Вештачка интелигенција

A Different Perspective on AI Showdown: Evaluating DALL-E 3 in ChatGPT vs. Midjourney

Summary

In recent weeks, there has been a lot of buzz around the application of DALL-E 3 in ChatGPT Plus. By integrating text-to-image capabilities into ChatGPT, artificial intelligence has made image creation incredibly easy. However, curiosity struck me – how does […]

A Different Perspective on AI Showdown: Evaluating DALL-E 3 in ChatGPT vs. Midjourney

In recent weeks, there has been a lot of buzz around the application of DALL-E 3 in ChatGPT Plus. By integrating text-to-image capabilities into ChatGPT, artificial intelligence has made image creation incredibly easy. However, curiosity struck me – how does DALL-E 3 compare to Midjourney? Having had extensive experience with Midjourney, I was familiar with its greater customization options compared to DALL-E in ChatGPT implementation. But how does it stack up without any additional adjustments? Let’s have a traditional AI showdown, pitting DALL-E 3 in ChatGPT against Midjourney in eight image comparison tests. And since it’s the Halloween season, we’ll focus on that theme. Each prompt includes an artistic or presentation style and an image to be created.

To conduct the tests, I provided identical prompts to both AI models. I had to prefix the introductory sentence of all Midjourney prompts with “/imagine,” while all DALL-E prompts started with “image.” In all other aspects, the prompts were the same. Midjourney automatically generates four images, while DALL-E creates two. I ran the DALL-E test twice to have four images for comparison with both AI models. I then compiled all the images in Photoshop to see all eight candidate images side by side. The evaluation criteria were as follows:

1. How well does the AI model match the given style?
2. Does the AI model incorporate the specified elements of the image?
3. Which image is my wife’s favorite, and which AI model created it?
4. Which image is my favorite, and which AI model created it?

After collecting the results, I will announce the winner. So, let the Halloween AI showdown begin!

FAQ

What is DALL-E 3?
DALL-E 3 is an artificial intelligence model that combines text and images to generate new visuals.

What is Midjourney?
Midjourney is another AI model used for image generation, offering more customization options than DALL-E.

Why was this comparison conducted?
This comparison aimed to determine how DALL-E 3 in ChatGPT compares to Midjourney when it comes to image creation for specific thematic demands.

Who is the winner of the showdown?
In a specific test, DALL-E 3 won in creating images of friendly witches, while Midjourney excelled in creating images inspired by Snoopy and the Great Pumpkin.

How is artificial intelligence used to create these images?
ChatGPT Plus and text-to-image tools are utilized to generate images based on given prompts.

Can these images be used for commercial purposes?
Considering copyright restrictions, images created by DALL-E 3, including those depicting Snoopy, cannot be used for commercial purposes without proper permission and licensing.

Source: ZDNet – AI Showdown: DALL-E 3 in ChatGPT vs. Midjourney