Video Gamer is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more
ChatGPT can now generate images. Let’s break it down.
ChatGPT has become something of a phenomenon thanks to its uncanny ability to create natural language and accurate code, but can it create images? Yes, and they are shockingly impressive.
In September 2023, OpenAI integrated a new version of its DALL-E image generator into ChatGPT. Named DALL-E 3, the technology can produce more convincing images than its previous iterations. Where before it may have struggled with letters, numbers and human body parts, DALL-E 3 can produce images that hits those features with accuracy.
OpenAI adding DALL-E 3 to ChatGPT has levelled up the chatbot to be a hub for generative AI, which can produce text, images, sounds, software on its own.
Jasper AI
Copy AI
Originality AI detector
How to generate images on ChatGPT
ChatGPT can now create images because of the ‘Vision update’, which combines GPT-4 with DALL-E 3, forming GPT-4V. To access this new update, you must be a ChatGPT Plus subscriber.
- Step 1: If you haven’t already, sign up for ChatGPT Plus
- Step 2: Log into ChatGPT and select GPT-4. In the drop down menu, click DALL-E 3 Beta
- Step 3: That’s it. Write ChatGPT a prompt for an image you want to create.
- Step 4: To download an image, hover your mouse over it, and click the Download icon
It really is that simple. Similarly to other text-to-image generators, DALL-E 3 will then generate four images to choose from.
DALL-E 3 prompts for ChatGPT
Here are some examples for prompts that you can try out on ChatGPT.
- A pop art style image of a frog in a forest
- A photograph of a family in a garden looking serious
- An oil painting of an ocean with fisherman boats
- A surreal painting of a tiny mouse in a massive shoe
- An illustration of a dystopian cityscape
- An enchanted forest
- A painting of the artic with northern lights, ice, shades of blue, polar bears
DALL-E 3 can use the prompts to create your images and develop your ideas. To dow
Can ChatGPT create images without DALL-E 3?
While ChatGPT itself cannot generate images, it can create detailed descriptions of images, based on users’ prompts. From here, you can then use a text to image synthesis program to create an image that matches that description.
Essentially this means the image is created just with an extra step added in between. ChatGPT is especially good for this kind of method as the language it outputs is very natural, just like the language the image generator has been trained on.
Read more: What is ChatGPT used for?
For some, this extra step might prove redundant, as you could simply type out the descriptions of the image yourself. It is true that going straight to the image generator will save one step, however, do not underestimate how detailed an image description ChatGPT can create in an incredibly short amount of time, providing it is not at capacity that is.
Image generation alternatives to ChatGPT
Jasper Art
Jasper Art generates 4 images for each input you give it. It’s very versatile and can generate images in multiple different styles depending on your text input. The images it makes have no watermarks and each is unique.
Midjourney
Midjourney uses text-to-image generation to create four images based on prompts given by the user. There is an extensive list of prompts that can be fed to Midjourney to adapt the style, size, tone, and a lot more. Find out how to edit images on Midjourney here.
Starry AI
Starry AI has more than 16 different styles for creating images. It also gives complete ownership of the images it creates over to you, to use as you wish. It’s very beginner friendly which good templates and processes to walk you through how to get it to generate the kind of images you want.
Frequently asked questions
Can ChatGPT make text art?
Yes, though with some questionable results. It is able to make ASCII art, though sometimes gets it wrong.
