🖌️ AI Art Tips Thursday #7: Midjourney 5.1 is here!

💡 AI Art Tips for the Day

By popular demand, I was supposed to talk about prompt weights in Midjourney. However, since Midjourney just released the 5.1 version, I thought it would be good to cover it while it is fresh on our minds. I will return to talk about the tips about prompt weights next week.

So what has been actually communicated for the update from Midjourney officially:

It doesn’t look like the change is that much compared to from v4 to v5, but it is hard to tell from just how it is shared in the announcement. The proof is in the pudding, so let us give v5.1 a try to see the real difference.

For the test, I am going to try just short prompts and would try both a prompt without humans as subjects and one with humans as subjects. I would try the v4, v5, v5.1 and v5.1 raw model so that we can get some comparison going.

Alright, check the results below 👇️ 

I am not sure about you but I don’t see much difference in this example between v5, v5.1 and v5.1 raw model. All versions interpret the prompt quite nicely. The only interesting point is that in the v5.1 and v5.1 raw model the water view is more prominent in all art generated.

Also, why do they suggest that many Sydney Opera Houses? 🤷‍♂️ 

Let’s try some examples with humans 👇️ 

Now you can see some more obvious differences. It is interesting that in both v5.1 and v5.1 raw model we can see the variations with a manga style whereas v5 is solely for the photo style. Also, you can see that in the v5.1 raw model only one of them is the manga style so I guess that is why in the announcement they mentioned that this raw model is more similar to v5.

To reduce any variations due to different seeds, I also try the different versions using the same prompt with the same seed. Let’s see how do they look like 👇️ 

The comparison is quite different now. We can see the emphasis on photo style shows up more in v5 and v5.1 raw model when the variation from the seed is removed.

Cool, so in conclusion, this is what I found by playing with v5.1 and comparing it to v5:

• General output style-wise, v5.1 is more similar to v4: the interpretation of the simple prompt won’t always be a photo.

• If you would like to try a more improved model with the same emphasis on a photo style, then v5.1 raw model is your choice.

Try this in your next prompt

{your prompt describing the art} --v 5.1 or --v 5.1 --style raw

🔤 Word of the day

Bokeh

n. is the aesthetic quality of the blur produced in out-of-focus parts of an image, caused by Circles of Confusion. Bokeh has also been defined as "the way the lens renders out-of-focus points of light"

As always, you can check for more details about this in this Wikipedia.

If you put this word in your prompt, here is what you can expect to get:

🖼️ Image of the day

Well, the face of the little kid is a little off but the overall image is 🧑‍🍳🤌 . Give me a great Studio Ghibli feel.

💕 Share the love, get some love!

I am doing a weekly draw to get people referring the newsletter to others with some cool avatars.

If you can refer at least 3 of your friends, you will be in this weekly draw of 3 to get 3 cool avatars using your chosen portrait with the styles of your choice.

So if you enjoy this newsletter and want to get some cool avatars for your socials, here is your link to share your love of the Prompt Craftsman with your friends. You are the best! ❤️