Technology — Codex North

How to set up Windows 11 almost without killing yourself

Sat, 15 Mar 2025 00:00:00 +0000

Here we go again. New laptop, new Windows update.

When you boot up the machine for the first time, do Shift + F10 to open command prompt. Run oobe\bypassnro. The computer will reboot, and you can now make a normal offline account by saying “I don’t have internet” at the WiFi screen, after selecting region.

Then the begging for eavesdropping we’ve all come to love:

Say no to Windows Hello facial recognition
Say no to location tracking
Say no to location tracking again
Say “Required only” to diagnostic data (Such a fuck-you)
Say no to them spying on your handwriting
Say no to “tailored experiences” based on their spying
Say no to advertising ID (wtf)
Say no to presence sensing

And now we wait…

While we wait, I will say that it seems they’ve reduced the begging slightly from Windows 10. I can appreciate that. Though, this probably just means that they spy more by default, without presenting you the choice to opt out.

I won’t go into fully stopping the spying here, because as I mentioned in the Windows 10 article, these anti-telemetry tools all seem really sketchy to me, and they seem to get new replacements all the time. Therefore, I don’t feel I can safely vouch for any of them.

Finally, it finishes and we are greeted with the new pile of turd taskbar.

Imagine actually thinking it was a good idea to move the most important buttons away from the corners by default. First thing to do is go to Settings > Personalization > Taskbar so we can:

Hide the search button, because it’s completely useless
Hide task view, because we have Alt + Tab
Hide widgets, because they are retarded
And of course: Taskbar alignment > Left.

Now we go commit programicide and uninstall:

Capcut (Are you serious?)
Office 365 six times in Danish, Swedish, Finnish, Norwegian, English and normal(?), slowly, one at a time, because only one of these uninstall windows can be open at once
- By the way, I timed it. This garbage managed to waste 3 minute and 7 seconds to uninstall the Finnish copy of Office 365. On a brand new multi-thousand-dollar laptop!
Do that again five times for OneNote in Danish, English, Finnish, Norwegian and Swedish
OneDrive
LinkedIn
Solitaire & Casual Games (To be replaced by Zachtronics Solitarie Collection, of course)
Family by Microsoft Corp.
Outlook (New)
Journal
Microsoft Bing Search, Clipchamp, News, Teams, To Do, Whiteboard, Power Automate, Quick Assist, Weather, Xbox, Xbox Live

And now we drop by ninite.com for the essentials (Must-haves are 7-zip, VLC, K-Lite Codecs and Everything), before also grabbing:

FilePilot, because Explorer sucks, and they’ve made it even worse now
LibreWolf, or any other browser you prefer
PowerToys, for Always On Top and various other tools
AltSnap, to make Windows windows usable
OpenShell, not because it’s good (it’s not), but because my normal Windows search completely broke one week in and now never returns any results.
qView or Cactus Image Viewer, because Photos sucks

Do these registry edits (Here’s a script that does them for you):

Restore the functional right click menu: HKEY_CURRENT_USER\Software\Classes\CLSID\{86ca1aa0-34aa-4e8b-a509-50c905bae2a2}\InprocServer32 Only needs to contain an empty (Default) String Value.
Fix wallpaper quality: HKEY_CURRENT_USER\Control Panel\Desktop New DWORD (32-bit) JPEGImportQuality set to 100 in decimal.
Disable Window’s reaction to Win+Shift+F23 (aka. the “Copilot key”): HKEY_CURRENT_USER\Software\Policies\Microsoft\Windows\CopilotKey New DWORD (32-bit) SetCopilotHardwareKey set to 0. The Copilot key should now just bring up the start menu, and should be a lot easier to rebind to something useful with PowerToys or AHK.

This cat and mouse game keeps going. Microsoft keeps making their product worse, and we keep fixing it for them. It’s genuinely comical at this point how far removed from the reality of their user’s experience they are.

What in the fuck is a loom

Sun, 15 Sep 2024 17:02:18 +0200

I once worked with a group of people where using cool new webapps was all the rage, and I can’t stand it.

