I want to start dropping older things I built in here now and then, so they do not just rot in a drawer. Someone might find them interesting, and I get a bit of backtracking for myself. So let me start with a project that genuinely entertained me at the time.
It was part of a quote for a client we were delivering an LED wall to. As part of the visualization I built him an interactive 3D demo with the real dimensions of the wall, and crucially with the real dimensions of the so-called pitch, the spacing between pixels. So he had something to compare and could see for himself what each option actually meant.
And then the thing that always happens to me happened: I was enjoying it, so I got stuck on it for a few days. A plain visualization grew a pile of gimmicks:
- a fully controllable reference figure, so you can read the scale,
- a VR mode,
- a realtime Telegram integration,
- and a million other completely useless things.
In the end it was mostly a benchmark of what the models of the time could do. I threw it together around October 2025, back when models were not as loaded as they are now and you genuinely had to herd them through work like this. Watch every step, walk back every wrong turn.
And that is the most interesting part in hindsight. With today's models it is quite possible I would one-shot a project like this given a good brief, and it would probably look a lot better too. In a few months, what took days became a question of one good prompt. That is exactly why these things are worth keeping: they are timestamps for how fast all of this is moving.
It also makes sense without the toys. Pixel pitch is hard to explain in words, but when you can spin the wall, step closer, and watch where the image falls apart, the decision suddenly gets easy. Showing beats describing.
The demo is still live: visualiser-led.xctech.cz. There is also a video on Facebook.
Over to you: do you have one-off projects that would be a shame to let fade away? Bring them out.