Many of us, who have a creative imagination and ambition, but lack the skills and talent to etch it masterfully on a canvas, have had no real way to express that creativity in the real world. That is, until very recently. Enter AI. This is the age of artificial intelligence and it has brought some very interesting AI tools that can help you bring your imagination to life without lifting a paintbrush. Midjourney is one such AI platform that turns natural language text prompts into mesmerizing images. But to use this tool to its full potential requires learning about its various features and keywords and the right way to write a prompt. If you have not seen the same success as some of the viral photos on social media, you can learn from our Midjourney prompts guide and improve the quality of the images drastically. Let us take a look.
What is Midjourney
For the unaware, Midjourney is a generative AI platform created by a San Francisco-based independent research lab Midjourney, Inc. Midjourney generates images from natural language descriptions, called “prompts”, similar to OpenAI’s DALL-E and BigSleep’s Imagen.
What are prompts?
Prompts, as explained above, are simply a description of the artwork that you are trying to create using Midjourney. For example, “A realistic visual of a starry night with a full moon against a backdrop of mountains”. The idea with the prompts is to be as descriptive and as specific as possible to create a piece of art that is close to what you would have wanted to see.
Writing better prompts
One of the reasons many Midjourney users do not see a good result with the images they create is that the prompts are either not descriptive enough, or lack certain keywords that help the AI understand what the user exactly wants. Additionally, to really unlock the AI’s full potential, you need to make use of the parameters in the prompts. So what are all these and how to use them? Let us break it down.
Let us continue the example of a prompt we gave above which was “A realistic visual of a starry night with a full moon against a backdrop of mountains”. An inexperienced user would end up using a simple prompt such as “a night sky with mountains”. However, the problem with the second prompt is that it has just two keywords ‘sky’ and ‘mountains’. The AI will treat them as constants and will fill the rest randomly, which can make the end result look vastly different than what you had in mind.
In the first prompt, there are four keywords which are ‘realistic visual’, ‘starry night’, ‘full moon’ and ‘backdrop of mountains’. With more keywords, AI will focus on adding these elements to the image, which will give you a final result much closer to your expectations. However, there are even more ways to stylize your art and get an image that is exactly what you were looking for.
To make your art stand out, you can use a style type and give the AI a specific direction to work with. Some of these can be ‘Japanese anime’, ‘cyberpunk’, ‘Pixar movies’, ‘surreal’, and more. You can use any genre of art to experiment with.
You can also give the name of an artist to ask the AI to generate an image in their style. For example, “waving summer afternoon with a cloudy sky and a town road with people rushing to their homes in Raphael style”.
You can then choose the degree of stylization. An image with high stylization will look very close to the art style of an artist or the genre and a low stylization will only appear inspired by that style. To control your stylization output, use the format “- -s (number)”. For example, “a cyberpunk city with flying cars – -s 2000”.
Chaos is a depiction of abstraction of your art. This parameter can be added with the format “- -chaos (number)” and picking a number between 0-100.
You can also add a preferred resolution such as 4K, 8K, HD, etc.
This is a very important parameter to set. This determines the length and width of the image. If you do not specify an aspect ratio, Midjourney will produce a 1:1 image by default. But you can customize it for your usage preference by adding a specific number. Do note, you can only use the standard ratios like 3:2, 1:1, 4:3, 16:9, 1:3, 5:1, and more.
Pro tip – If you are a journalist looking to use images for your articles, use the aspect ratio of 16:9. For Instagram posts, you can select 4:5. Reels, TikToks, Stories, and more require an aspect ratio of 19:6.
Filtering words from a prompt
You can also tell the AI specifically to not add an element to the image by using the format “- -no (object)”. For example, “red car – -no wheels”, will create an image of a red car that does not have any wheels.
Adding Camera and Lens
One cool feature that you can make use of is adding a specific camera and lens to make the image appear as if it has been clicked using them. For example, you can write, ‘a playground in the middle of a city with kids playing Sony Alpha 7, ISO1980, Leica M’.
You can similarly specify a particular type of lighting condition such as cinematic lighting or volumetric lighting to give your artwork a better effect.
Use these tips and guidelines and we are sure you will be able to create some spectacular artworks that will amaze everyone.