Presently:

• My editor is poking at my footage of Code BEAM Stockholm Lite, very curious to hear what he has to say. I tried a bunch of new things. Discomfort, but also challenge, brain says "Yay!"
• Still need to dump (offload from devices and gather in correct place) and process (remove random shots of my kids) the footage from the local thing.
• Trying a new segment, will see how I like it, see box below.

Library of the moment

Evision
github | hex

I love Cocoa's work on this library. So the heaviest lift is by the library OpenCV which is, much like FFMPEG, global international infrastructure in the form of an open source project. Evision provides good Elixir bindings to this library with all of the powerful possibilities that entails.

You can do face detection, segment out people from their background, super-sample multiple photos, stitch, HDR. You name it, it does it. And a lot of more modern models and techniques get slotted in without issue. I recently ran into this U2net model and asked Cocoa about it. Suddenly I had a working example of the object detection parts and I've wrangled that to remove my background in a photo. You might see me poke that code in an upcoming video if it makes the cut.

Why should you get familiar with it?
Say your user should upload a profile photo. Would it be cool to be able to remove the background and make it transparent? Would it be cool to provide a recommended crop based on where the face actually is? Would you want to do anything with parallax, facial landmarks, background blur? For still images this is very doable. You could do so much with LiveView file uploads and this that it isn't even funny. Pretty exciting.

It also works well with Nx.

Meticulousness

I am not being down on myself when I say that I suck at particular things. It can be a bummer, but overall it is worthwhile to have an understanding of what you are good at and what you are bad at. Or to put it less limitingly, what comes easy to you and what comes hard.

Some of what comes easy or hard is a pretty clear-cut skill. Writing comes easy and quick to me. Talking about tech comes pretty easy. Designing systems on paper. Writing code. I find them all easy and well-oiled practiced things. There is some subjectivity as to whether I am good at them. But I don't struggle to execute on it or do it consistently. Skills I'm bad at are rather uninteresting because it is everything I haven't tried to learn and some things I have only kind tried to learn or where I didn't follow through. Why I didn't follow through is probably more interesting.

The other type of thing that comes easy or hard is more of a general approach and mindset. I find it easy to spawn an idea, build out what the prototype and the next three iterations should be and just start going off on building it. I start easy and I execute fast. This has been rewarded time and again by my surrounding environment and incentives. I really struggle with meticulous work. Deep design, intricate structures, sweating all the details. I appreciate that ability immensely in others. I love a well-designed thing and I do try to sweat the details as far as I can but the incentives have never been there and I'm not sure I have the brain-layout for it.

Some people can consistently clean their tools, mis en place their things, methodically implement a spec from A to Z. I have had moments where I've managed to do parts of this in a variety of ways. It has felt either special and magical or like hell itself. I should probably practice it more, I should probably work on lowering the incentives to hurl myself forward. But I struggle. I'm very used to the .. hurling (sorry?).

I struggle with long-term follow-through and simple repetitive tasks. If I can systemize it I can definitely do it but it has always been a challenge. I need external pressures for it.

I think I suck at meticulous mechanical work with my fingers because I have shaky hands. But also low patience for failure and fiddliness. My wife is brilliant at it, decades of knitting and a mind for details, a deep harmony with calm things. We are really different and complementary in a lot of ways.

I believe I generally produce good results. I've worked with enough developers where I've delivered code, libraries and they've been happy. I'm pretty sure I do fine in that regard. I think that mostly comes down to a lot of practice, meaning I can do it quite well even though I'm almost always moving faster than ideal. There is also good pressures in place with other people depending on what I do which keeps the worst impulses at bay.

I've worked with people who can't switch off sweating the details, to their detriment. One end is not better than the other. But I think I am off to the side of where the sweet spot is for producing the best results. I think this makes me very suitable for excitedly showing people things, getting ideas going and generally carrying on like I do. I often feel it hinders me in trying to build anything truly beautiful or impressive.

As a creator of things you also always know the warts. Some would be impressed with my videos. I know all the things I wish I did better. Some might point at Underjord itself, and I really like how it works and how I run it, but I am intimately aware of all the things I wish were already achieved.

I have a delusion that I want to execute to a very high degree, top of the line, leading edge. I'm not sure I actually want it. But I feel like I want it. And while I like my strengths just fine I am painfully aware of my shortcomings. I think we all deal with this. I guess there are all-rounders out there. I wonder what that's like.

What are you terrible at? More importantly, what are you really good at and how do you know? Reply to this email or at me on the fedi @lawik@fosstodon.org.

Thanks for reading. Much appreciated.