53 - Ryslan

Ryslan is an 18-year-old designer from Russia. He started designing in 2023, working from his phone before saving up for his first PC. About six months ago he left the studio he was working at and started focusing on his own projects.

Since then he's been sharing his work and thoughts on Twitter, taking on new projects, and planning to keep doing this for a long time.

Twitter (X) → x.com/Roward_design

I've already made a few thumbnails for Borja, so I already knew what kind of style he was looking for. After he sent me a few really good photos, which made a huge difference to the quality of the final result, I got started. The idea was to make it look like he was struggling to hold a heavy barbell, with blood shown on his chest.

For style inspiration, I used these two thumbnails — I had actually made one of them for him before.

First, I just made a quick sketch in Nano Banana by giving it a simple prompt: "draw a sketch of this photo, without too many details, using a black marker on white." I picked one of the generated sketches and added the man's body to it, so I didn't have to generate everything from scratch and could keep it pretty close to the original photo.

After that, I described the original photo to ChatGPT and started generating. This was probably the longest part of the whole process — getting the prompt right and making it generate what I needed took quite a few attempts. But after around 40-50 generations, I got this version. I liked it because the composition was different: it looked like the guy was closer to the camera, while the barbell was closer to him, and overall the composition felt pretty much perfect to me.

Now comes my favorite part of the process — when you realize the thumbnail is actually turning out really interesting, and you can finally start polishing all the little details.

The first thing I wanted to change was the background. I wanted to make it cleaner, so I refined it using Nano Banana. The prompt was pretty simple: remove some of the gym equipment from the background.

In the end, I just combined the backgrounds from two different generations. For example, I really liked the dumbbells on the left from one version, so I kept those and mixed them with some elements from the other one.

After that, I just manually touched up some small details, like making the background brighter and so on. The only thing I'd really point out is the face replacement. I simply copied a close-up photo of the face and enhanced it using Banana 2 (4K).

The prompt was: "Transform this photo into a detailed 4K photo, with a slight skin glow and a light layer of sweat on the skin."

And that's basically how I got the final result I sent to the client.

They asked me to remove the blood and make it look like he's in pain, like he has a shoulder injury, and this improvement made it look way better.

Communication: Discord
Editing: Photoshop, Magnific, Ideogram

If you found this edition of Unlayered helpful, please consider sharing it with someone who might benefit from this workflow too! 💙