Weeknotes 94

I was up until 3:30am on Monday night recording and editing my RubyConf talk, but I nonetheless got the video submitted before the deadline as long as you measure “Monday” in the right time zone.

There are a million things wrong with it but it’s a huge relief that it’s done. I still need to publish the accompanying blog post and GitHub repo in time for the conference but that’s a couple of weeks away and feels much less daunting than the video itself. It’s done.

I made several regrettable decisions about the video. My wrongs:

1. I recorded the talking head footage in 1080p at 60fps. This was lots of extra hassle, produced huge video files, and in hindsight wasn’t really worth it. I should’ve either bought a better camera (or used my iPhone?) or settled for 30fps given that most people won’t notice or care.

   My webcam is unreliable and finicky and its naff bundled software is not well-suited to professional applications like this. The only way I could consistently record at 60fps was to start QuickTime Player capturing the camera at its default 30fps, open the naff bundled software and spam the controls until it hiccupped momentarily into 60fps mode, then hit record in QuickTime and hope that it stayed that way. Not the ideal setup when you’re trying to record a single continuous 30-minute take.

2. I tried to record a single continuous 30-minute take. I think it’s fair to say that this was exactly as stressful as making the film Victoria or that Russian one.

   I should have just structured the talk around self-contained sections that could be shot separately, or else made the stylistic decision to cut in and out of a full-screen talking head to hide the edits, but I didn’t realise any of that until I’d wasted enough time on bad takes that it was too late to change the design.

3. I recorded myself against green screen without checking that I could get a decent chroma key in ScreenFlow. I discovered afterwards that ScreenFlow’s chroma key tool is bad so I had to install the Final Cut Pro trial and do the key there instead.

   Final Cut Pro is tricky to use and it’s not obvious how to export video with an alpha channel. It turns out that ProRes 4444 supports it, but the resulting files are so massive that I could only fit one of them on my laptop at once. This led to the high-wire act of having to delete each failed take to make enough room to record the next one, running the risk that it would turn out worse than what I’d just deleted. Again I could’ve dealt with this better if I’d planned for it and bought an external drive to store these big intermediate files, but I hadn’t so I didn’t.

4. I bought a foot switch for advancing slides so that both my hands would be free to gesture while I talked. This was a nice idea but ultimately I needed to stand close to the camera so I could speak near the microphone without having to bring it into shot, so you could barely see my hands anyway. I should’ve just used a normal handheld clicker and nobody would’ve noticed.

5. I had to do noise reduction on the audio because I still couldn’t get quite close enough to the microphone to avoid a bit of hiss and room echo. This seems really hard to do well so I probably shouldn’t have bothered trying to fix it. Or maybe I should’ve “invested” in a lavalier microphone so I could both sound better and stand a more relaxed distance from the camera.

6. When designing the content for the slides I didn’t think much about leaving room for my superimposed head until it was too late, so I had to do some extra work to scale and slide my face to make sure it wasn’t obscuring too much. It would’ve been easy to just block out the bottom corner from the start if I’d bothered to consider this early enough.

I mean, I do try, but clearly making videos isn’t one of my strengths. I found all of the above unnecessarily stressful and, despite the lessons learned, I’d be very reluctant to do another prerecorded talk.

I’ve been using a new iPad mini 6 for the last couple of weeks. It’s really good.

The most obvious improvement over the ageing iPad mini 4 it replaced is that it’s just so much faster. This quantitative upgrade is large enough to become a qualitative one because now I can face doing things on it that I didn’t previously have the patience for, and even basic stuff like browsing the web or reading email is easy and quick enough that I do it way more often.

Spending more time on a smaller screen has made me realise how great the iPadOS 15 text size controls are — it’s not at all obvious, but you can easily set different font sizes for different apps. I love this because it lets me have nice high information density for tasks where that matters (e.g. email) while making more frivolous apps like Tweetbot big and chunky so I can easily scan them when I’m sleepy.

I wish Apple did something like this for audio volume. Being stuck with a single global volume feels anachronistic: it’s not a physical knob, so there’s no fundamental reason why cranking Netflix up loud should mean that Instagram ads automatically get to deafen me. Volume should be per-app, or there should at least be a configurable default that the system resets to on unlock or asymptotically approaches over time if no audio’s playing.

I’ve been enjoying the new Self Esteem album, particularly I Do This All The Time and the title track Prioritise Pleasure. The latter fits well with my personal motto, “fun is all there is”, which I say to myself all the time but don’t think I‘ve ever written down anywhere.

Although I generally enjoy the hubbub of city life, the easing of coronavirus restrictions has meant a return of the perpetual distant throb of bass from one source or another, which is something I could do without. During the day is fine but it’s hard to have a sense of humour about it at 3am or whatever. It rarely bothered me in the Before so I’m hoping my brain can slowly return to normal and stop getting distracted by stuff like this.