Loom? Miro?? Jira? Figma?? Nearby? Around? Veed?!??!? Yes, these are all real, and yes, none of them do anything new, beyond having a nice domain name and logo. Yet, they do expect you to make a business account and pay for their premium features on a monthly basis.

Remember IRC? MSN Messenger? Well, now you can have the same thing, but slower, and in your browser, if you pay for Meet, Around, Whereby, Teams, Slack, Zoom or Appear (One of these isn’t real). Video chat was solved 30 years ago. Putting it in the browser, and masking the webcam feeds to be circular is not a value-add, and is definitely not worth $15.97 a month.

Remember screen recording? Now you can have that for $12.50 a month per “creator”, but slower, and inside your browser, and with a version that doesn’t just give you a video file you can share. It gives you a link to a website, where there is a video file that other people might be allowed to access, as long as the servers are up, and they like having JavaScript enabled in their browser. Oh, and it’s been nouned, so what you have made is not called something confusing like “a video”. What you have made is of course called: “a loom”.

Now, remember, it’s not just you using this, it’s your entire company. After a while, when your company is nicely infested with looms everywhere, then all your internal documentation becomes 2-hour long rambly videos that you start linking one another. This is your life now. And re-doing all the documentation in a reasonable format is not an option, of course, because that would take time and effort. Better to just keep paying for loom.

Dear Lord, all mighty, wash this planet clean and set it ablaze for we are sinners far beyond saving

You know video editing, right? Now you can have that for $12 a month, but slower, and in your browser, and with restrictions like “up to 720 minutes of subtitles per year”, “Up to 1080p quality” and “up to 25 minute-long videos”, if you pay for Veed! For that price, they’ll even remove the ugly watermark they slap on top of your videos! And if that isn’t enough for you, you can subscribe to the Pro option for $24 a month, where you get 1440 minutes of subtitles per year! Wow! And if you pay us $59 a month, you’ll even get to export in 4K!

I am dead fucking serious. You don’t pay them to solve any real problems, you pay them extra to remove their own arbitrary resolution limits and watermark. This is a real company, and this is their real product in the year of our lord, current year.

Remember taking notes? Well, with Miro you can have that for as low as $8/mo.! But slower, and in your browser, requiring WebGL, with everything placed in virtual Post-Its, and with insanely annoying access policies that can be set per user account that you have to register to get access.

Oh, I forgot to mention! All these services need you to make an account with your email. They have EULAs, ToS, and privacy policies that you definitely will read very carefully. These companies and their data will definitely never get sold to another company with a shady past. That never happens. Not to mention, your employees will simply love making accounts on all these platforms! They’ll love all the emails they’ll get about “tips” on how to use the latest features!

Remember GitHub issues, and having a TODO list? That has a mediocre webapp replacement too! Jira! Now you can track your GitHub issues… on another website! Great. Don’t you feel productive?

In fact, that’s what all these services are about; feeling productive. They don’t assist you in doing anything at all. If you want to be productive, produce product. Go create. Do the work. Ship.

Take a Miro brainstorming session, for instance: Having everyone’s cursors show up on screen is just distracting. As is seeing everyone else throwing sticky notes around and accidentally occupying one another’s space. This gives a feeling of intensity, of busyness, of doing something! Despite probably being less focused on the ideas you are putting down, due to having to think about the location, scale and paper color of these ideas in a limited 2D space shared with others. Now that you have to juggle all these things, you feel like you’re doing more, which you are! Issue is, you’re doing less of the actual task.

I was going to find out how to do a specific thing in our Unity project. There was apparently an article about this on Notion, but in that Notion article, they decided to make graphs of the information with some embedded Miro boards. Because how cool isn’t it that Notion is actually just Chromium, so we can embed anything we want!? However I’m not signed in to Miro… inside Notion. So, I can’t see any of that information, and the login button just pops open my real browser.

Despite all of this seeming completely bass ackwards in the most obvious of ways, people still fall for it. They still pay for it. With the Netflix logo tattooed on their forehead and Coka-Cola on their cheek, they go around enthusiastically calling it “a loom”, while scoping out more funnels to fall into.

