A few years back, I grew my own Steuben grapes at home and decided to try my hand at making wine from scratch. Steuben is a deep blue-purple grape, so I was expecting something rich, dark, and red once it was ready. I bottled it up, tucked it away, and let it age for three years, figuring time would only make it better.
Now one bottle I opened up early and let in too much oxygen (for early tasting).
I was right that time would change it. I was wrong about what it would change it into.
The Reveal
Three years later, I finally opened it up for a taste test. First thing I noticed: the color. This was supposed to be a deep red wine, and instead I was looking at something the color of Scotch whiskey. Then I gave it a smell — and yep, that tracked too. It smelled like Scotch whiskey.
Then I actually tasted it, and that’s when things got weird. It wasn’t whiskey-flavored at all. It was wine that was clearly, unmistakably starting to taste like vinegar.
Color of whiskey. Smell of whiskey. Taste of vinegar. Not exactly the tasting notes I was going for.
So… Did I Accidentally Make Vinegar?
Turns out, yes — probably. And the strange part is that all three of those weird signs are actually the same problem wearing three different disguises.
Somewhere over those three years, my wine picked up more oxygen exposure than it should have — maybe the seal wasn’t perfectly airtight, maybe there was too much air space in the container. Once oxygen gets in and stays in, two things start happening at the same time:
Oxidation turns the color brown/amber instead of staying red, and produces compounds that smell a lot like sherry, brandy, or — apparently — whiskey.
Acetic acid bacteria (they’re just floating around in the air) get to work converting the alcohol into acetic acid. Which is, chemically speaking, just… vinegar.
So it wasn’t three separate flaws. It was one slow, three-year process of my wine quietly turning itself into vinegar, and I just happened to catch it partway through.
The Silver Lining
Here’s the fun part: this is basically how wine vinegar has been made on purpose for thousands of years. Wine, air, and time is the traditional recipe. I just did it completely by accident.
So instead of pouring three years of effort down the drain, I’m calling this a win. Homemade Steuben wine vinegar it is — salad dressing’s gain, wine glass’s loss. Next batch, I’m sealing things up a lot better.
Cheers!
Posted inWine|Comments Off on This Tastes Like Vinegar, Smells Like Scotch, and Started Out as Wine
I’ve always had a vague idea of what Tor was — something to do with anonymity, something associated (fairly or not) with the “dark web,” a tool journalists and privacy researchers namedrop. What I didn’t actually understand was how it works under the hood. So I spent some time digging in, and it turned out to be one of those rare pieces of internet infrastructure that’s both genuinely clever and much simpler than its reputation suggests.
Here’s what I learned — and, since I ended up building it directly into one of my own apps, how I put it to use.
What Tor Actually Is
Tor stands for “The Onion Router.” It’s a free, open-source network run by The Tor Project, a nonprofit that’s been maintaining it for over two decades. The core idea: instead of your device connecting straight to a website, your connection gets bounced through a chain of volunteer-run computers (called relays) scattered around the world before it ever reaches its destination.
That’s the whole premise. The genius is in exactly how that bouncing happens.
The Three Hops: Entry, Middle, Exit
Every time you connect to something over Tor, your traffic doesn’t take one hop — it takes three, through what’s called a circuit:
Entry (guard) relay — the first hop. This one knows your real IP address — who you are — but has no idea what site you’re trying to reach.
Middle relay — the second hop. This one knows literally nothing except “traffic came from relay A and needs to go to relay C.” It can’t see your identity or your destination.
Exit relay — the last hop. This one talks to the actual website on your behalf, so it knows the destination — but it has no idea who originally sent the request.
That’s the actual privacy property, and it’s worth sitting with: no single relay in the chain ever knows both who you are and what you’re doing. Not because they’re trustworthy — because the system is designed so it’s structurally impossible for any one of them to have both pieces at once.
The Onion Part: Layered Encryption
Ever wonder why it’s called “The Onion Router” and not something forgettable like “The Layered Router”? Because whoever named it understood marketing. Peel back one layer of an onion and there’s another one waiting underneath — same deal here, except instead of making you cry by the time you reach the middle, it’s just making sure nobody in the chain ever sees the whole picture at once.
Before your traffic ever leaves your device, it gets wrapped in three layers of encryption — one for each relay in the circuit.
Each relay can only peel off its own layer:
The entry relay peels off the outermost layer. That layer tells it “send this to the middle relay” — and nothing else. What’s left is still encrypted.
