When AI Starts Scoring Photos Like a Real Teacher
1 min read
When AI Starts Scoring Photos Like a Real Teacher
We’ve all seen AI critiques that feel half-baked. Vague descriptions. Generic advice. No visuals, no accountability. That ends now.
This isn’t a gimmick. It’s a real tool — built by a real photographer — that scores images like a seasoned photography teacher would. Goker-ish GPT doesn’t just comment on photos. It draws on them. It explains composition, balance, flow, light direction — and it does it with sharp, human-grade annotations.
The Mission Behind the Project
This isn’t just a technical test or a fun experiment. It’s a serious step in a longer creative path. A few days ago, I decided to launch a dedicated photography art project, and the features of Goker-ish GPT will help refine and support that vision. For context, I also hold an associate degree in photography — but this project goes beyond credentials.
Anyone serious about improving in photography knows how rare it is to get solid, actionable feedback. That’s where Goker-ish GPT comes in. It’s the AI photography teacher I wish existed years ago — one that critiques like a pro, draws like a professor, and doesn’t waste your time.
What Makes This Different
No vague advice. No fake polish. Just:
- 📸 Rule-of-thirds and composition guides
- ✍️ Hand-drawn style annotations — jittered lines, elliptical arcs, and all
- 🔍 Clarity on subject isolation, negative space, flow, light direction
- ✅ A clean score with category breakdowns, just like exam submissions
- 💬 Teacher-style comments, placed on the photo like real feedback
Watch the Demo (Unlisted YouTube)
In this video, Goker-ish GPT receives two real images and the prompt: score this
. There’s no warm-up. No second chances. It delivers spot-on critiques — visually, precisely, and with confidence.
This is how AI should analyze images.
Watch the full scoring breakdown: 👉 Watch the demo on YouTube
Try It Yourself
Want to see how it handles your photos? Go to goker.me and start typing. The GPT isn’t listed — just begin with a prompt on the home screen.
If you want a photography teacher that draws on your photo and tells you exactly what’s working (and what isn’t), it’s waiting.
Building It Was No Joyride
Let’s be real: setting this up wasn’t smooth. At first, it was just a casual chat. I asked the GPT to annotate and score a photo. It took nearly 30 minutes to get a halfway decent result.
To make it actually useful, I had to rewrite the context and shape it into what is now the "Goker-ish" instruction layer. I won’t get into every detail, but the entire process took over four hours to fine-tune. Custom Python libraries were written from scratch to mimic how a photography teacher would annotate an image — from sketchy jittered lines to elliptical arcs.
Color-coded feedback is coming next. The foundations are built. The tool is ready.