Photographic Quality Image Generation Settings and "Train Your AI"

Hello all,

I absolutely love this “AI Generation” for my blog posts and image creation! I am hoping someone can can give me some tips on what I am trying to do. There are so many options available when it comes to image generation. I am trying to get as close as I can to “life like” images that look like photographs of people. My testing during the last couple of days is making my usage higher than I would like it to be.
I mean I am generating images and I do not mind paying for that. I am just not saving many of them right now until I get everything fine tuned. Currently my settings are at “None” as much as I can. I just started testing out the color options. I have uploaded a screen shot.

The other part of my question is a recommendation of the format to type of what I am looking for with results. In my current description I am trying to get the IA to understand that I want a photograph of two men wearing different styles and different colors of outfits. I have not been able to accomplish this yet. I know I am asking a lot but if this can be accomplished with the proper prompt info that would be great. I will add my input information below.

The last part is knowing if there is a way to train the “AI” to learn what I like by knowing which images I save and the ones I do not keep?

My input information:

“Generate images that look like photographs of 2 Nigerian men at a wedding smiling. Their wardrobes are different.
[1 man is wearing a blue colored “3 piece” Agbada outfit.]
[The other man is wearing a traditional orange colored “aso oke” matching shirt and pants]”

The application you are using is not OpenAI software. You can tell this by the Stable Diffusion tab, which is a different image generator.

The selection of different options on the right side likely just injects text into the request for an image generation.

It is probable that it uses the OpenAI API image generator, which might be rated at “tier 3” as far as image quality, behind DALL-E 3 (now slowly being made available in ChatGPT Plus), and then followed by DALL-E 2 at (purchase credits for image generation).

An option you might pursue, with a Microsoft account, is the Bing Image Creator. It has also been upgraded to DALL-E 3, and does a better job of making people and faces than any prior generation.

Knowing the prompt style that produces results, I give you:

Two smiling Yoruba Nigerian men at a festive traditional wedding, in an impromptu posed photograph. The first man is wearing a blue three-piece Agbada traditional garment. The second man is wearing a traditional orangea aso oke, with matching and complementary shirt and pants.

1 Like

Oh my Lord! Thank you very much for your detailed explanation and advice. I am just getting started with the “AI” applications. You are right, the software I am using is not OpenAI. I needed to set up and fund the OpenAI account to use the plugin though. I just contacted those folks as well.

I am definitely going to check out the Bing Image Creator.

Did you generate that image there?

Yes, that is Bing with DALL-E 3 that created the image.

I took your advice and it works!

He needs a hat too.

I have had great luck with making images on Bing. Thanks again!