How good are you at using AI to generate photorealistic images?

Here is the one I came up with:

My prompt was:

Create an amazing space scene with the majority of the scene showing small stars off in the distance and looking towards a nebula that consists of multiple colors. On the right side show a planet that we are overlooking to see the space scene.

So I asked Copilot to give me an Image of the Horse Nebula with Zeus riding on it. It told me "Here you go! A cosmic scene featuring the Horsehead Nebula with Zeus in all his glory. Enjoy the view!" The image it gave me was

Then I said: "I don't see Zeus in that image. Can you add him?" Then AI told me "Looks like I might have gotten carried away with the cosmic vibes. Here's another try with Zeus right where he belongs." Then it gave me this.

On a more humorous note, this is a cartoon I asked AI to make. Text is my own.

In order to reply with a picture, I couldn't add a photo in reply to your comment, but I actually have a couple already.

And this:

Last night while I was in a state where I was halfway between that "still asleep" and "waking up" state, a cartoon strip concept came to me, and I had to attempt it this morning. The girl is named Jean, the boy is Rod, and the Dog is Barry. Hence, Jean, Rod 'n' Barry. (Yes, it's a total play on the famous Gene Roddenberry). And yes, they will be space explorers. Now, I just need to figure out how to get AI to reuse the characters in different scenes.

While on Copilot, I did ask AI to redo my initial image and make it look more like a real Tomahawk steak. This looks much more realistic, and I will add, perfectly cooked.

Some of those that have frequented here for a while may remember my Pittsburgh roots, and also may recall my affinity for the Pittsburgh Steelers. Here's an image I created for tonight's Steelers vs. Browns game. (We like to call them the Brownies.)

A gourmet tomahawk steak served on a black plate with creamy mashed potatoes, topped with herb butter and green pesto sauce, garnished with fresh basil leaves and roasted garlic cloves, photographed under soft natural light for a luxurious dining presentation, realistic and appetizing.

Era of Dinosaurs:

I had another go with AI this morning, this time using X AI (a.k.a. Grok) I supplied an image of myself to use for reference. Unfortunately, at the moment, AI apparently still can't use images of "real" people. However, I think I got it to come very close. The image I gave Grok for reference was this one of me.

I spent about an hour working to tweak it, but this is what AI finally came up with. Thought it was pretty close.

UPDATE:

I had reached my maximum allowed edits the other day (I wasn't happy about the food in front of the chef), so I went back today and gave Grok the following instructions:

Can you edit this to make the chef coat white and remove the food on the table and replace it with a New York strip steak resting on a wood cutting board. The steak should have the appearance of being grilled with crosshatched grill marks and some juices running on to the cutting board. Leave everything else in the picture exactly the same."

It then produced this:

So far, I have liked the results and accuracy of Grok AI vs. Copilot AI

I can generate pretty good photorealistic images using AI, especially when given clear, detailed prompts. I can help create things that look realistic, from portraits and landscapes to objects and scenes, though the more specific and detailed your prompt, the better the result. I can also tweak the output based on your feedback, so you can get closer to what you're imagining.

If you have a specific idea in mind, feel free to share, I have done this guys:

This is the most realistic "photo" I've made to date. I asked Copilot: "Can you create an image of wings being cooked on a grill with a sheet of crumbled foil underneath them". If I hadn't told you this was AI generated you would never know.

To celebrate Friday the 13th, I had Copilot create this one"

Hey Mark,
Love the throwback to the old OZZU photo competitions—those were fun times! AI image generation has definitely come a long way. That Tomahawk Steak example looks mouthwatering and impressively rendered. I’ve been experimenting with Copilot and Midjourney too, and it’s wild how detailed and realistic things can get now, especially with lighting and texture.
Would be awesome to see a community challenge again—maybe a themed photorealism contest using AI tools? Could bring back some of that creative energy from the early days.
Looking forward to seeing what others share!

It's been a little bit since I added to this. But I tried an experiment and am curious what you think. I also learned that not all AI is created equal. I gave ChatGPT, Grok AI, and CoPilot the same exact instructions. If these were on a menu and you had to choose, which one would you be inclined to eat? I'm not going to tell you which one generated which image (except one of them always generates a watermark lol). These were the instructions:

Can you create a photo of Smash Burgers with Sharp Cheddar, Caramelized Onions, and Sweet Potato Fries simulating the feel of eating in a burger joint?

These were the results:

I stepped out of the box this morning and actually installed Stable Diffusion (an AI image and video generation program -- open source) on my PC. Realize that I'm a complete novice at using it and have only just started reading the very beginner user guide. I know this image has flaws, and I haven't gotten to the part where I learn how to correct them. But for my first generation of an image knowing practically nothing I don't think I did too bad. This is not a program for the faint of heart. Even though it's free to run on a PC, there are several steps that you must know what you're doing to install it. (including but not limited to installing Python on your PC) Complete installation instructions are here.

My second try (learned how to increase the image size). I've stood on beaches like this at both sunrise and sunset. This one isn't too far off from the real deal.

After a morning of trial and error I was able to come up with this. (It did require a small bit of editing in gimp - her lefthand fingers were showing under the purse). The keywords I used to create the original were:

a young long blond hair woman on a city sidewalk, purse, photography, photo from about 10 feet behind, entire scene full color, gentle breeze, realistic color all buildings, street view, other people walking, back view of woman, natural arms and hands

I learned how to find and install and use Stable Diffusion checkpoints and it was a gamechanger. This was made using cyberillustrious checkpoint model with keywords:

a tanned young blonde about five feet tall and 100 pounds in a bikini wearing sunglasses on a beach walking in the ocean, waves covering her feet, blue water, blue sky, full midday sun, photorealistic

Note: it took about an hour of experimenting and altering keywords to get to this image. I'm not happy with the right foot, but I haven't learned to fix that error yet and was hoping the waves would actually cover her feet but was way happy with the rest of the photo. Brian, this time I got the sand correct.