Oi Hive,
Was checking in on @starkerz the other day, and we were briefly catching up. Happens he and @theycallmedan are working on a book! Pretty nifty. Long story short, they were waiting for the book cover to be done, so I asked who was going to make it. "You, if you want". Well, why not ;). Thought it would be cool to see if I could create something with Flux AI, because I haven't been using it for a long time, and needed to re-check all the settings. Got to stay awake these days, because a lot of stuff is going on!
Just made my morning coffee, and did a few prompts. This can be such a tedious job as it can break or make the image, literally.
I haven't read anything yet about the book, so I just had to go with the title and let my imagination run wild. Obviously, it's about Hive, but also about governance. Matt mentioned something about "everything you know to be able to argue against centralized blockchains". Interesting, and I can't wait to read their book.
So, let's get going with a quick keyword focus analysis.
Community, blockchain, governance, crypto, future-tech, innovation. I kinda get that these things are quite important to visualize.
I thought it would be cool to show you guys the process of AI generation from your local home, instead of those dumb credit-hungry
FLUX AI
I'm running Flux AI locally on my desktop. It's a bit more complex than the ones you get online. But it gives you a lot more control. Most importantly, the size you're generating images in. I just used a simple workflow to generate quick seeds, and stopped until I liked something to scale it for the book cover. Even though I'm using my Nvidia 4080 INNO3D Geforce RTC X3 SUPER, generating an take still quite some time, especially if you're scaling up the size ;).
The smaller ones that come with a standard 512x512px preview, take about 1.5 seconds, which is ridiculously fast. I kinda messed up, because the aspect ratio impacts the image generation a lot. It's just a shame when you 'finally' generate something cool, you can't rely on it again if you missed a few steps for further optimization or actual use. This is why you need to know exactly what you're doing when working with AI. Prompt engineer jobs incoming in 2026 ;).
I waited about 10 minutes to generate an absolute failure. At least my kitchen looks nice now. Should do some push-ups for the next generation. lol. Who said prompting can't make you get buffed, eh? lol. This reminded me that I first need to do some test generations with some models and options, like a SD Upscaler for example. You definitely dont want to be looking at a mosaic canvas because the pixels are just stretched. SMH. I clicked on the wrong upscaler lmao.
So here's the deal:
5 second generation will get you a: 
Using the wrong Upscaler will get you 15 seconds extra rendertime: 
And after 12 minutes, and using the correct upscaler you'll get this:
Looks like we have everything set-up.
Let's disconnect the upscaler, and pump some generations for fast previews and use something we want. Let's also check out the prompt carefully so we can get cool stuff too.
Prompt #1
Hyper-realistic, nighttime cityscape. Silhouetted figure, outstretched arm, rooftop vantage. Glowing, intricate digital sphere above, pulsing with light, extending into starry sky. Deep blues, cyan neon, and warm orange-red city lights.
|
|
|
Prompt #2
Create a hyper-realistic, cinematic-quality background image set in a futuristic city at night. The skyline should be filled with detailed skyscrapers illuminated by a mix of deep blues, cyan neon, and warm orange-red accents. In the foreground, a close-up of the shoulders of a black colored silhouetted person, standing on a rooftop, reaching and looking toward the sphere in the sky, wearing a interconnected node pulsating light costume, conveying a sense of agency and visionary leadership. Above the city, render a large, high in the sky but centered, blue glowing digital sphere composed of intricate, interconnected nodes and threads of light, symbolizing a decentralized blockchain network. The network should appear vast, alive, and evolving, with light pulses and digital trails extending outward into the sky. Style should be highly detailed, photorealistic, and atmospheric — blending cyber-futurism with emotional depth. Super realistic photography, perfectly composition.
|
|
|
Along the way, there were dozen's of generations that looked cool, but were kinda looking handicapped or just off. There was always something going wrong in the generations, which makes it such a tedious job, as expected. It's like getting a warm snack you absolutely love, but it was cold when you ate it. Uufff.. not eating that kind of shit, I'd toss it away, as I did with the AI generations.
Making the actual book cover
Oh well. so, the generations for the actual book cover take so much more time, so I checked out the low-res images to see which one was most appealing to me. I absolutely love the woman in a cyberpunk styled costume with the huge sphere in the center. I scaled that one up to complete my journey in Photoshop.
I mean, this generation took, FOREVER, but certainly faster than a real artist, (sorry artists). A whopping 15 minutes.
So, in Photoshop I changed the hue of some colors, added a blur-path, and made it more glowing. It looks quite futuristic if you'd ask me. Let's see if I can make another one so the boys have another option to choose from.
|
|
I got your Back
If you need a front, you need a back. Don't worry, I got your back. Let's make the back of the book as well. I wanted to add the summary as well, so instead of making it super busy with another image, I figured having a textured solid background would be much better. Especially if you wanting to add text on top of it.
Hope you guys learned something along the way about Flux!
Good day, well spent. Hope you guys had a nice day too.
Cheers,
Ruben