The middle relay peels off the next layer. It learns “send this to the exit relay” — still nothing about content or origin.
The exit relay peels off the final layer, revealing the actual request, and sends it out to the real website.
Each relay only ever sees enough to pass the parcel to the next hop — never the whole picture. The website on the other end sees the request coming from the exit relay’s IP address, not yours.
How a Circuit Gets Built
None of this is manual — when Tor starts up, it automatically:
Downloads a directory of currently available relays (thousands of them, run by volunteers — individuals, universities, privacy organizations — all over the world)
Picks three of them for a circuit: a guard, a middle, and an exit
Negotiates that layered encryption with each one, hop by hop, before any real traffic flows
A circuit typically stays alive for about ten minutes before Tor rotates to a fresh one — so even if you’re browsing for a while, you’re not stuck on the same three relays forever.
Why It’s Slower (On Purpose)
This is the part people run into first and complain about: Tor is noticeably slower than a normal connection. That’s not a bug — it’s the direct cost of the three-hop design. Your data is taking a longer, more roundabout path, encrypted and decrypted at every stop, often through relays with limited volunteer-funded bandwidth. Speed and this level of anonymity are fundamentally in tension, and Tor deliberately picks anonymity.
What People Actually Use It For
The “dark web” association is real but wildly overstated relative to actual usage. In practice, the everyday reasons people reach for Tor are a lot more mundane:
Journalists and their sources communicating without exposing who’s talking to whom
Researchers and ordinary people who’d rather their browsing not be tied directly back to their home IP address
People in countries with heavy internet censorship or surveillance, getting around blocks
Anyone traveling who wants to reach a service the way they normally would from home, without a random hotel or airport network coloring how they’re seen
Where FFmpeg Commander Comes In
I build FFmpeg Commander, an app for downloading, converting, and editing video on your own machine. One recurring headache users ran into: trying to download a publicly available video — say, from a national broadcaster’s own site — and getting blocked because the site only allows connections from IP addresses inside that country. Even when the content is 100% legitimate to access, a normal home internet connection from the “wrong” country gets an instant no.
Once I understood how Tor’s relay system actually works, the fix clicked into place: if a download gets blocked for exactly that reason, the app can now automatically retry the request routed through a Tor relay that happens to be sitting in the right country. No separate app to install, no account to create, no manual setup — it’s built directly in. It only kicks in for that one specific situation (a location-based block), and it’s honest with you when Tor genuinely can’t help — a locked-down or paywalled video is a different problem entirely, and no amount of relay-hopping fixes that.
It’s a small feature next to everything else the app does, but it’s one of my favorites — mostly because building it meant actually learning how the thing works, instead of just knowing the word “Tor” and vaguely nodding along.
Tor and the Onion Logo are trademarks of The Tor Project. This post is my own account of learning how the network works — I’m not affiliated with the Tor Project.
Posted inUncategorized|Comments Off on I learned Tor, what it is and how I used it!
I was building a Linux version of Screen Capture — the free screen recorder I’d already shipped for Mac — and the first real test on a Linux machine gave me a video file that opened fine, played fine, and showed absolutely nothing. Just black, start to finish. Audio was there. The window I was trying to record wasn’t.
No error, no crash, no obvious clue. Just a black rectangle where my desktop should have been. That’s a specific kind of frustrating — the tool thinks it worked.
Diagnosing It: Ruled Out at the System Level
My test machine runs Wayland (Ubuntu 26.04, GNOME/Mutter), and the app’s video backend tries the modern, Wayland-native capture path first: PipeWire, via the desktop’s screen-cast portal. That’s the sanctioned way to record a Wayland session — you get a real permission prompt, and it’s what OBS Studio, Firefox, and Chrome all use for screen sharing on Wayland today. It’s supposed to just work.
On this machine, it didn’t. PipeWire capture came back black — and to make sure it wasn’t something dumb in my own code, I tested OBS Studio’s native PipeWire source on the same machine. Same result: solid black. That ruled out the app entirely. Whatever was wrong was below the application layer.
The actual root cause: my GPU is a GTX 1080 — Pascal generation, from 2016 — which is permanently capped at NVIDIA’s 580.xx “Legacy” driver branch. That branch’s DMA-BUF/GPU-buffer-sharing handoff with Mutter has a real bug. The compositor and PipeWire negotiate a buffer, but what actually lands in it is nothing.
