So, can ChatGPT draw pictures? Well, as of this article, no, ChatGPT does not draw pictures. However, DALL-E can. DALL-E is a project of the OpenAI.com team who also develop ChatGPT.
See the video of this article on YouTube.
I was experimenting with ChatGPT in order to create short stories. While I was using ChatGPT for this purpose, I started wondering if ChatGPT could draw images. It cannot, but I discovered an AI bot that does: DALL-E. In using DALL-E I supplied the following text, which was extracted from a short story I created in collaboration with ChatGPT, for it to draw a picture:
Using realistic drawn cartoon art style draw a breathtaking view from a castle's balcony, overlooking a vast expanse of a dense forest, a distant lake with small village nearby, and distant mountains.
It drew four pictures, as it typically does. Here are two of the pictures the DALL-E AI drew based upon the above description:
|
 |
Using DALL-E, like ChatGPT, can be a bit frustrating at first. However, over time you learn what specific words and phrases help ChatGPT, DALL-E, and other AI bots, provide the results you want. Unlike ChatGPT, DALL-E does not remember previously drawn images from previous text descriptions. Also, DALL-E, at this point, does not support direct image upload as a guide for generating subsequent illustrations.
There are several AI websites that generate pictures based text prompts. Here are a few examples:
-
DALL-E: DALL-E is an AI model developed by OpenAI that generates images from textual descriptions. It can create a wide variety of images, including animals, objects, and scenes.
-
Artbreeder: Artbreeder is a website that allows users to create and explore images generated by AI. Users can provide text prompts or combine existing images to create new ones.
-
RunwayML: RunwayML is a platform that allows users to create and use AI models for a variety of tasks, including image generation. It provides a range of pre-trained models that can generate images from text prompts.
-
Deep Dream Generator: Deep Dream Generator is a website that uses AI to create surreal and dreamlike images based on user input. Users can provide text prompts or upload their own images to generate new ones.
It's important to note that while these AI-generated images can be impressive and interesting, they may not always be accurate or reliable representations of the original text.
This input text yielded the four images below:
The dense forest is rich with detail. The ground is covered in a carpet of fallen leaves and moss, softening footsteps as a boy and a girl traverse the forest floor. Ferns and wildflowers dot the landscape, adding vibrant pops of color to the greenery. Small creatures, like squirrels and birds, can be spotted in the branches, observing the young explorers' progress.
 |
 |
 |
 |
For both ChatGPT and DALL-E, wording and wording order does matter. Take the following illustration for example:
Note that in the illustration the boy is holding a mop; however, here is the original text description entered into DALL-E:
bring to life an illustration in an art style that delights children, inspired by the character Pryor. Pryor, a young boy with a mop of brown hair, leads the way with a spirit of determination. He wears sturdy hiking boots, practical outdoor attire, and carries a backpack filled with exploration essentials.
Note that the description says, "boy with a mop of brown hair," but DALL-E interprets it and separates it such that the boy of brown hair is holding a physical mop. The other three images are similar illustrations with a boy holding a mop of some type:
 |
 |
 |
So wording matters. But with practice and experimentation, you can create great pictures with DALL-E and other AI picture generating technologies.
That's all for now. Thanks for taking the time to read this article.