Artificial intelligences are evolving by leaps and bounds. And not only because they are becoming much more predictive and methodical, but also because they are gradually reaching users so that anyone can have access to it to perform their function. Today we can find many Artificial Intelligences for many types of activity. But one of the most striking, currently, is the AI that allows us to generate images from a text that we pass to it.
It is very likely that in recent weeks we have heard about DALL-E. This is one of the best-known artificial intelligences, whose main purpose is to create images from a text that we indicate and, little by little, improve the results and become more professional. Depending on what we ask of it, the results can be more or less complex, reaching, in certain requests, images that border on the unimaginable.
However, being the best known does not mean that it is the best. And it is that DALL-E has come up with a very tough rival that, without a doubt, is going to steal the spotlight. That’s how it is Midjourney.
Generate AI images with Midjourney
The first thing to keep in mind is that, in order to use this AI to generate images, you need to have an account. Discord. This is because, unlike others, its developers have opted for a bot for this platform to send requests and receive drawings. This also allows us to use it at any time and in any place, even from the mobile.
We must enter their website, and the first thing will be to “Sign In” to join our Discord account and give permission to the bot.
Once done, the next step will be to click on “Join the beta” to join the bot channel on Discord and start using the program. Once joined, we will see a long list of channels. The ones that interest us are the «newbies» channels, since they are where we will be able to send the image generation requests.
We enter any of them, write “/ imagine” in the chat bar, and then the image we want to generate by AI. We, for example, are going to ask that we want an image of Pikachu surfing on a pizza. We write it (preferably in English) and press “Enter” to process it.
The bot will take around 60 seconds to send us 4 images, which we can scale, and even vary, using the buttons that appear below them. We can even tell the bot that we want to regenerate all the images in case we don’t like any of them.
In our case, the generated image was this:
Meanwhile, the same request in DALL-E has returned us these related images.
Everyone can decide which image they like best, the one that best suits what they were looking for.
Limitations
At the moment, like most AIs, Midjourney is in beta. Of course, it seems that it is somewhat more advanced in development than its rivals, although it does not achieve absolute perfection either. For free, we will be able to generate up to 25 images, and we will not be able to use the advanced parameters. In case we want to take full advantage of it, we will have to choose one of the payment methods:
- Basic: 10 dollars per month, which allows us to use the GPU of the servers for up to 200 minutes.
- Standard: 30 dollars per month, and allows us to use the GPU for up to 900 minutes (15 hours) and allows us to use the “/ relax” parameter to have more time to create the images.
- Private Visibility: A module that, for $20 a month plus, keeps our creations private instead of showing them in chat to everyone.
- Incremental Billing: If we use up the plan, we can pay 4 dollars for each additional hour of use.
