When You Can No Longer Read Your Smartwatch

No matter what smartwatch you decide to buy, I expect a significant proportion of non-athletes fall into a similar pattern with them: initial and continued usefulness from reading and responding to notifications; some initial interest in sleep tracking that trails off after a couple of months; and the occasional messing around with things like timers. And steps – 10,000 steps…

I got the first Apple Watch release, then after abandoning it as a solution to a problem I didn’t have I decided to try again about 5 years ago.

While I liked being able to leave my phone on silent all the time – something I adopted, and stuck with right from Day One – it was a pretty low ratio of utility relative to charging needs. After the battery performance started to drop off, I couldn’t find a daily pattern for charging it that suited me, so I decided to abandon ship for a Garmin Solar 2. Perfectly acceptable in terms of what I wanted, but with a 22-ish day battery life.

However, the LCD screen real estate is tiny and in the last 18 months I’ve found myself having to reach for my glasses to read the notifications.

This set me on a long drawn out, low-intensity piece of research: what about having something screen-less, but with a super long battery life. This was kicked off by looking at a Whoop. I have quite a high barrier for entering into subscription contract #643 in an apparently infinite series, so I abandoned that after about 30 seconds.

What I eventually stumbled on just before Christmas was something called a ‘Yanmis’ tracker. Amazon had a protracted shipping estimate (from China probably), so I had a look on eBay and found one for £14. It also has an estimated 40 day battery life.

The design is… compromised: in what is probably an attempt to look like a Whoop, the strap only attaches to one side. The app (QWatch Pro) appears to be generic, and used by a bunch of cheap fitness trackers. The documentation it came with was free of detail and – well, pretty much any useful information whatsoever. I migrated to Android and haven’t really looked into the permissioning model that closely. I’ve basically set the bare minimum, at least for notifications to work.

It’s still ticking all the right boxes for me. It also means that I can wear my mechanical watch again (or at least not feel like an idiot hiding two watches with long sleeves.)

Migrating off Windows 10 to Linux

There are probably tens of millions of people contemplating what to do with their PCs once Windows 10 goes end of life and, up until a couple of months ago, I was one of them. I had originally thought about upgrading the CPU on my 7ish year old machine to something that supported Windows 11. This, in turn, would have necessitated a new motherboard. I was estimating about £700 for it. I decided against it, partly because I nearly wrecked the motherboard fitting the heatsink when I built it, but mainly because my usage has evolved. It is to all intents and purposes a file server.

What clinched it for me was discovering that there is now full read / write support for NTFS on Linux. The last time I looked at it (probably more than a decade ago) it was read-only. This allowed me to shuffle multiple copies of data like photos while I did the install (the machine has a couple of M.2 SSDs and a fairly large HDD), and I can still use Grub to boot back into Windows for occasional emergencies.

I was really impressed with the NTFS support: it worked absolutely seamlessly… for a while. But it is a little glitchy over the longer term. I made a couple of knuckle-headed mistakes like managing to crash the default file manager on Debian (Thunar). It appears to be single-threaded for at least some operations, so trying to close its windows during file copies can have unhappy consequences.

After a while the volume auto-mount began to fail and then start working again for no apparent reason; then undeletable files started to appear in directories I was using frequently. I reiterate, this is undoubtedly down to me just not paying attention every now and then – but the warning signs were there.

It’s fine to get you over the line – so is read-only, for that matter – but I wouldn’t plan to use it for the long term.

AI and the Pixel Pro 10 Photos – Some Early Results

I have started to explore the camera and photo editing capabilities of the Pixel 10 Pro, which I took delivery of a few days ago. I have been a long-time Apple fan, so this is a return to an ecosystem that I’ve not tried in (my Amazon purchase history reliably informs me) more than 12 years – but that’s another story.

On the whole I’ve been very impressed with the camera. However, taking a rather drab picture and then trying to breathe some life into it has proved an informative exercise.

Pentonville Road Facade
Pentonville Road Facade

Ignore the leaves in the top right, which are knocking my eye out even now. The auto-erase feature works quite well, although there is a bit of a UI car crash if you are trying to draw around something that abuts the side of the frame. Let’s ignore that for now and focus on a couple of other details.

The first is some unusual artefacting on the edges of the window frame:

Although there is what looks like the signs of compression, there is a pattern around a lot of these high contrast boundaries. It reminds me of some of the weirdness I used to see with HDR software: the reassembly of the bracketed exposures sometimes went awry, and edges had strange shadows and overlays. My working assumption is this is a burst of images that are then assembled into a single shot – and not entirely successfully in high contrast areas.

What really piqued my interest was what happened when I tried the auto-framing feature:

Auto-framed wide exposure
Auto-framed wide exposure

Even though I didn’t select the widest lens, the camera took a picture with it at the same time. The framing is quite pleasing, from a basic geometric perspective. However, it has introduced a few unwanted elements into an already anaemic picture.

First is the lifeless building on the left; second is the tomb-stoning. Last, is the plain weird artefacts in the foliage:

Foliage artefacts

Having looked at the picture on a monitor, there are bands of distortion in the centre left and right of the image. I have no idea what’s going on here: there is clearly some sort of wrinkle in what, at least at fact value, appears to be a non-generative feature of the software.