Test: ChatGPT vs Microsoft Copilot on image creation

ChatGPT Image Creation

I tried comparing the quality of ChatGPT and Microsoft Copilot when it comes to creating an image. I learned that they both use DALL-E tool, but the results are still different as you will see the article.

I chose the topic of chihuahua dog, since I find it amusing. The prompt engineering is everything in the world of AI visuals. You know the right prompt, you get the best output!

So my prompt was as follows:

create a picture for me of a chihuahua wearing a christmas hat and christimas tree in the background

I logged in ChatGPT and entered the above prompt. I got the following image:

Now, using the prompt, I want to add some deer and a sleigh to this picture. My new prompt is:

add to the image 6 reindeer and sleigh

The new image created by ChatGPT is below:

I see the added reindeer toys, but they look somewhat hazy in the background and a chihuahua looks a bit intimidating. So I wanted to make the dog friendlier. My prompt was:

in the picture make chihuahua smile and friendly

Unfortunately ChatGPT through me a out of limit error, meaning that I have used up all of my credits for the day.

So the limits imposed allow me only to create 2 images per day, which in my opinion, is very little value. Purchasing a Plus version could be a solution, but for now I would check what is free out there.

Trying Microsoft Copilot

Microsoft has released its Copilot version that is working seamlessly with Office and also Edge browser. To activate it click on small blue Copilot icon on the top right of you browser:

Right away you get your right panel opened with Copilot window. Now I will try entering exactly the same prompts that I did for ChatGPT.

For the first prompt, I got the following output:

Wow, I see 4 pictures of Chihuahua wearing a Christmas hat! This is quite a different experience from ChatGPT’s one picture result. Now with Copilot, I can choose between 4 different pictures which one I like the most. Out of the 4 pictures, I see 4 different variations of moods of the dog: interested, smiling, cute and surprised.

To further adjust the image, I typed the second prompt. I got the following results:

If you zoom in, you can see the quality difference when compared to ChatGPT. The reindeer in the top left picture are in the foreground and sharp unlike in ChatGPT which was hazy. The color choices are better, such as including a green sleigh, which is nicely contrasting with reddish Christmas hat, giving it a more present look.

Now to add some fun, I tried making the dog smile with this prompt:

in the picture make chihuahua smile and friendly

Wow, what a delight – a smiling Chihuahua with a Christmas hat, reindeer and sleigh! That would make anyone laugh. And Microsoft Copilot was not limiting my image creation the same way as ChatGPT did.

From my personal test, I conclude that Microsoft Copilot provides us with a better image creation possibilities than ChatGPT. However, prompt engineering is key, so make sure to create those perfect prompts to get the best possible images!

Let me know if you have different experiences in the comments below.