Celebrating 51 Years of Life
I’m 51 this year, and at the same time Tobyblog turns 11. Today there won’t be any celebrating to speak of, it being Good Friday. My birthday frequently seems to fall on the holy days around Easter (or Pascha as we now refer to it), so it’s often overshadowed but never entirely forgotten. Sunday Kassi will be baking me a special blackberry cake that I expect will rival anything she’s ever made for me in the past. In fact, I think I should make a photo gallery of all my previous year’s birthday cakes. There have been some outstanding ones. A old photo of the blueberry cake she made for me one year still stands as my most popular photo on Flickr for some reason.
AI has been a recurring topic of my conversations of the last few weeks, spurred on by the sudden popularity of ChatGPT’s image creation capabilities and its ability to mimic the style of popular artists, a feature that has sparked a fair amount of controversy both online and closer to home, with one of my close friends recently confiding to me that he took some heat over his frequent posts of family photos rendered in the fashionable Ghibli style. The whole subject is of some interest to me. I’ve known about Chat’s image rendering for a little while now and have played around with it some, but after discovering that they restored internet access for “deep research” queries I decided to turn it loose on this blog, asking Chat to do a deep dive on Tobyblog.com and generate an image of me that would tie together all the loose themes of my posts and capture the essence of what this site is all about. The result of that is the image posted above.
I won’t pretend that’s an accurate rendering of me personally, but as a kind of badge for the site I don’t think it’s all that bad.

A request for a more realistic, less idealized image result in the above, which does what I asked, but without the compositional flare of the first image. “I” also look frankly a little more pathetic in that image somehow. I won’t be using that one.

This one is completely unusable, and without real personality, but it does look like it could go on a sales brochure or something. All I did was ask it to modify the first image to remove the facial hair. Apparently it wasn’t able to do that. I’ve had much better luck asking it to modify the image just rendered. At any rate, I included it to demonstrate the actual output from ChatGPT. This is kind of a default style, I’ve noticed. Flat, warm colors, illustrative. It’s essentially just Midjourney or Stable Diffusion behind the scenes.

Prior to this experiment, this is how I typically used Chat’s rendering abilities — to create images of cats to amuse my cousin Carl.