Your Kool-Aid man made me want to try for a realistic one, but the couple I tried failed. (Plus Isaac Asimov is supposed to be on the Galapagos Giant Tortoise, not the Commodore Pet!)
So I went a different route.
For those first two, I used Sora with uploaded reference photos for Isaac and the Pet. In the name of experimentation I tried the prompt again in ChatGPT with no reference images.
Isaac looks at least as good. The Commodore doesn’t have the built-in tape recorder and the monitor is pretty small, but I can live with that. Neither version created him staring at the screen without noticing the terminator as I asked for but I can’t say I hate how they did turn out.
The prompt
Summary
An over-the-shoulder shot seen from just behind a T-800 endoskeleton that is facing away from the camera. In the background Isaac Asimov is sitting in a battered swivel office chair staring at the screen of a Commodore Pet computer on the the cluttered desk in front of him in his book-filled home office. He is too distracted to notice the dangerous robot looming behind him. Isaac should be in sharp focus and the T-800 that is in the foreground should be slightly out of focus. Photorealistic Kodachrome dslr photo with shallow dof and forced perspective.
I noticed that back with your BJ and the Bear one earlier. In the first image, the one from the perspective of inside the cab, their truck is perpendicular to the road rather than driving on it.
That’s because that’s a later model PET, the CBM 4016. It had a full-sized keyboard instead of the ridiculous square-grid calculator-style keyboard, but had to give up the integrated tape deck to make room.
Close, but you are linking to the “album”. You want to right click and link to “image location” or whatever your browser calls it. You’ll get a direct link to the image file with the .jpg/.png image.
I asked for an illustration of this short vignette generated by OpenAI’s “Monday” April Fool’s joke (still available in custom GPTs I think) and I was surprised it got the essence of this stupid little story correct in one shot.
I had largely disappointing experiments today. Yesterday I worked on an idea for a person with Sarracenia pitcher plants growing out of their head. I never got quite the body dismorphia horror I was originally imagining, but I git a few decent images. Then I decided to pair her with my claw “character” in some vacation photos.
First I put them in front of the Chicago Bean and asked for the images of their backs to be reflected in the surface. I got the Bean, but no reflections. And it insists on distorting the shape of Claw Guy or giving him legs even though I ask it not to. (I asked for a “touristy tshirt” without specifying text.)
Failing with the Bean, I picked Akihabara next. For that I wanted other people on the street staring at them (including someone in maid cosplay) and an Engrish tshirt. It mangled that one so I dumped the crowd. Then asked for the Engrish to be more colorful. Then asked for a chibi mascot character.
The most successful discovery in those images being, I think, the ability to invent entirely cromulent Engrish (the most advanced description being “confusing Engrish text in colorful fonts with an odd chibi animal mascot”). I had to include every example.