This isn’t Wayland refusing to hand over the screen. PipeWire is the sanctioned, working path — it just hit a real bug in an old GPU’s driver branch, several layers below where the app or Wayland itself has any control.
Worth being honest about the outlook here too: NVIDIA’s Legacy branch gets security and critical bug fixes, not new DMA-BUF/buffer-sharing development — and Pascal is thoroughly aged out of their active branch. So this isn’t really a “wait for Wayland” problem. It’s a “wait for NVIDIA to backport a fix to a driver branch they’ve functionally stopped extending” problem, which in practice means it’s unlikely to ever get fixed on this hardware. That’s exactly why the app can’t just assume PipeWire will work — it has to check.
The Real Fix: Try PipeWire First, Validate It, Fall Back When It’s Broken
The app’s backend selection doesn’t just attempt PipeWire and hope. It attempts PipeWire, then actually checks whether real frame content came back — not just whether the pipeline opened without erroring, which is exactly the kind of false-positive that would have shipped a “working” recorder that silently produces black video on hardware like mine.
If PipeWire delivers real captured content, the app uses it. That’s most people — AMD and Intel graphics, and NVIDIA hardware on a current driver branch, should get Wayland-native capture with zero manual steps.
If PipeWire’s output validates as black or empty, the app falls back to FFmpeg’s x11grab instead — which needs a genuine X11/Xorg session to read from.
One trap worth naming here, because it’s an easy wrong turn once you’re in fallback territory: XWayland being installed does not mean you have a real X11 session. XWayland lets old X11 applications run as individual windows inside Wayland — it doesn’t hand x11grab a copy of the whole desktop. You can have XWayland fully installed and running and x11grab still has nothing usable to capture. I made this assumption myself early on, and it’s wrong.
Getting the X11 Fallback Working, When You Need It
If the app falls back to x11grab — or you just want to check ahead of time — this is how to confirm which session you’re actually running:
echo $XDG_SESSION_TYPE
If that prints wayland and the app has told you PipeWire capture isn’t producing real content, here’s what’s actually confirmed to fix it. Some systems show a “Ubuntu on Xorg” or “GNOME on Xorg” option behind the gear icon on the login screen — worth trying first, it’s the lightest possible fix. On my machine it wasn’t reliably there, so:
Install Xfce: sudo apt install xubuntu-desktop
If asked which display manager to keep as default, keep gdm3.
Reboot.
On the login screen, click the gear icon and choose Xfce as the session.
Log back in and re-check echo $XDG_SESSION_TYPE — it should now say x11.
Once that says x11, x11grab has a genuine desktop to read from, and the app’s own “Detect Capture Method” button will confirm it’s working.
The Extra Step: I Built an App That Figures This Out For You
Diagnosing a DMA-BUF handoff bug in a legacy NVIDIA driver branch isn’t something I want to hand anyone as “here’s how you record your screen on Linux.” So I built the actual app — a GTK4 interface driving FFmpeg 9 underneath, recording your screen and your microphone and the computer’s sound together into one file, same as the Mac version already did.
A few decisions that came directly out of everything above:
It tries PipeWire first, automatically — the correct, modern, Wayland-native path that most users on healthy hardware should get with zero manual steps.
It doesn’t trust that PipeWire “opened successfully” — it validates that real frame content came back before committing to that backend, so a driver-level bug like mine doesn’t silently ship a black recording.
If validation fails, it falls back to x11grab automatically, and a built-in “Detect Capture Method” button tells you right in the window which path is actually active — no terminal commands required to find out.
Everything is bundled into the download — FFmpeg 9, the GTK4 runtime, GStreamer, PulseAudio/PipeWire client libraries — because Linux distros vary too much to assume any of that is already installed. No package manager, no root password, no install script.
It’s a ~305 MB download because of that bundling. Extract and run: tar -xzf LinuxScreenRecorder_*.tar.gz, then ./run.sh.
Marked Beta, on purpose — it’s newer and less tested than the Mac version, and I’d rather say that upfront than have someone find out the hard way.
Here it is actually recording, on a real Linux machine, x11grab and all:
If your Linux recording comes out black, it’s worth knowing it’s probably not your fault, and it’s probably not even Wayland’s fault — it’s most likely one specific layer, your GPU driver’s buffer handoff, having a bad day. The app checks for that automatically now, so you don’t have to go find it yourself.
In the future should I get a different graphics card, I will try the Pipewire method again, cheers!
So there is this great UFO story out of Poland, it’s kind of famous also, but people seem to only think UFO sighting happen in America, no where else, how American and typical – just saying!
