Using a test project as an example, I’d like to share some helpful ways for designers, artists, and other creatives to think about how to use AI-generated imagery in their creative process.
(View the final test project page here).
Background
Many creatives out there may be feeling apprehensive about AI to create art and platforms such as Midjourney. You may find yourself asking "when will this take my job?". Or you may be skeptical about the utility of AI in real world applications.
I for one have had all of these doubts, but after doing some investigation I'm now more open to the use of AI-generated art. While these tools may in fact impact the creative industry in unforeseen ways, I believe the potential for greater creativity that AI provides could create more possibilities than expected.
The Problem
So far, much of the AI-generated art we see shared online are examples of users attempting to create the most pristine fantastical images all within the confines of Midjourney or other AI image platforms.
While these images are fascinating, I’m afraid we’ll be missing the greater potential of these tools unless they can be used for real world applications.
Project Brief
I wanted to apply AI generated images into the context of a project. After some exploration, it seemed like a web landing page would be a great way to do this as it requires many aspects of design: layout, typography, graphic elements, and imagery. It also would require the images to support a greater narrative or structure.
I wanted to explore the following types of imagery common to many real world projects:
- Portraits
- Landscapes
- Detailed Patterns
- Graphic Symbols
I also wanted to the end results to fit together, tell an overall story, and not look too fantastical or specifically AI-generated.
Midjourney Process
I was curious to see if I could convey a mysterious story along a topic I've never designed for: knights. Why? It just sounded fun!
I began by providing Midjourney with prompts combining medieval imagery with varying styles from different eras: tin type photography, Caravaggio paintings, and black ink drawings. I was interested in how the platform could be used to combine disparate themes and styles together.
The figure below shows how I directed the Midjourney bot to create a visual asset by using prompts and iterating on a few selected images.
As you can see, just like art created solely by a human, you still need to provide guidance, using artistic influences and curiosity to curate and direct the process. Using this process of input, selection, and iteration, I generated many different images to use in the landing page.
Post Midjourney Process
The results were fascinating and unexpected, but I knew I'd need to modify the images to fit into a cohesive creative direction.
To create the final assets that were incorporated into the page design, I brought the exported images from Midjourney into Photoshop and enhanced them by cropping, removing backgrounds, deleting unnecessary elements or layering multiple images together. I also made color corrections or contrast adjustments as necessary to fit the overall web design, and in one case drew additional layers by hand.
I developed a design system that utilized geometric arches as a graphic concept, along with typography that referenced fiction book covers from the 1960s and 70s. A simple color palette of parchment, bright orange, black and white to tie the various images and graphics together in a bold look.
You can see an example of how all of these different elements were pulled together in the hero banner area of the landing page, shown below. What I find exciting is how one can use AI as a starting point for very different types of visual assets - even a graphic symbol such as the logo.
Final Design
The final outcome of the landing page design is much more interesting than I expected. It's definitely a much richer experience than had I used stock imagery. If this was a commercial project, hiring an illustrator to clean up and refine the art could push this even further. I can also imagine that these images could be used as starting points or inspirations for professional illustrations.
Below are some screenshots of the final layout. You can see the live page here.
Insights
This process of using Midjourney to create imagery was eye-opening, and helped to alleviate some concerns and I had with using AI as a tool for art. Here are a few takeaways I learned:
AI is a great tool for:
- Creative direction exploration
- Creating multiple images in a short amount of time
- Getting more artistic and unique images than stock libraries
- Finding Unexpected solutions
- Idea generation
AI is (so far) not so great for:
- Getting exactly what you have in mind
- Sourcing specific real objects or animals (tends to want to create mashups for everything)
What's important to note is that all of this still requires thought, experimentation, trial and error to select, edit, and place the images with the bigger picture of the overall design and narrative in mind. For me, that's ultimately what makes this process rewarding.
Conclusion
I’m looking forward to harness this technology for real projects where it makes sense and in ways that enhance the overall viewer experience. Already, it has helped me think more deeply about my own artistic influences and push me into new directions for creative expression.
Ultimately, my hope is that AI platforms can be harnessed by artists and creatives for positive outcomes: to bring about more artful experiences that connect with other humans in impactful ways.
