I recently collaborated with AI to create a brand illustration system. In this case study, I will explain the process of building a brand illustration system using AI. I will cover everything from generating a rasterized image from mid-Journey to converting it into a usable vector image that can be imported into Adobe Illustrator. Additionally, I will discuss the best practices for extracting foreground, midground, background, and main character objects so that they can be used across different designs for marketing purposes.
While building the brand illustration system through Mid Journey, I faced several challenges. I had to prompt the image generator for hours to get the desired results. I encountered issues like random animals floating in the air, disjointed limbs, backward bodies, and oversized foreground, midground, and background elements. After a while, I decided to vectorize the rasterized images being generated by mid-Journey to get better results. I then manually went into Adobe Illustrator to make the necessary corrections instead of relying on the image generator to make the corrections.
The Good, the Bad and the Ugly
Prompting AI and Image Generation
When starting this process I noticed right away that there would be issues with using the images as is and that I would need to use Adobe Express to convert the images to SVG files and go in and edit the errors myself in Adobe Illustrator. In this example, the obvious is some sort of floating reindeer ended up getting randomly generated along with the features of the cactus star looking like a different style than the rest of the creative.
Prompt: _Western_UIUX_warm_colors_Moab_cowboys_cactus_tumble_weed
I had spent a considerable amount of time prompting the AI image generator and it was frustrating at times the results that where produced as I kept refining my prompts and getting completely random results for hours. In this example the proportions of the cowboy hat is too large and somewhat comical.
Prompt: _beautiful_website_in_style_of_western_impressionist_cactus_cowboy_
This example was a common accurance and something happened in the image generation that distorted the head or limbs of an animal. This helped me to understand that the images being generated are being somewhat smashed together in sometimes pretty random ways like in this example the front of the horses head has become a mashup up of a wagon and random things that I can't quite make out.
Prompt: _Western_UIUX_warm_colors_Moab_cowboys_cactus_tumble_weed
Finally I had found an image that I thought would work. The horses limbs looked distorted or even missing and would need to be recreated after converting the file into SVG. The proportions of the cowboys were also too alongated and would need to be corrected as well in Illustrator. The colors would also have to be changed to create more contrast within the creative.
Prompt: _Western_UIUX_warm_colors_Moab_cowboys_cactus_stylize_
Converting to SVG
Separating elements, recoloring, and simplifying creative in Adobe Illustrator
After converting the AI generated file from a rasterized image to a SVG file using Adobe Express. I then brought the SVG file into Adobe Illustrator and using my shape builder tool I simplified the layers for recoloring and adding the desired warmth that I wanted in the creative. Stripping the landscape down created a feeling of an open canvas to place objects on and create the desired narative in the creative.
Once everything was recolored I address the issues with both the human and horse anatomy. To achieve this, I referred to some horse images and recreated them in the same style as the creative design.
Creating the object library
The completed system
Brand Illustration Services
Campaign assets
In conclusion, developing a brand illustration system with the help of AI presented challenges. However, the end result is a stunning and functional system that can engage users across marketing efforts.