So anyways, I wanted this story to be better known, so I used a video toolbox I developed ffmpeg commander to first translate the movie into English with Burned Subtitles and this worked pretty well, but then
I was thinking, and this was about a Beer in and no more than 2 Beers out / I was thinking since this video was old probably 8mm source, so not Even VHS, that I could improve the old video first and then translate it.
below are a few screen shots so you can get a genera idea of the location where it happened, and there is even a monument to this event.
It happened on May 10th, 1978
With that said, the whole research taught me a lot about how video restoration even works, in the end I was pretty successful and the tool does a lot of what I wanted.
In addition to that there is a great web site that give a lot more details, turn on your language translation for this one because it is in Polish only…
Who designed it: Polish painter Andrzej Fogtt conceived the monument and made the first model. It was unveiled 15 October 2005, funded by Fundacja Nautilus.
Why a cube: Fogtt’s black cube is a direct reference to Stanley Kubrick’s 2001: A Space Odyssey — specifically the famous scene with the mysterious black monolith. The Polish art source states it plainly: “Czarna kostka na postumencie nawiązująca do słynnej sceny z filmu” (“The black cube on the pedestal referencing the famous scene from the film”). The cube stands for mystery and the cosmic unknown — borrowing cinema’s most iconic image of “something alien and unexplained has arrived.”
One honest nuance: in the actual film the artifact is a tall rectangular slab (the “monolith”), not a cube. Fogtt took that idea — the enigmatic black object that shouldn’t be there — and rendered it as a cube balanced improbably on one corner, which is why it reads as gravity-defying and otherworldly.
The whole monument is a stylized figure, too: the sources note the design reads as a simplified alien — the tapered granite base with its inclined “head,” topped by the cube. So it’s part homage to Kubrick, part abstract alien.
The Final Video you can see here –
Posted invideo editing|Comments Off on Emilcin 1978, UFO, The Visit, or on the Threshold of the Mystery // Restored with English Subtitles
Every photo and video on your phone is worth something to someone, just maybe not to you. Whoever is backing up your photos to the cloud is running your face through recognition models, scanning your location history, and training AI on your family’s memories, quietly, in the background, while you pay them every month for the privilege. I got tired of that arrangement a while back, which is part of why I already run a home server. Now that same server backs up every photo and video from my phone too, and I own the whole thing.
Hopefully I didn’t scare you! Read On!
The app doing the work is called Immich, and it’s free and open source. Think of it as your own private Google Photos, except it runs on hardware you own instead of a server farm you’ll never see. I run it on my TrueNAS box, the same machine that already handles Plex for my movies and music, so it slotted right into a setup I was already running.
The App Just Runs in the Background
Immich has a mobile app for both iPhone and Android, and once you install it and point it at your own server, it quietly backs up every new photo and video as you take them. I’m running an iPhone 17 Pro, and it grabs everything, including the big video files that eat up storage the fastest. It works the same way on older iPhones and on Android too. You set it up once and forget about it.
The Real Reason I Started Caring About This: Storage
Here’s the part that actually got me to set this up. Video eats phone storage faster than almost anything else, and if you shoot much of it, you know that “iPhone storage full” message always shows up right when you need to take something important. Once Immich has backed a photo or video up to my own server, I can delete it off the phone and get that space back, and I still have the original, full quality, sitting safely at home. No more choosing between deleting memories or paying for more cloud storage every month/year.
It’s Not Just Storage, It’s Actually Smart About It
Immich runs its own AI locally to organize everything, and it holds up well against what the big cloud photo services do, just without your data leaving your house to make it happen. Search “beach” and it finds your beach photos. Search a person’s name and it groups their face across years of photos automatically. Searcha cat, there a cat!, LOL
There’s a map view that plots where every photo was taken, so you can scroll back through a trip just by browsing the map. All of that runs on my own hardware. No subscription, no cloud processing, no company on the other end learning anything about my and your life, f them is what I say and you should be also thinking this, if you are not, you are a fool!
Worth Setting Up
If you already have a home server running for something else, adding Immich is a weekend project at most. If you don’t, it’s still worth considering, because the alternative is paying a subscription forever for something that’s quietly mining your family photos in the background, your kids, your wife, your pets, did I scare you yet ? LOL
I don’t know about you, but I’d rather own my own memories outright, and now my phone’s storage stays free for whatever I actually want to shoot next. Give it a try.