I had a little fun with SwarmUI again recently, often generating images to post into a chat at work or to print them on small table cloths.
So, here they are:
A bunch of images
Category: AI Images
Butterfly
Stable Diffusion is like a box of chocolates. You never know what you’re gonna get.
Aidin’ Aiden – SuddenFx
I was trying to create a new wallpaper for Aiden and my Laptop that I mainly use to connect to it. I created an AI generated image (of course) of the logo/icon thing of Aiden:
After some attempts with Qwen Image Edit, I tried Flux2.dev and got a pretty good image:
It was cool, but lacked some effects to make it pop. I wanted scanlines, chromatic abberation and so on. You know, all the cheap effects seen in indie games. 😉
I thought about launching Affinity Photo or something and look for effect filters that I could download. But then I thought: “Let’s do a quick test if I can create the effects using javascript and css or something.
Shortly after, Claude Code connected to Qwen 3.6 produced SuddenFx.
It is named after a Visual Basic tool that I did like… over 20 years ago… o_O
I think the “real” came after I sped up the effects using native C++ DLLs for the rendering and Visual Basic only for the UI. It was still… very slow. ^-^
But it had a very similar idea and so the new app SuddenFx using Javascript, vibe coding and luck came to be.
SuddenFx ’26
I wanted it to be built, so filters could be added easily, each with a definition of their parameters and a global strength parameter for all of them. They should be reorderable and there should be a quick toggle for each of them and a solo mode to only see the effects of a single filter temporatily.
There is also a compare mode like the ones in similar apps. When downloading the result, the original generation metadata is added again, so it can still be dropped into SwarmUI. There is also a drag source image to download per drag & drop into other apps.
The development of this went really smoothly and was one of the best vibe coding experiences so far. And pretty quickly, I got my wallpaper:
After creating my wallpaper, I let the agent come up with more filters to be used especially for AI generated images and it came up with “Clarity” and some others that really improve the often flat look of AI generated imagery.
Unesroga – The next room (Spoiler alert)
After finishing the demo, I can now work on the next room. At the end of the demo, the player just opened the door to it. When clicking on it, there’s just a “To be continued…”
And that is done at the moment. I’m trying to keep this as much spoiler free as possible, but if you want to discover everything on your own, then:
Do not read any further!
Unesroga – Demo 2 finished
I’ve finished the work on the second demo of Unesroga.
The last things to do was to generate some texts for the completely irrelevant books in the shelf of the starting room. Still, I wanted them to be in there and there will be more in the future.
I had to fix more stuff in the layouting code of the books after adding new ways to style the text elements with a little markup, because changing the font or size of the text made the previous size calculations invalid.
There was an incredible struggle to maintain the correct number of linefeeds because of the way I first split the text into parts (including styling, alignment and some other metadata) during the calculation and split to pages step and at the same time reconstructed the text, so it could be drawn later on.
In the end, I got rid of the double parsing stuff and used the calculated parts directly, which produced some other problems, but eventually it got all resolved and now it is stable and works in all cases when there is plain text, styled elements, page styles or conditions or custom pages with images etc.
The demo contains roughly 75% of the game. There are two more rooms planned and I have a basic idea of what will happen in one of them.
But before I start the work on that, I will wait for the initial reaction to the new demo and see if I have to do some quick fixes to it, so I can release it to a larger group.
I am curious to find out whether the puzzles will be solved quicker (or at all) than the one in the first small demo. ^-^
Reunion
You were a nobody, an urchin living in back alleys, struggling each day to get enough money to eat. So, when this stranger offered you 20 bucks just for a few hours of work on the graveyard, you immediately said yes.
You did not understand what you were doing, placing weird stuff on the ground, drawing circles, lighting candles. But then he sat down next to you and explained.
“Tonight”, he said with a mad grin on his face, “I will return my wife to life. We will finally be reunited again.”
When you saw the movement behind him, you ran as fast as you could. You did not turn around when the screaming began… or ended.
Only on the next day, in broad daylight, you returned to pick up his money and pocket watch. You also picked up an old tattered book. You could not read, but recognized some of the symbols you had drawn onto the ground.
Maybe someone else was interested in this? For a few bucks you would even help them set it up…