Artificial Intelligence (AI) has become extremely popular lately, but what is less known is how you can use it effectively. AI can be used for multiple purposes such as generating texts, images and videos. This article dives into the most important details you need to know to generate images effectively.
To generate a perfect AI image you must use a combination of keywords that generally consist of the following components: subject, medium, style, color and light. For example: Mountainous landscape with a river, village background, starry night, Ruby Red, moon light, style Vincent van Gogh.
Generating images with AI is always done via a text command. A prompt generally consists of the following parts:
Not every item needs to be used in the prompt. A few examples of the effects can be seen below:
Prompt: Landscape
This example uses a very simple prompt containing only one word. The AI has generated a landscape, but because no further keywords were provided, the AI had to come up with many details itself.
The AI is trained with images from the internet. This allows the AI to only generate images with keywords that are common on the internet. So the name of a celebrity can be used to generate an image of that person, but if you use the name of your neighbor, for example, you will not get images that look like your neighbor. The same applies to logos of unknown brands and objects of which there are few images on the internet.
Prompt: Mountainous landscape with a river, village background
In this example, many more keywords were used in the prompt, so the AI had to come up with fewer details. The more keywords you add, the more you can direct the AI to create what you have in mind. However, if you add too many keywords, you may find that the AI cannot process everything and therefore does not add certain things.
Prompt: Mountainous landscape with a river, village background,
pencil sketch
By adding a medium you can also change a lot in the image. For example, one medium could be 'pencil sketch', but other options include: oil painting, watercolour painting, pen and ink, chalk art, and photography.
Prompt: Mountainous landscape with a river, village background, starry night,
style Vincent van Gogh
By adding the style of an artist, for example, you can have the AI generate an image in that specific style. Other style examples are: fantasy, sci-fi and realistic.
Prompt: Mountainous landscape with a river, village background, starry night,
ruby red,
style Vincent van Gogh
By mentioning a color you can ensure that the AI tries to bring out this color in the image.
Prompt: Mountainous landscape with a river, village background, starry night, ruby red,
moonlight,
style Vincent van Gogh
By mentioning lighting conditions such as moonlight you can ensure that the image is illuminated in a certain way.
In the negative prompt you can enter keywords that you do not want to see in the image. The rules for the negative prompt actually work exactly the same as the rules for the normal prompt, except that it concerns things you don't want to see in the image. So don't write 'no bridge', just write 'bridge'.
Prompt: Mountainous landscape with a river, village background, starry night, ruby red, moonlight, style Vincent van Gogh
Negative prompt:
bridge
As you can see, no bridge has been added to the image this time.
You can adjust the weight of a keyword using the following syntax: (keyword: factor). The factor is a value that if it is above 1 the keyword is more important and if it is below 1 it is less important.
For example, if you want to see more of a village in the image, you can do the following in the keywords: (village:1.9). This indicates that you want to see more of a village by a factor of 1.9.
Prompt: Mountainous landscape with a river,
(village:1.9)
background, starry night, ruby red, moonlight, style Vincent van Gogh
Now that you know exactly how to generate the perfect AI images, you can try it yourself with our own AI image generator.