i started using visualmind.ai to add meta data - it used to take me an h our to process 50-100 images.  with much more detail & only a few mistakes - it even distinguishes among various mosques in Istanbul

now i can process 100-200+ in about 10' - 1' setup. the rest to copy csv to captions & workflow spreadsheets & unzip to an upload folder

anecdotally i seem to be getting more D L of recent uploads (w better tags, desc)

Quote from: cascoly on October 28, 2024, 13:36

i started using visualmind.ai

Are you using the paid service or is this free? My attempts with chatGTP have not been too accurate

ChatGPT requires you to enter a descriptions - VM takes the image & generates title, description and tags. for simitars they actually write different descriptions.  they're able to recognize many features - eg rhine river based on a caste picture.  Even translates from Greek name on floor mosaics.  you can also prepend/append text or tags (eg country or city where it's not obvious, or byline for SS editorial)

error rate about 1-2%

there's a small number of free submits you can try, but the credits only cost about 2-3 cents/image. i just bought 2000 credits for euro 30