Problem · All agencies

“ChatGPT gives generic titles without the place, the event or the names.” Why it happens and how to fix it

You ask a chatbot for a title and get “Beautiful lake with a church on an island”. It is correct, and no buyer searching for the real place will find it.

Short answer

This is not a ChatGPT fault. Any AI model only sees the pixels, so it cannot know the place, the event or the names. In PhotoTager you write one context note for the project and it is added to the prompt for every file.

Why it happens

01 The photo does not say where it was taken A model can see a lake with a church on an island. It cannot see that it is Lake Bled unless you tell it.
02 A chat window starts from nothing each time You have to paste the same facts again for every photo or every new chat, so most files end up described from the pixels alone.
03 A chatbot does not know each agency’s rules Adobe Stock, Shutterstock, iStock & Getty Images and Alamy have different limits and formats. A chat answer is one block of text, so you copy, trim and reformat it by hand.

How PhotoTager fixes it

  1. 1
    Write one context note for the projectAdd the place, the event, the names or the species, up to 500 characters. It goes into the prompt for every file in the project, so you type it once.
  2. 2
    Titles, descriptions and keywords follow the agency rulesYou get a title, a separate longer Adobe Stock title, a description, and up to 49 keywords for Adobe Stock. The limits for each agency are built in.
  3. 3
    iStock keywords are matched to its vocabularyAfter every title run, a free match checks your keywords against the iStock & Getty Images vocabulary.
  4. 4
    Export a CSV for each agencyDownload the file for Adobe Stock, Shutterstock, iStock & Getty Images or Alamy, matched to your file names.
  5. 5
    Test it on one photo firstThe Prompt Playground gives you a few free runs a day (5 without an account, 10 with one). Run one photo with and without a context note and compare the titles.

A chat window or PhotoTager

WhatA chat windowPhotoTager
Place, event and namesYou paste them again for each photo or chatOne context note per project, added to every file
Agency limitsYou check and trim by handBuilt in for Adobe Stock, Shutterstock, iStock & Getty Images and Alamy
Adobe Stock titleOne answer you adapt yourselfA separate Adobe Stock title next to the main title
iStock & Getty Images keywordsFree-form phrasesMatched to the iStock vocabulary, free
OutputText you copy and pasteA CSV for each agency

Questions

Is ChatGPT bad at stock photo titles?

No. Any AI model, from any company, only sees the pixels you give it. Without the place, the event or the names, the title will be generic. PhotoTager also uses OpenAI models; the difference is that your context note goes in with every file.

What should I put in the context note?

The facts a buyer would search for and the photo does not show: the place, the event, the names of people or species, the date or season. It can be up to 500 characters and applies to the whole project.

Can I test this before paying?

Yes. The Prompt Playground has a few free runs a day (5 without an account, 10 with one). Run one photo with a context note and without one to compare the titles.

How many keywords does Adobe Stock allow?

Up to 49. PhotoTager’s Adobe Stock export keeps to that, and titles are best kept to about 70 characters.

Does one context note fit a mixed shoot?

It works best for photos that share the same place or event. Put different shoots in separate projects, each with its own note.