Midjourney is the most popular word artneural network. Her work is so beautiful that it can win an art competition. If you've seen a music album cover, a quirky movie art, or another image clearly generated by a neural network, it's most likely made with Midjourney. But this neural network has several big disadvantages: starting from March 30, it has ceased to be free, the price of a paid subscription is from $10, and payment using Russian bank cards is not available for those who wish. Therefore, many people are trying to find a free analogue of Midjourney that is not inferior in quality. I found some good options and I want to share with you - perhaps you will find the “same” neural network that can turn all your ideas into reality for free.
To illustrate what can be expected fromof the neural networks listed below, I gave them the same requests to generate images. They are diverse and moderately complex - neural networks should definitely cope with the creation of such images.
Here are three made-up queries:
- a vintage car drives through the desert on Mars;
- a black cat sits in a chair in a hat and eats pancakes;
- a girl with purple eyes stands on the field and holds a gun.
In Russian services, it is better to write requests in Russian, and in foreign ones - in English. Verified by personal experience.
Another popular neural network in 2023 is ChatGPT, which composes texts. We have a selection of his unexpected abilities.
Content
- 1 Kandinsky 2.1 neural network from Sberbank
- 2 Neural network "Masterpiece" from Yandex
- 3 Bing neural network that draws pictures by words
- 4 Playground AI service for creating pictures
- 5 Neural network Dream by WOMBO
Kandinsky 2.1 neural network from Sberbank
In April 2023, Sber introducedneural network Kandinsky 2.1. This is a new version of Kandinsky 2.0 launched a year earlier, which contained 2 billion parameters. The updated version will include 3.3 billion parameters and, accordingly, works much better.
One of the features of the Kandinsky 2.1 neural network is support for more than 100 languages - the popular Midjourney could not boast of this.
Kandinsky 2.1 also supports four modes of operation:
- image generation by text description;
- mixing two pictures into one;
- creating different versions of the same image;
- image completion (the neural network complements the pictures so that you can see what is “behind the scenes”.
There are several ways to useneural network Kandinsky 2.1. It is best to use the dedicated Fusion Brain site, because it is as clear as possible: in the upper left corner you can select different tools (for example, to remove unsuccessful image fragments and then replace them with another query, and in the lower left you can select a drawing style.
But there are other ways to run Kandinsky 2.1:
- official bot in Telegram;
- Rudalle website (there are other neural networks from Sberbank);
- skill "Turn on the artist" in applications and devices with the voice assistant "Salute".
IMPORTANT: sometimes the Kandinsky 2.1 neural network does not work and usually this is due to high load. If there are problems, it is best to try again in a few minutes.
I tried to enter the queries mentioned above, and here are the results that the Kandinsky 2.1 neural network gave me.
AGAIN I UNDERSTAND: The Kandinsky 2.1 neural network understands more than 100 languages, so queries can be written in any way you like. Personally, I used Russian.
Using a neural network, I found several advantages over Midjourney in Kandinsky 2.1:
- the neural network is completely free and has no restrictions;
- images are created quickly, from 30 seconds to 2 minutes;
- there are several ways to use the neural network, ranging from an intuitive site to a Telegram bot;
- with the right query, the results are no worse than Midjourney;
- Kandinsky 2.1 supports several modes of operation and styles;
- It is possible to specify the size of the image.
See also: How do neural networks affect the climate and the environment?
Neural network "Masterpiece" from Yandex
At the beginning of April 2023, Yandexintroduced the Masterpiece neural network, which is available as an application for Android and iOS. At the time of this writing, this is a beta version of the service, but despite this, the neural network can surprise.
To use Masterpiece, you need to downloadit through Google Play or the App Store and launch it on your smartphone or tablet. Then everything is extremely simple: you need to log in to Yandex services, click on the “plus” at the top of the screen, enter a request and click on the “Generate” button.
Here is what I managed to do by entering the queries listed at the beginning of the article. At first, the image will be in low resolution, the quality increases a few minutes after the image is created.
In the Masterpiece application, you can publish your work and see what others have done. This is no less exciting activity - sometimes the results of other people's work cause a smile.
Here are the most obvious advantages of Masterpiece:
- the application is completely free and available on any smartphone;
- The application interface is extremely simple;
- creating images takes a few seconds, everything is very fast;
- There is an opportunity to share your work and watch others.
Did you notice? As of April 1, 2023, we have prepared an articlewhich was written entirely using the ChatGPT neural network. It talks about a scientific discovery that allowed scientists to extend life to 125 years. All this is not true, read the refutation in the article “What article on Hi-News.ru was written by the ChatGPT neural network”.
Bing neural network that draws pictures by words
At the end of March 2023, Microsoftopened for everyone access to the Bing search engine with the function of creating pictures from a text description. To use it, you need to turn on the VPN, sign in to Microsoft services and change your location to the United States. After that, you need to go to the Image Creator website - that's it, you can generate images.
Details on how to use all this, Itold in the article “How to use the Bing neural network to create text and pictures”. In the same place, I talked about the advantages of Bing over Midjourney, so for now let's just look at the results of his work. I think it turned out very well.
Here are the obvious advantages of the Bing neural network, in addition to free access:
- fast image creation within seconds;
- extremely clear interface;
- the presence of history on the right side of the screen, allowing you to look at previous work.
While writing this article, I found a very interesting piece of news from 2017. It says that Microsoft is trying to inspirelife in the Bing search engine with the help of artificial intelligence. It seems that work on this has been going on for more than five years, and the fir tree has finally been achieved. Do you use the Bing search engine? Write in our Telegram chat. After the introduction of the neural network, I began to use it often.
Playground AI service for creating pictures
Perhaps the Playground AI service is the mosta little-known analogue of Midjourney. on our list. But in vain, because although it is not completely free, it allows you to create up to 1000 images per day with a resolution of up to 1024x1024 pixels using the Stable Diffusion neural network (there is support for DALL-E 2, but it is paid). For the average user, the Stable Diffusion features and the set limit are more than enough.
The Playground AI interface is extremely clear.On the left is a column with a choice of drawing style, a field for entering a query and choosing a picture to use as an example. On the left, you can select the neural network version, image size, and other options.
Here are the images we managed to create using Playground AI with the Stable Diffusion 5.1 neural network.
The advantages of Playground AI are as follows:
- limit of 1000 high-resolution images per day;
- clear interface;
- a large selection of different styles.
You will be interested: Why the neural network for writing essays ChatGPT cannot be banned in schools
Neural network Dream by WOMBO
Finally, it is worth mentioning the Dream neural network - itas simple as possible, but at the same time gives good results. To create an image, you need to enter a description in English, select a style and name it on the Create button. Everything, the drawing is ready.
In fairness, it should be noted that often it generates too cartoonish pictures. Here's what he came up with on my request.
The advantages of Dream are obvious:
- the most simple interface;
- popular drawing styles are completely free;
- images are created in 10-15 seconds;
- There is no limit on the number of images created.
Liked the article? We have many of them and there will be more! In order not to miss them, subscribe to our channels in Zen and Telegram.
There are many neural networks for creating pictures - eachfind something to your liking. If the options listed above are not enough, check out another selection of neural networks for generating images. Among them, too, there are interesting options.