Faa Automated Image Keywords / Descriptions

It helps me tremendously because sometimes my mind just goes blank when thinking of key words or descriptions. I definitely edit them but it gives a good starting place.

I can't say if it is improved. Had not used it much until now.

I do see improvement with descriptions when there is recognizable information in the image. People, objects, animals, places, text, etc. gets decent descriptions. Other then that the descriptions end up very generic, not good for seo and needs a lot of twerking. I see many new uploads using only the generated description. I am sure that won't matter for FAA's internal search, but may not help at all with external search and indexing.

The generated keywords are still not great with little to no improvement as I see it. Still, better then nothing and it is all a good start.

I did never get the right place even when it was very famous. So no real improvement.
It is a nice starting point but not the final result You and the customers want to read. Not overworking it is a big mistake in my opinion.

Its one of those things I wish would work using a button, because I always fill in content, so I never see the generated one. A click and oh those are keywords I never thought of etc... Would be a nice add on.


---Mike Savad

I think it has gotten better. I like to use it as a starter, then modify the description to make it more "human".

One thing that I use if I am trying to identify a building or monument, etc., is Google lens.
If you open the Google app, then select the little camera icon next to the search box, take a picture of your image/item.
You will have an opportunity to add key search words, then Google scans for similar objects.
This has saved me a lot of time identifying distant items.

It is a wonderful "starting place" but needs to have my influence added to better describe/identify my photographs.

I uploaded three abstract pieces, it can pick up the colours, the descriptions were bland and generic, all described as vibrant, energetic. Still rewriting them. AI descriptions definitely do better on more figurative pieces, a lot of reference material I suppose.

James,
vibrant seems to be the absolutely favourite word of this AI. I get it for may be 60 or 70% of my images.

I'm sticking with gpt to gather what is hopefully facts and to make it all flow better. Though its often wordy, I think it reads better than what the site can do... One of those cases where my AI is better than your AI. I still try to write parts myself and then blend it in later.


---Mike Savad

Mike strikes a cord there with "my AI is better than your AI". I use 4 different AI sources, all for different purposes, with different types of images, what works best for each.

The AI here I will use by leaving some of it, but very little. I use Gemini most of the time for creating descriptions, titles and keywords. ChatGPT is second and used for creating social media posts as well. I like that Gemini and ChatGPT retain my chat history on the side panel. I can always go back to specific ones and continue without prompting again. I do use Microsoft Copilot. Its pretty good for everything but they, disappointingly, do not display the chat history. If I close it and go back, I have to prompt all over again.

Its interesting how different the flavors are. Like gemini is useless when it comes to anything I ask it to do but I never used it for story writing, they added it to my phone and its just not helpful other than to hear an early realistic voice. I don't like theirs. The microsoft one is impossible to use perplexity isn't that bad, google tends to give me a lot of wrong info without links to check.

Still even with gpt I have to go back and say , remove this, and that, and less of this more of that, make it flow better, shorter, longer and its a pain but easier than doing it myself.


---Mike Savad