[ back to Tom 7 Radar ]

p
e
r
s
o
n
a
l
Aesthetic images, Skidder 2021, and c. (31 Jan at 22:21)
So goes January. That went by simultaneously quickly and slowly, as is the way of our times. Definitely looking forward to winter being over and getting vaccinated, although I'm hanging in there and finding ways to be healthy and modestly productive.

I spent most of my weekends on the latest secret project. It's a machine learning one, which has the usual contours: It does kinda work, but it's not magic, so you spend so much time fiddling with things. It takes forever to run because of the worldwide shortage of GeForce 3080s (thx scalpers) so you're still using the venerable but aged (pronounced age-ed) 1080. And since you're me, for some reason you have to write everything from scratch, so that takes also extra-long, and you don't benefit from decades of expert ideas and tuning, but on the other hand it's a different (and perhaps more pleasurable) kind of work, and perhaps you have a genuinely new idea from the naive exploration. But probably it just comes out mediocre. Anyway, my software has some nice visualizations; here's an aesthetic image:

Weights
Weights


Now I have at least three videos well underway, but as you've no doubt sussed this year has ushered a new kind of constipation in actually finishing them!

I took it a little easy (like 5k a day instead of 10k) on the running since 2020 ended, since I was having some knee trouble, and that seems to be completely better now. Nothing interesting there; winter isn't the greatest time for adventure runs, though I have made a few long trips out in my basically-infeasible new project to run all the streets in the county. I'll tell you one thing, the absurd hilliness is not confined to the city limits!

Like I've mentioned before over at Destroy FX HQ we've been updating our audio plugins to modern times, and de-squinting some of the tiny cruddy old UIs. I finally got done this redesign of Skidder:

Skidder redesign 2021
Skidder redesign 2021


Up next is the venerable but aged Buffer Override and then the never-actually-released classic BrokenFFT.

Gamey-wise, I played through ZeroRanger, which is a stylish and fun bullet-hell shooter, although towards the end I did get tired of its mean streak. I'm still in the midst of A Monster's Expedition, a great puzzle game. It reminds me a lot of Stephen's Sausage Roll, although the puzzles are much easier (they are still very pretty, but most of them are small enough that you can completely wrap your head around that feeling of "this must be impossible! unless....!") so it's less of an endeavor to play. Still need to get back to SSR one day! I'm also playing INFRA, which is an indie walking simulator about a city infrastructure inspector who goes around photographing code violations in its venerable but aged physical plants. I love the concept and I'm having some fun, but I wouldn't really recommend it. Finally, I finished the new Paper Mario on the Switch (was like you'd expect) and just took up Cadence of Hyrule. This is a sequel to Crypt of the Necrodancer, which I liked a lot and got fairly expert at, set in the Kingdom of Zeldaland. It's way more forgiving than Necrodancer—it's barely even a Roguelike since you keep a lot of stuff when you die (and other generosities). So it's already feeling quite easy, but it has many charms and lots of great Zelda pixel art.

I believe that is all!
c
o
m
m
e
n
t
jonas (catv-176-63-12-49.catv.broadband.hu) – 02.01.21 13:35:50
Might I recommend you the timewaster called shapez.io? This is a single-player video game played with keyboard and mouse that is an open-ended sandbox simulation game where you build factories that make abstract shapes. I recommend it not for its sandbox aspect, but for the two nontrivial puzzles that it comes with. These are your types of puzzle: ones where part of the difficulty is that the game doesn't quite tell you what all the rules are, and you have to figure out the remaining ones using experimentation. The first part of the game is the tutorials that teaches you most of the other rules. The game runs in browser. You can play the free demo version at https://shapez.io/ . The demo does not unlock all the tools for the sandbox part, but it unlocks enough that the rest of tools don't change the nature of the two puzzles. (After the speed upgrade screen changes so it no longer asks for the first 15 upgrade shapes, it will require three final upgrade shapes. Making the second and third shapes are the two puzzles. In the non-demo version, these also unlock level 20 and 26.)
c
o
m
m
e
n
t
Lucas (172.103.226.56) – 02.03.21 13:17:19
Have you written any games lately? I was flicking through your YouTube hits and noticed you had performed a bit in Ludum Dare events.
c
o
m
m
e
n
t
William (105-186-38-95.telkomsa.net) – 02.16.21 20:44:39
I just tried to read the memory address of a game boy advance emulator so I could store the data structure of a pokemon into some python shitpost, and I'm finding it to be really tedious. Any tips on how to make working with emulators not painful?

Also, do you have any recommendations for books that influenced your work?
c
o
m
m
e
n
t
Tom 7 (pool-74-109-237-238.pitbpa.fios.verizon.net) – 03.01.21 09:49:04
jonas, Shapez.io looks good! I added it to my wishlist and will give it a go once I finish some current games. Thanks for the recommendation.

Lucas: No new games for a while. I should really do a Ludum Dare event since I almost always come out with something worth the time, although in the last year or so I've felt more attracted to longer-lived "care and feeding" kinds of projects than the "unhealthy weekend" kind. So my video game work has been of the form where I write pages of notes on game mechanics in some notebook before bed. That's less fun for people other than me (as the paper is unplayable) but I like to think that when my brain truly sponges over in my old age I'll just be able to live off the prerecorded creativity in those books and ideas.txt, etc..

William: Most of the emulator code I've worked with has been pretty annoying. It might be a combination of the fact that the hardware being emulated is itself pretty annoying (so there are all these weird special cases and hacks) and the way those projects tend to develop. I only have significant experience with NES, but I did spend dozens of hours ripping out needless stuff from FCEUX and cleaning it up for my purposes, basically turning it into a library. (i.e.: Try to hide the stuff you don't need behind a layer of abstraction.) This both reduced the minor and major annoyances of working with it, and also gave me a lot more familiarity with what was going on under the hood! I don't know if it's the kind of tip you're looking for, but I will say that a lot of this stuff just ends up being a grind, and that figuring out how to successfully suffer through those grinds (whatever works for you!) is indispensable. Maybe there are some people out there for whom it's effortless, but it certainly isn't for me.

If you're asking about my programming work, I honestly don't think books were/are that influential. I do like to read non-fiction but for computer stuff, I think that books always seemed a bit out of date, and I think the best source of inspiration was either (a) trying to figure it out from scraps of code and text files pre-internet in the 90s, (b) classes and (especially) stimulating colleagues during college and grad school and then the same later at work. But, outside of programming (or more abstractly influential), some that randomly come to mind are "Thinking, Fast and Slow", "Anthropic Bias: Observation Selection Effects in Science and Philosophy", "Inadequate Equilibria", and the full anthology of SIGBOVIK proceedings.
p
o
s
t

a

c
o
m
m
e
n
t
[ Tom 7 Radar  •  Tom 7 on Google+  •  on Twitter  •  on Facebook ]