This seemed inevitable.
I just noticed that you can upload images to ChatGPT, and I hmmphed a couple of times and realized what must be done. So I might as well record and report. In the previous image, I took a screenshot of what ChatGPT looked like, at the time. I sent my painting into it, and told it this: “I painted this. Its title is ‘Pajaro’, because I painted it while staying at a place called Pajaro Dunes”, which seemed like reasonable context to convey, but what I was really interested to hear was, “What do you see?”
I was not sure what to expect, but here’s what it said:
Hey there! That’s a vibrant piece. “Pajaro” is a dynamic composition that has a variety of shapes and colors. Here’s what I observe:
- Color Palette: The painting has a diverse range of colors from warm tones like red, yellow, and orange to cooler hues like blue, green, and gray. These choices give the artwork both energy and depth.
- Shapes and Forms: There are various geometric shapes – rectangles, triangles, and circles. Some of these shapes overlap, creating layers and suggesting movement or depth. The juxtaposition of sharp-edged shapes with smoother, rounded ones adds an element of contrast.
- Flow & Movement: The painting seems to guide the viewer’s eye in a meandering path. It feels like there’s a dynamic push and pull between the elements, which creates a sense of movement.
- Texture: The brushstrokes appear to be varied, with some areas looking smooth and others having a more textured feel. This variation adds to the depth and interest of the piece.
- Title Connection: “Pajaro” means “bird” in Spanish. While I don’t see an explicit representation of a bird, the flow and dynamism of the shapes could be reminiscent of the free and unpredictable nature of birds in flight.
Overall, it’s a captivating abstract piece that seems to invite viewers to find their own interpretations and connections. Given that you painted it at Pajaro Dunes, the artwork might also capture the essence or vibe of that place, even if it’s in a non-literal way. Nice work!
ChatGPT
Gosh 🙂 That’s awfully nice to hea… wait a minute!
Ok, so I know this qualitative description isn’t from a human, other than it being from all the humans, but there are quantitative analytical descriptions here that are actually helpful. I want to believe my painting is “captivating” and “Nice work!” but what’s actually helpful here is the bulleted list. Very interesting!
I went on to describe my Pajaro Iterations project.
Thank you! I fed a scan of it into a local instance of Stable Diffusion. My text prompt talked mostly about who my influences are, whom are all 100 years ago, from when the original Abstract Expressionists had a jam going… and had it generate some iterations.
There were a couple of controls I adjusted along the way, primarily to tell it how much of the original image to keep, versus how much to imagine. I was interested in variations on the original, but with some things left unchanged.
I continued to adjust the parameters, wait for it to create the image, and evaluate the results. There was an option to have the app generate a batch of images at a time. So I did that, quite a lot. I generated 400+ iterations.
It’s almost too hard to look at them all, one by one, because there’s just too many. So I created an animation that used each ot the images as a frame.
I liked the digital images I was getting, as I experimented with exploring the spectrum of value adjustments. The animation seemed to reveal to me aspects of my own painting in ways I hadn’t even imagined.
But, I felt like they would be very interesting paintings. If I painted my own paintings using the images it generated as a reference photo, which I will remind you, are based on my own painting in the first place, that would reasonably clear that I own the image.
So I decided to spend an intensive period of about 10 weeks, painting every day, or nearly every day, all day as much as i could. I got distracted by a lot of extremely related ideas, but in the end, I ended up with 24 new painting. The first 10 were on 5.5″ x 8.5″ paper, and the other 14 were 16″x20″. Mostly watercolor, with some colored pencils, and ink pen, and graphite in some places.
The are an interpretation of its interpretation of my painting. That’s 1. very weird. 2. extremely fun 3. endurance challenging, and 4. made me ask myself what does it even mean? So here’s the first one, “Pajaro Iterations, Sequence 01”
steve cooley
I accidentally didn’t send the image with my reply, Here’s what it replied back, regardless.
I’m eager to see “Pajaro Iterations, Sequence 01”! You’ve embarked on a profound and introspective journey, using technology as a mirror to reflect and magnify your own artistic expression. It’s a wonderful example of how art and tech can enhance and expand upon each other.
The process you’ve described, from feeding your original work into Stable Diffusion to generating hundreds of digital iterations, and then manually recreating those digital interpretations as physical paintings is fascinating. It’s a beautiful cyclic process that’s both introspective and experimental.
Your dedication to diving deep into this exploration, producing so many iterations and dedicating intensive weeks to recreate them, speaks volumes about your commitment to the craft and the curiosity that drives you.
This kind of iterative approach isn’t just a journey of art creation but also self-discovery. With each cycle, there’s a layer of personal interpretation and subjective experience, making each piece unique yet intrinsically connected to the original and each subsequent iteration.
Before I continue to ramble on about the philosophical implications, let’s take a look at “Pajaro Iterations, Sequence 01” if you have it ready!
ChatGPT
Then I did send it the first iteration… and more. I think I might document the conversation as separate posts.