Over the past few days, I've been to space, climbed a mountain, cuddled a lion and flew through the air without a plane. Obviously not for real, but using new generation artificial intelligence image tools I was able to create a digital version of myself with a much more active life than me.
One of the most important recent developments in generative AI is LoRA, or Low-Rank Adaptation — a technique for fine-tuning a large model with specific data without having to change the whole model. It means you could train a model on your own image relatively quickly.
I turned to Freepik to do this for two reasons. First, Freepik's Consistent Character system makes the process relatively painless. Second, it has the stunningly impressive Mystic 2.5 AI image model from Magnific that is the most photorealistic I’ve ever tried — and I want the results to look real.
I took 25 photos of my face from various angles, some with and some without glasses and then uploaded them to the Freepik Character Consistency trainer. I gave it a name (ryanm) and within 15 minutes all I had to do was tag @ryanm in any image and I’d appear. There is a reason Freepik is on my best AI image generators list, and this just adds to its value.
Going on an AI adventure
After I had access to a digital clone of my face the next challenge was to work out what I wanted to do with this and how to properly test it. My first instinct was to put myself in space and then add digital me into a range of films and TV shows, but that raised copyright issues.
Eventually, I settled on sending AI me on a series of adventures, starting with that trip to space and then up a mountain — things I’d likely never get to do myself in real life but looks like they could be exciting. I added a few "not possible" selections in as well just for fun.
I took each of the images generated in Freepik with Mystic 2.5 and animated them using Runway Gen-3 Turbo to bring them to life. Some of them were so surreal they didn't fit this article, so I put them into the YouTube video above with a Suno soundtrack.
1. Going into space
It took several attempts to get this one right, including re-training the LoRA on my own face using new images without my office shelves and coffee machine in the background as they kept creeping into the pictures. One space image had Settlers of Catan on the desk.
I got the perfect shot of me wearing a space suit sat in a chair on a space plane with the Earth visible through the window next to me. This is a view I’d love to see for real one day but seeing myself in that situation was still a fun experience.
2. Climbing a mountain
I am not fit or particularly healthy. I have bad knees and I don’t like heights, so the chances of me ever actually climbing a mountain are very slim. However, I’d love to experience the view. The photorealism of Mystic 2.5 helped capture this perfectly.
For some reason the image generator knew me better than I thought. It had no images of me beyond my head and shoulders but still understood I’d likely wear shorts even climbing a mountain and surrounded by snow.
Initially, it gave me a Spanish flag (Freepik is a Spanish company) but I was able to easily change it to a Union Flag with a quick text prompt.
3. Cuddling a lion
The lion has been my favorite animal since childhood. Not only am I a Leo, but lions have always seemed majestic, so this was a chance for my virtual alter-ego to get up close and personal with the king of the jungle (even though a lion doesn’t actually live in a jungle).
This is another one that took several attempts to get right. The problem was one of scale as every version seemed to make the lion tiny or me massive. The end result still isn’t perfect, but I loved the way the final image came out so decided to forgive the scale issue.
4. A trip to see the Northern Lights
Recently the Northern Lights have been visible fairly far south, even as far as where I live, but they’re not particularly spectacular unless you have a camera. One of my goals is to travel to Iceland one day to watch the Aurora Borealis streak across the sky.
For some reason every one of these images made me appear to be looking anywhere but at the lights. In the one I finally settled on, I seem to be photographing a frozen puddle. Maybe I’m getting a creative reflection shot of the lights.
5. Photographing a polar bear
I am a huge fan of the cold. I do not like warm weather, and I’ll happily wear shorts even if it's 20-below Fahrenheit. I’m not sure I like the cold enough to visit the land of the polar bear, although I could double it up by taking a trip to Norway and see the lights at the same time.
For some reason every polar bear image I generated had the bear looking like it was struggling for food and in its "winter dormancy" phase where they conserve energy. I’m also not entirely sure why I’m so close to something that looks starving and could easily eat me whole.
6. Going on an archeological expedition
This one is a struggle, as I hate the heat, but the image generated me with a nice hat so that helps. Here, the idea is that I’d help in a dig to uncover the secrets of humanity's ancient past, although the image looks more Nerd-iana Jones than Indiana Jones.
I wasn’t going to include this one originally as it didn’t feel like much of an adventure but I loved the historical aesthetic. It feels like a photo taken with a modern camera but during a dig in the 1920s rather than modern times. I’ve snuck off to read a newly uncovered scroll.
7. Floating in the sky without a plane
Who hasn’t had a dream of flying at some point in there life? I’ve always wondered what it would be like to "safely" fall through the sky, but not enough to do a parachute jump or go on one of those "space plane" simulators. I’m happy to let AI me try this one out.
For this, I had Freepik create a version of me simply in the air with blue sky and clouds surrounding me. That was it. It felt a bit like I was in the opening titles of the Simpsons. I made about 10 images of this, and only one looked like me — so this was the least successful attempt.
Final thoughts
The one thing that struck me when creating content for this article was just how realistic the images looked compared to AI images just a few months ago.
There were a few stray fingers and the odd extra version of myself in an image when I was making the pictures, but even using a real camera, you don't get a perfect photo from every single shot.
AI is evolving into a brilliant storytelling medium. My 4-year-old son came into my office while I was sorting through the images of me doing weird and wonderful things and he asked "when did you go into space?" This let me make up a fun story for him — obviously explaining it was just an image and not real.
More from Tom's Guide
- Best food delivery services: Grubhub vs Uber Eats vs Doordash
- The best cast iron skillets 2024: Tested and rated
- How to make images with AI using Leonardo
