This week I published my first YouTube video. I’ve dabbled in vlog-style stuff before, but this is the first time I’ve made what I’d call a “real” video.

Do I love the output? Not really. But here’s the surprising part: I absolutely loved making it.

I might just be becoming a YouTuber.

Why video?

I’ve been on a personal branding journey for a while now (an article - or maybe even a video! - is coming on that soon). People keep telling me that video content performs really well and that I need to start doing it.

At the same time, I’ve also been told that your first 100 videos will be crap and nobody will watch them.

That sounds contradictory, but I suspect both are true: video performs well eventually, but the first stretch is just about getting through the awkward early learning curve.

So here we are: video #1. It’s a bit crap, but for a first attempt I don’t think it’s that bad. I both hate it and feel weirdly proud of it.

Here it is - I’d love to know what you think:

The making of…

From start to finish, this video took me two weeks - though most of that was procrastination. The actual production time was maybe 10 hours for a 21-minute video.

I didn’t set out with a topic in mind, but I did have two goals:

• Solve an actual problem that I (and probably others) have.
• Showcase some skills I’m proud of.

My day job is web development. I’m skilled in that area, but I find it hard to make content about it - the space is so crowded already that I’d just feel like more noise. That’s probably not true (and I’ll likely make web dev videos eventually), but for now I leaned on two hobbies I love that overlap nicely with my professional skills:

• Homelabbing
• Smart homes

Both require a mix of technical skills, both give me interesting problems to solve, and both are fun. Perfect for content.

The problem

I live on a busy road. Pulling out of my drive often takes minutes, and at peak times it’s rarely worth the effort.

It would be genuinely useful to collect traffic data and display it in a chart, so I’d know exactly how busy the road gets at different times.

That became the starting point for a three-part project, each with its own video:

• Part 1: Gather traffic data and chart it.
• Part 2: Use the data to highlight good times to leave the house, displayed as a widget on my smart home dashboard.
• Part 3: Add buses to the tracking and compare real-world data with my local bus timetable to see how accurate it actually is.

Part 3 was actually the spark for the whole project. A few weeks ago I needed to collect my car from its service. I walked to the bus stop, and watched the bus turn up ten minutes early. I missed it.

I resigned myself to the walk, only to see another bus drive past me 20 minutes later!

I realised that the bus timetable was wildly inaccurate and had the idea that tracking it might be an interesting project.

The solution

I already had a CCTV camera pointing at the road, with Frigate NVR capturing the feed. The plan was straightforward:

• Use Frigate’s MQTT integration to push events to my MQTT broker.
• Process that data in Node-RED (already running my smart home).
• Store it in InfluxDB.
• Visualise it in Grafana.

The results were better than I expected. Frigate’s AI doesn’t catch every single car, so I expected very rough data - but it turned out to be pretty accurate, even at picking up cars going the same direction in quick succession.

The video making process

I recorded my first attempts, but quickly realised I needed to rehearse a few times before hitting record properly. My first run was me fumbling with InfluxDB and Flux.

I also wasn’t happy with the video quality, so I upgraded:

• Camera: Elgato Facecam 4K
• Mic: Blue Yeti USB

That helped a lot, though not everything went smoothly.

I wanted a multi-cam setup

I love Joe Scott’s “tangent cam” style and wanted something similar: one camera in front, one to the side. My Mac just couldn’t handle two webcams, crashing every time I tried. I didn’t want to drag my PC upstairs, so I shelved the idea. Maybe next time I’ll use my iPhone for side shots or even invest in a standalone video camera.

I wanted to record camera and screen separately

I used OBS, which is great - but it insists on compositing in real time. There’s a plugin called Source Record that can record separate streams, but again, my Mac choked on it.

This caused a problem: in one section, my camera overlay blocked an important part of the screen. I had to reshoot that bit without the overlay.

Someone recommended screen.studio which looks brilliant, but it doesn’t support CCTV overlays and it’s not cheap, so OBS it was.

Editing was harder than expected

Initially, I tried recording in one take. A few people suggested splitting it into sections, which made filming easier. I even bought a clapper board as a joke, which turned out to be genuinely useful.

The downside: stitching it together. I often didn’t leave enough silence before or after clips, so the edits came out jarring. I also speak too fast, leaving little room to cut cleanly.

Audio was inconsistent: some clips louder, some quieter. I haven’t mastered normalisation in Premiere yet. Add in background noise - a cat knocking something over, a motorbike roaring past - and it wasn’t exactly smooth sailing.

And despite having a 4K camera, OBS spat out slightly grainy 1080p footage. Good enough, but not what I was aiming for. That’s another thing to fix before video #2.

It was SO much fun

Despite the hurdles, I came away buzzing. Making this video was fun. Really fun.

I learned a lot: scripting helps, sections are easier than long takes, and production quality matters more than I expected. Next time, I’ll probably work from a loose script and maybe rope ChatGPT into polishing it so I don’t ramble as much.

But the main thing is this: I enjoyed it enough to want to keep going. Parts 2 and 3 are already lined up, and I’ve started thinking of new ideas beyond that.

So yeah… I guess I’m a YouTuber now.

If you watch the video, I’d love your feedback. Tell me what I did well, but more importantly, tell me what I did badly. Leave a comment on the video, or if you’re worried about being too blunt (you’re not), drop me a message on LinkedIn.