AI
3. januar 2024
AI Image Generation and Its Potential
I know—you come across it everywhere: LinkedIn, the news—you name it! I am, of course, talking about AI—but in this post, I will focus on a branch you may not be as familiar with: image generation using an AI model. To be even more specific, I have used ‘Midjourney’ as my point of departure.
So what is it?
AI image generators are complex tools that use advanced technology to generate images. But how do they actually work?
As with ChatGPT (which I will cheekily assume you know), in simple terms it is machine learning trained in pattern recognition and characteristics based on the data it is trained on—images, in this case. It uses what it has learned to generate new images based on the prompts (descriptions) you provide.
Pros and Cons of AI Image Generators
As with any other technology, there are, of course, both advantages and disadvantages to AI image generation. I will try to cover the most prominent ones I have encountered.
One of the biggest advantages is time savings. Once you understand how to prompt your way to the result you have in mind, it can be done extremely quickly, depending on how many iterations you need. In most cases, you can have a solid starting point significantly faster than if you need a human to create it.
Following on from the above, there is, of course, also the financial aspect: when you do not need to outsource to a person who spends more time, you can typically get a good output in less time and at a considerably lower cost.
A third notable advantage is creativity—and this should in no way be understood as implying that a designer is not creative, but rather that you can prompt an AI to generate an image of almost anything you can think of. Through iterations and perhaps a few changes to the prompt, the likelihood of ending up with a usable result is relatively high.
And following on from the latter, we naturally arrive at some of the disadvantages of AI image generation. If you want a very realistic image (think a photograph) featuring, for example, a person, some challenges can arise: the wrong number of fingers on a hand, or an arm that is not quite positioned as it should be. That said, they have become much better at exactly the above.
In addition, another major challenge is the lack of control. You do not have much control over the result the AI generates from your prompt. You can, of course, try to adjust it—add things that MUST be in the image, or things that must NOT be in the image—and then hope the AI understands. Alternatively, you may have to start over and try a different way of phrasing your prompts.
Last but not least, there is also quite a bit of discussion about the ethics of using AI. However, this is a fairly complex question and not something I will go into further here.
How good is it, really?
The potential of AI image generation is enormous. This should not be read as designers becoming redundant—on the contrary, there is a high chance they will become even more valuable. With a tool like ‘Midjourney’—and yes, that is what it is: a tool—a designer can use AI to arrive at a rock-solid starting point much faster than if they had to draw it from scratch.
In addition, a designer can also use AI as a kind of sparring partner—always available and able to provide quite detailed suggestions in no time.
It can also produce visual elements that, for a designer, can be extremely resource- and time-consuming to create manually—and it can do so in a fraction of the time. And once the designer has cracked this code, they will also be able to significantly increase the volume of their deliverables without increasing their working hours. This also means they free up more time to handle tasks that may not be solvable by an AI—because yes, that type of task undoubtedly still exists.
In addition, there is the “creativity” that AI brings to the table, because even though designers are already known for their creative abilities, they can benefit greatly from the “suggestions” an AI can provide. It is also not unlikely that the AI will produce something that is “outside the box” and thus help the designer explore new perspectives.
What should you take away from this?
First and foremost, it is important to remember that it is a tool—and yes, a really great one—but nothing more than that. It will not replace humans 1:1.
It can be extremely helpful in a design process—perhaps you are a bit stuck, lack a little inspiration, or something else entirely.
It can save a lot of time, which in most contexts also means cost savings.
It has advantages and disadvantages, but in my view the advantages outweigh the disadvantages.
It is already being incorporated into existing software; a very good example is Photoshop, which in its beta version has AI integrated and can be used while you edit images.
I hope this has been easy to understand and that it may have given you at least a little desire to try your hand at it.
Thank you for reading!
