Last night I gave a webinar on Basic AI for Genealogists for the Toronto Branch of the Ontario Genealogical Society. The highlight of the entire session was the different types of images that we can create using AI. It was such a hit that I thought you would like to read about it as well.

A quick refresher on the basics:

You work with the major AI tools (ChatGPT, Claude, Gemini, etc.) by typing in directions, called “prompts”. In order to be most effective, a prompt needs four parts:

•Role – What expertise or knowledge do you assume the AI has already?

•Task – What do you want the AI to do?

•Format – What format do you want the results in?

•Limits – How many different results do you want?

Additionally, is there any background or important information that you want the AI to know before starting?

Here’s an example of a basic prompt.

Two of the most common ways to create an image using AI is to either describe the image that you would want to see using a prompt only, or supply the AI program with an original image file, and prompt the program to modify it.

Some Examples

Let’s do the text only version first:

Prompt: You are a professional genealogist with a specialty in Italian history and genealogy. Create a cartoon style image of a southern Italian woman from the 1850’s in traditional dress.

Result:

Prompt: You are a professional genealogist with a specialty in Italian history and genealogy. Draw a southern Italian mother and children from the 1850’s eating a meal in a hyper realistic style.

Result:

Keep in mind that AI programs create the image from scratch every time you use the prompt, so subsequent versions may be different than what you see above.

Try to describe the image as completely as possible. You can specify artistic style, colors used, background and setting, time of day, etc. The more information you specify, the closer the image will be to your desired results.

If you want to modify a photo or image you already have, you can attach it to your prompt. Look for a plus sign, paperclip, or other icon that indicates attaching a file. For example, below is the photo of my great-grandmother that accompanied her naturalization records. It’s one of the only photos I have of her.

Prompt: You are a professional photo restorer.  Please colorize this picture.

Result: I love this picture! I didn’t notice her jewelry in the original picture, but it stands out here.

Now let’s do something fun!

Prompt: You are a children’s book artist. Please create a picture of this woman sitting in a rocking chair in a living room, knitting.  The room should be pictured as a tenement building in Brooklyn in the 1930s.  The picture should be in comic book style, suitable to give to her five year old great grandchild.

Result: How cute is that?!

One of the great things about AI is that you can start over if the results aren’t what you wanted. For example, I asked for a coloring book style picture of my immigrant ancestors arriving by boat in New York Harbor in 1840. The image included the Statue of Liberty in the background…however the Statue of Liberty wasn’t erected in the harbor until 1886. If you’re using images to accompany reports, in books, or for any official work, make sure you check the details! AI programs will sometimes make things up. (this is called a “hallucination”)

Additionally, if you are using a free version of any of the major AI programs, you will probably be stopped from creating more than three or four pictures in a 24 hour period. This doesn’t usually apply on a paid version of the program.

I would love to see any images you create! Share them on the Family Tree Tech Facebook page!