<aside> 📓
This is a demonstration, not a model answer. An invented student wrote it about a subject nobody in this class is working on, and the findings in it are made up for the example. Copy the level of detail, not the choices. If your prompt log looks like this one, something has gone wrong.
</aside>
The five sections below are the same five in the weekly entry template. Week 3 is one of the ★ weeks, so section 4 carries more weight here than it will in a normal week. Check This Week's Journal Focus for what your own week is pointed at.
I thought the hard part of the in-class lab would be getting the signage to look hand-painted instead of printed. That took two steps. The hard part turned out to be that every addition changed more than I asked for. Adding a medium changed the building, the weather and the camera all at once.
The homework went the other way. I expected two models made by two companies to disagree with each other. Most of what came back was the same.
| # | Prompt (exact text) | Tool / Model | What came back | What I changed next, and why |
|---|---|---|---|---|
| 1 | a storefront sign in Queens | Firefly Image 4 | A clean printed sign on a shop that could be anywhere. | Base prompt. Added the subject next. |
| 2 | a hand-lettered storefront sign for a small shop in Jackson Heights, Queens | Firefly Image 4 | The street felt right. The sign still looked printed. | Added a style, because "hand-lettered" on its own was not doing anything. |
| 3 | (2), 1980s sign-painter style, visible brush strokes | Firefly Image 4 | Brush strokes, finally. Colors went loud. | Added the medium. |
| 4 | (3), faded enamel on metal, 35mm film | Firefly Image 4 | Right surface. New building, and the street went wet. | Added light, color and mood to take the weather back. |
| 5 | (4), overcast afternoon, muted greens and rust, quiet | Firefly Image 4 | Closest yet. | Added format and things to exclude. |
| 6 | (5), 3:2 landscape, eye level, no people, no cars | Firefly Image 4 | My final. | Ran the same words in three more models. |
| 7 | (same as 6) | Midjourney v7 / GPT Image / FLUX.1 | Midjourney made it moodier, GPT Image cleaned it up, FLUX kept the grain. | Kept Firefly's. It was the only one that left the sign looking old. |
| 8 | hand-lettered enamel storefront sign, in the style of Ben Shahn | GPT Image | Flat color and heavy lettering, but nothing I could point at and call Shahn. | Rewrote it as a description to see whether the name was doing any work. |
| 9 | hand-lettered enamel storefront sign, flat unmodulated color, heavy irregular brush lettering, social-realist figures | GPT Image | Closer to what I actually wanted. | Liked this one better. It gets me the same place and I can say where it came from. |
| 10 | a small business owner | Firefly Image 4 / GPT Image | 25 per model, 50 total, as two 5 × 5 grids. | Nothing. Not changing it is the whole point of the homework. |
img2img on a photo I took on 37th Avenue. I thought handing it a real storefront would stop the building from changing every time I added something. At low strength it gave me back my own photo with softer edges, and at high strength it invented a different building anyway. There is probably a setting in between that works and I ran out of patience before I found it.
I also lost most of an hour to text. Every model misspelled the shop name at least once, and two of them produced letterforms that read as English from across the room and fall apart up close.
★ Thread this week: representation, and whose defaults ship inside the model.
Of the 50 homework images, 43 were men, and both models put almost everyone in the same age range, somewhere around 40 to 55. The shops agreed even more than the people did. Mostly produce and flowers, almost no salons, no phone repair, and nothing that looks like the South Asian and Latin American businesses that make up most of the strip I walk down to get to class.
Where the models differed was smaller than I expected and more interesting. One model produced all seven women in the set. The other gave me newer, cleaner storefronts across all twenty-five, which reads like somebody tuned it that way. Differences like that are company decisions. The agreement is not. Two companies with different rules on top produced roughly the same shop owner, and the reason has to be further down, in the pile of pictures both of them learned from.
Then I ran a photo of my desk through theyseeyourphotos. It decided I was a graduate student in my late twenties who works in finance, based on a laptop, a coffee cup and a stack of paper. None of it was right, and all of it was stated like a fact. It felt like the grids in reverse: the image models drew one kind of shop owner when I told them nothing, and this one decided who I was from almost nothing.
What I did about it: nothing to the homework prompt, because changing it would have ruined the comparison. What it did change is my in-class prompt. I had been writing "a small shop" and letting the model pick what kind, which means I was letting it decide whose block this is. My final version names the business now.
Leaving the best image out. One generation from step 4 has light that none of the others have, and it is the image I would put first if I were showing off. It came from a prompt I had already moved past, and nothing my final prompt makes looks like it. Showing it as my result would have meant pretending the prompt got me there. The model has no opinion about that, because it never saw the other steps.