpt, and the prompts have been half a page or longer.

here’s a generic office image i created using chatgpt, which is typical of chatgpt’s style:

 

and another after i asked it to follow the accountant’s accelerator blog header image style:

 

after chatgpt has created your image, here are a few things to check:

  • if people are in the image, make sure they are realistic-looking. i’ve seen six fingers, three legs and that sort of thing before.
  • if you don’t like the graphic it came up with, you can always do a follow-up prompt. sometimes, starting a new chat works just as well, especially if the image is very different from what you want.
  • don’t ask chatgpt to edit images, though. it’s pretty much a waste of time. however, you can go into the openai api and use dall-e to edit images. this requires using credits, which is a pay-as-you-go system. you can access it via https://labs.openai.com/ .

give images a try in chatgpt, or one of the many other ai-powered image generators out there. it’s just one of thousands of time-saving tasks that this technology can help you with.

want to learn more about chatgpt, including all of its new features? check out our courses on chatgpt, offered as a deep discount bundle or separately.