A.I. Art Poster
Ride on Time - Yamashita Tatsuro
(1980)
Plum Estate, Kameido - Utagawa Hiroshige
(1857)
Concept
I came up with the idea for this poster while listening to the 80s pop album, Ride on Time, and mindlessly staring at the Hiroshige print on my wall. I decided to attempt to use A.I. to create an alternate cover for the album in the style of Ukiyo-e woodblock prints, hoping to show an amusing amalgamation of art as it appears across the centuries (note the album title!)
Experimentation
I began with the Midjourney and Dalle-2 A.I. but was disappointed in the lack of resemblance to Yamashita Tatsuro in the figures they were generating. So, I turned to the Dreambooth A.I. with the recently implemented "stable diffusion" which allowed me to supply a set of input images from which it would generate a model that would force certain patterns to appear in all generated art (in this case, the individual's face and body/poses/outfit).
It was easy to generate hundreds of options with varying text input (specifying different mediums, compositions, colors, Ukiyo-e artists, etc.) and I chose one to work from. I then extended that image at it's borders, cleaned up blemishes and colors and replaced the face in photoshop, and added the album title and artist name, adjusting sizes and values to maximize balance and imitate original Ukiyo-e art.
Process and Printing
In printing this project, I decided to use key focal points, such as the face and text, as my registration marks. The goal was to maximize poster size by eliminating extraneous marks and to get maximum "focus" at these key points. Thankfully, this appears to have consistently worked quite well and gave exactly those results.
Three of the layers, C, K, and M, went down without issue, aside from some gain/pooling at the edges. Yellow, however, consistently failed to print along one edge, despite cleaning the screen and restarting. I also failed to clean the light table completely, resulting in some extra blemishes on all layers.
Final Print
The final poster, printing issues aside, is nice to look at and (silver lining) is a nice, rough, analogue alternative to the original, digital composition. The art is appropriate at this size (18x18") and appears as though it would read equally well at album cover size. There are some elements I would like to have pushed further, such as the use of gradients, the contrast between the blues, and the discrepancy in the two text sizes, but the overall effect is positive. As someone who's has a mild interest in and knowledge of woodblock printing and who is a big fan of this musical artist, I feel that this successfully evokes a familiarity with both. In other words, the A.I. did a great job at giving life to the idea in my head.
