It’s July 2022. Suddenly there’s a life-changing tool that generates images from text: Midjourney. Well, it wasn’t the first tool to generate artificial images. But suddenly there were high-quality images, instead of pixels. Yet there were still flaws. Midjourney v5 was released a few days ago. What has improved in six months?
In September I wrote an extensive article about Midjourney on Frankwatching. The quality was good, but you could still see that something was wrong with the captured images. Hands didn’t really look like hands yet and faces didn’t always have a fresh look. We are now half a year further. It is high time to dive into the program to see how far Midjourney has come in six months.
What is Midjourney again?
In a nutshell, Midjourney is a public server on Discord. You can access this server for free. With your free account it is possible to submit several dozen text requests. The program then creates four images based on your text. That can be short and concise, but also super detailed. It’s just what you want to see.
After a while you will run out of text requests and it will be possible to switch to a paid version. There are several subscriptions between $10 and $60 per month that give you unlimited access to the program. The more you pay, the more options you get in terms of quality and speed. It is also possible to use your search terms privately in your own server.
This is how you use Midjourney
- Go to https://www.midjourney.com/home/
- Klik op ‘Join the beta’
- Create an account on Discord
- Log in
- Select one of the newbie bots on the left
- Type /imagine and type out what you want to see
New in Midjourney v5
Last July we had to deal with the first version of Midjourney. Now we have arrived at v5. Here are the biggest changes in Midjourney v5, compared to v4:
- Upscaling to a higher resolution is no longer necessary. You get the generated images – all four – immediately in the highest possible resolution.
- It is possible to apply multiple aspect ratios, such as 3:1 or 1:4. In previous versions this was 1:2 or 2:1.
- You have more control over the tiniest details in your written prompt.
- Add multiple images and indicate how much you value each image in the final result. This way you indicate which photo you want to show more, compared to another.
- Toes and fingers now look more realistic.
- Reflections are more natural.
That all sounds good! That is why it is high time to see those changes in practice. Let’s see how photorealistic, super duper and high quality this tool is half a year later.
Applications of Midjourney v5
Let me get straight to the point: I am speechless. In my article on Frankwatching in September, you’ll see results of images that are nowhere near what’s possible right now. Yes: the images back then were bold, because there was nothing better yet. Even now I realize how quickly things can change in a short time. So here are some randomly generated images and related prompts in Midjourney v5.

(peaceful yoga school), (tranquil, serene, spiritual, harmonious), (Canon EOS R6), (Canon RF 24-105mm f/4L IS USM), (golden hour), (candid lifestyle), (Kodak Portra 400) –q 2 –v 5 –s 250

realistic cartoon style coloring page of a happy and cute bear reading a book in the woods, close up headshot, dynamic pose, upper body portrait, child coloring page, black and white, crisp lines, clean and simple line art, white background –ar v5 17:22

“Create an image of a man in the jungle, focusing only on his body and the surrounding environment. The man should be dressed in rugged clothing and carrying a backpack or other gear. He should be positioned in the middle of a dense jungle, surrounded by tall trees and lush vegetation. The lighting should be dim, with only small beams of sunlight piercing through the thick canopy above. The man’s body should be visible, but his face should be obscured, either by shadows or by turning his head away from the viewer. The overall mood of the image should be adventurous and mysterious, with a sense of exploration and discovery v5.

game asset 16bit pixel art isometric, large dungeon, with details, items and wall, medieval fantasy
This way I can go on indefinitely. I look again with the same enthusiasm as six months ago at the increased level within Midjourney. The images you see here were created in minutes. The only thing that stands out now is the lady’s toe in yoga pose. It feels unnatural. But further? Judge for yourself about the final results.
Midjourney v1 compared to Midjourney v5
To show Midjourney’s progress even better, I’m going to re-enter the search terms I used from September. First (left) I’ll post the image that Midjourney generated in September. In addition (right) I place the best result from Midjourney 5. In this case pictures speak more than a thousand words.
- ‘Photorealistic image of a guy writing a blog about AI’
- ‘Warrior knight ivory colored mythical creature.’
- ‘photoreal detailed batman cat.’
- ‘settlement in the middle of the forest in the style of Studio Ghibli.’
The quality of the images has grown enormously in six months. In these images, all I see is the writer holding two pens and writing on a laptop. That doesn’t feel natural. But if we look at the Batman cat, the end result is many times better than before. What do you think?
Do I still share the same opinion?
Half a year ago I said:
Midjourney is a wonderful tool for creating concepts. Designers can use this tool to create sketches and to pimp them to their own liking. This way you get the best of both worlds.
And that’s exactly the way I look at a tool like Midjourney. The speed at which the quality of these tools is going up is astounding. I can’t imagine what the quality will be like in half a year, but that will probably be even more bizarre.
I am convinced that Midjourney and all other AI tools, such as ChatGPT, are brilliant tools to increase human efficiency. A computer expresses your idea and you bring that expression to life. We have entered a new era. AI is here and will stay.
The Dangers of Photorealism (Controversy)
With advanced tools such as Midjourney, there are also dangers lurking. I am not the first to say a word about this, but certainly not the last. If everyone gets access to making photorealistic expressions for next to nothing, then there will also be people with bad intentions.
Neem the controversial news story about Donald Trump from last week. There were AI-generated images showing Donald Trump being arrested by New York police. Very fake, but in the wrong hands a powerful tool for spreading fake news. Take a look for yourself at the fake images that have been doing the rounds:
Regulations are lagging behind
Also in September I already talked about the lagging regulations. There is a lot of work to be done, both in the field of copyright for the expressions made, and in the distribution of images based on untruths. Apparently the government’s priority is not yet here.
On the site of the National Inspectorate for Digital Infrastructure of the Ministry of Economic Affairs and Climate, checked on March 22, 2023, there appears to be no regulation at all about artificial intelligence (AI). It does state that they are busy identifying changes and associated risks. How long will it take for a bill to roll out there? Joost should know.
Conclusion
Well, what can you say now. The past year has been dominated by AI almost everywhere and that will not change in the near future. Tools like Midjourney become faster and better in a short time. That data is there. AI is here and AI will stay. See this article as a good time to think about whether you can do something with it yourself. The world is changing at lightning speed and I’m curious to see how that will be in five years. Are you hooking up in the field of AI or are you letting it pass you by completely? The choice is yours.
Source: Frankwatching by www.frankwatching.com.
*The article has been translated based on the content of Frankwatching by www.frankwatching.com. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!
*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.
*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!