Remember the days when you had to take a picture with your own camera to get a nice picture for your website? Skip a few years and you’ll have Photoshop, search engines, and stock photo libraries at the click of a button.
Now, to make things even easier and faster, the industry has evolved once again. Artificial intelligence can now generate images for you based on simple text input.
In a short period of time, AI-assisted design is starting to drive more and more creative processes across several industries, including marketing, design, and programming.
But are the results worth all the hype? Find out for yourself.
In this article, we discuss how a design-driven company is starting to incorporate artificial intelligence into their online service.
An Australian based design company, Canva, created an app that allows you to describe an image and then create a simple AI using the app. The application is called Text on image.
We’ll show you how it works, what you can do, and then discuss the use of artificial intelligence in creative work, the pitfalls and potential.
How can I use it Text on image in Kanva
When composing an original work, the first thing you can do is brainstorm, sketch a sketch, or create an outline. With artificial intelligence, you can simply think about how to describe the image you want in plain text and then write it into the program.
Start with a few simple commands and see what you get. Constant revision of the text and evaluation of the result is the cycle of creation of artificial intelligence art. Rinse and repeat.
Canva’s Text to Image app uses a huge library of images as training data to generate output in three main styles: concept art, photography or painting. Once you’ve chosen a style, you can simply type in a text command to generate your image.
For example, if you want an image of a colorful bush with flowers, you can simply type “colorful bush with flowers.” See what you get.
You may be satisfied with the results you get on the first try. But if not, remember you can be more specific. For example, you can write “Colorful bush with flowers at sunset in watercolor style.” The output of this command should be more like what you’d expect. Rinse and repeat.
What kind of image formats are supported?
You can create AI art for different formats.
You can choose a custom size, or choose from popular social media formats (eg Facebook, Instagram), posters, videos, invoices, infographics, proposals, book covers and postcards – just to name a few.
A Text-to-Image application requires some specificity and refinement. This is at a stage where these creations are still relatively simple and may require human intervention to meet specific requirements.
Don’t be surprised if the AI produces weird and strange images.
Artificial intelligence in creative work – good or bad?
The text-to-image option is a great way to get instant graphic gratification, but it may not always meet the exacting standards you’d expect from a trained artist.
So is AI good or bad for creative work?
Artificial intelligence imitates the art that humans want to see, but not always exactly. He can interpret and generate aesthetically pleasing compositions in broad strokes, but not so much in details.
AI generated image of “Fantasy World Concept Art, Middle Ages”
If you want something uniquely beautiful or specific, you may still want to work with a trained human artist.
Concept art for CD Projekt RED’s medieval fantasy world of The Witcher 3: Blood and Wine by Andrzej Dybowski
For corporations, this could lead to a reduction in skilled workers as AI makes it faster and easier to get images. Concept art that might have taken weeks to build can now be created in seconds. But on the other hand, creating consistent high-quality AI art is an art in itself.
When choosing what kind of AI you need to do the job, you weigh speed against consistent quality.
You can find many examples of artificial intelligence being used in creative endeavors, but they are always related to human work and intervention at the time. Using AI to create elements in an overall design is quickly becoming the norm, but you still can’t use it to create everything—at least not yet.