Boredom: Is AI ready for Prime Time?

Stable Diffusion. Absolute Reality v1.81. Photoshop.

Boredom is a powerful tool.

In fact I wish my kids were bored more. They can't stand having "nothing to do". I suppose that makes me old. I remember entire summers where we had "nothing to do". We made up games like bowling with random objects and a soccer ball. I remember riding my bike for what felt like hours in the lobby of our apartment building (because it was too hot outside) and making up stories in my head about time travel and doing great things. I remember wondering what it'd be like to have to escape an apartment fire or an armed gunman from our 9th floor unit which is to say that I imagined having to climb around the railing of our balcony to escape into our neighbors' units.

Thank God for boredom.

I recently spoke of boredom with an old friend and said, "I haven't been bored since 1997". That's about the time that I got on The Internet. Prior to that I was already on bulletin boards services and prior that we already had direct dial-up where we played games like Command and Conquer (as long as we didn't get any incoming calls or random computer crashes or loss of connection). That was the time when "shit didn't always worked" and I credit much of my aptitude for suffering to that era when one should expect things to break and not function as designed. These days iPhones and even PCs are way too reliable. Even Photoshop built-in a autosave function so you no longer lose 5 hours of an edit to some random crash (which wasn't even that long ago, maybe like 8-9 years ago?). It's amazing how things don't break anymore so long as you take reasonably good care of your stuff.

I think that as "things" have gotten easier people have gotten "lazier" and less resilient. It's like that scene from WALL-E where everyone's fat because no one has to move anymore... that and they're stuffing their faces with milkshakes. In all seriousness though, I've found my own children's patience or tolerance for technological failures to be appalling. I mean they literally fall apart if they can't instantly log into Roblox or can't load a YouTube video. Sure, part of it is that they have limited screen time so they're freaking out over wasting time on troubleshooting and not getting to play. But come on. This shit didn't always work. Hell, it used to rarely work.

But enough about them, let's talk about me. The last several weeks have been a turning point in my relationship with AI. The first few weeks with Stable Diffusion were great. It was eye-opening how powerful this new tool was with its ability to create images on demand given the right prompts, add-ons (extensions), and knowledge of how things worked. I was truly stunned by the on-demand nature of all the AI image generators. It made short work of stuff that used to require a LOT of work to achieve. So the first few weeks I was head over heels in love with this new tool like a new love interest. Then the excitement fades, the blinders come off and you see it for what it really is. Just another tool. And at the moment a fairly limited tool. Can't draw hands and feet. SDXL broke a lot of extensions including but not limited to my favorite tool (ControlNet), and it seems like we're back at square one with the community needing to develop the tools for SDXL that they once created for SD1.5 to just get back to where we were before SDXL dropped.

One of my greatest grievances with Stable Diffusion is not being able to achieve a dutch angle with any of the XL models. That's where the camera turns so that the subject is no longer upright but rather diagonal to the frame. I used to do this a lot not because I was trying to create disorientation (as cinematographers do) but rather because my models were super tall/long and needed to fit more of them into the frame. But as I've tried with SDXL models, so far it can not be done and it's the proverbial "straw that breaks the camel's back". I went from being totally enamored with Stable Diffusion to suddenly seeing all of the limitations with what it *can't* achieve. It's like going from glass half-full to glass half-empty. While I understand that's a perspective problem (pun intended), the timing is coinciding with loads of other challenges that turn me off from making pictures particularly of the AI kind.

And I'm starting to understand why people gripe at the fact that "it's not real". For me it's not so much "it's not real" as it is "it's not real enough". I see through the very translucent facade that is AI-generated imagery and conclude that it's so very not ready for prime time. It's jenky like Windows 95. Like a house made of popsicle sticks held-together by Elmer's Glue. It's a 5-year old doing magic tricks and ruining the illusion because he/she hasn't doesn't have enough hand-eye coordination to pull it off but yet you still have to clap and tell them how amazing it was.

And it's pissing me off... not to mention making me really bored.

Look. I stand by everything I said about AI. It's not not true just because I'm bored. This post is simply to point out the chinks in the armor. To draw attention to all the little things that AI thus far does so poorly that it requires massive amounts of pre-production or post-production to get right. If that's the case then the value of the tool isn't nearly as high as I'd personally hoped for. AI is not Jesus. It can barely give us all five fingers much less save the world. Maybe I'm asking for too much. Maybe it's just a matter of time.

But maybe it's just not ready yet either.