I tried to generate a banner image yesterday on Midjourney, the image was having six fingers with three legs. 😂 But after I tweaked the prompts a little for like two hours, I finally got a satisfactory image. I didn't believe it could come out that good. To be honest, spending hours to generate just a single image is overwhelming.

On Oct 4, 2026, Flower said:

I tried to generate a banner image yesterday on Midjourney, the image was having six fingers with three legs. 😂 But after I tweaked the prompts a little for like two hours, I finally got a satisfactory image. I didn't believe it could come out that good. To be honest, spending hours to generate just a single image is overwhelming.

Was that a photo realistic image then?

I’ve had success on ChatGPT with getting images and having it make edits. I have the free version and sometimes I have to wait a few hours before I can get an edit made.

I really like my Drama Llama logo for my forum:

On Spark Promotion I've got a Halloween mascot that AI designed for me. It’s Gizmo as a Spider. I call it Gizder.

I did some back and forth with my resume and cover letter when I was applying to jobs. I now have a job so that seemed to work.

I've seen VS Code Copilot provide some useful suggestions, basically being ‘better’ Intellisense (but you still have to thoroughly vet the output).

I've been meaning to try out some agentic stuff but haven't done it yet.

Those six-fingered Midjourney results are hilarious until you are on a deadline and just need a clean banner.
I don't generate much visual art myself, since most of my experiments are software builds.

The AI project I'm happiest with is a healthcare voice prototype I built called Clara Health, which pairs a conversational voice companion with doctor discovery and appointment booking.
Getting the voice responses to feel calm and natural without awkward latency took hours of prompt and pipeline tweaking, pretty much like your two-hour banner session.