Using this stuff is a massive NGMI-signal. The people that don’t get swept up in any of this garbage are of a higher ilk. I have been blessed with a great circle of such shipper-friends, and encourage anyone wanting to actually achieve something to seek out the same thing.

Imagine using Pro Tools

Sat, 22 Apr 2023 20:53:07 +0200

Note: This article was written at a point in time. This information will eventually (hopefully) become out of date.

I’ve been working in Reaper since 2016. For working with audio (as opposed to composing electronic music), it is the best DAW I’ve ever used. In fact, it is one of the best pieces of software I have ever used in general. I don’t think any other DAW can measure up. And that’s big, coming from me, because aside from the notable exception of Cubase/Nuendo, I am probably the person I know who has used the most DAWs, and still do on a regular basis. LMMS, FL Studio, Ableton Live, Reaper, Renoise, BitWig, Ardour, I have and use them all. Some are good, some (like LMMS) are bad, but none of them are anywhere near as awful as the God of Shite itself… AVID’s Pro Tools.

A few months ago I had the displeasure of preparing a bunch of Reaper projects to be delivered over to a Pro Tools user, and he asked me what files would be mono and what files would be stereo, as if that might cause a problem. That’s not his fault of course, because he’s used to a DAW where that might be a problem. Imagine using a DAW where you can’t just toss stereo and mono files together on a track.

Imagine using a DAW that makes working with different sample rates something you need to think about at all. Well, I suppose you might have good reason to make people want to avoid resampling if your DAW has this fucking shite of an anti-aliasing filter. Here, compared to the resampling algorithm shipped with Reaper, r8brain.

Pro Tools 2021	r8brain

Analyses courtesy of Infinite Wave Mastering

All that purple you see in the background is noise, and that laser blast you see coming back down is aliasing. The retort is always “you probably won’t notice”. But trust me, you will, and when everyone else is doing a vastly better job than you, you have no excuse to be this bad. Especially when you’re charging many hundreds of dollars for your program every year, all while getting beat by a buy-once-own-forever, $60 DAW with extremely lax DRM.

Imagine using a DAW where people don’t render, they play back their whole project and re-record it to a new track to avoid problems with “using the bounce mix feature”. Yeah, what a feature. Imagine even speaking about faster-than-realtime rendering as a “feature”. It was added to Pro Tools as a “feature” in 2013, where as Cubase has had it since the 1990s.

And for anyone who does anything vaguely challenging for your CPU, rendering by re-recording live playback is an absolutely idiotic idea. Imagine recording a file and getting underruns. I suppose it’s a good thing that Pro Tools by default stops everything it’s doing and throws up an error box if one single underrun occurs. (An underrun is when you hear a click/pop due to high CPU load)

Imagine having ownership of your DAW be bound up in a subscription fee and fucking iLok. Imagine not being able to just download your DAW from the website and install it in mere seconds, if you ever need to. Or just carry around your more-than-fully-featured DAW setup on a memory stick because the entire installation is less than 140 MB, as opposed to the requirement of 15 fucking GB for a Pro Tools installation. That’s more than 100x larger, with far less features and far worse user experience!

Imagine being able to put up to 10 whole effects on a track! Wow! 10 entire effects! All on one track?? You bet’cha! 10 entire effects, all on a single track. Isn’t that amazing? (sarcasm)

Imagine not having industry standard normalization tools built in to your “industry standard” DAW.

Behold, Reaper's industry standard normalization tool

Imagine being happy that in 2022 you were finally allowed to change your keyboard shortcuts.

Imagine being told to disable Hyper Threading because it can cause CPU errors in your DAW.

Imagine paying $600 every year ($300 every year if you’re ok with having a “Limited audio post-production workflow toolset”) for this absolute hunk of garbage. (And that’s with the 50% discount for paying upfront for the full year!)

Imagine using Pro Tools.

PlasticSCM: The Nightmare SCM

Sun, 12 Feb 2023 19:15:19 +0100

Context

I’m working as a composer for a videogame. We use the Unity engine (Yes, I know. Not my decision.), and something called PlasticSCM for source control. It is the single worst SCM software I’ve ever experienced. It makes git feel like absolute heaven.

Constant empty changes

The Plastic people supply their own GUI, which is nice. The GUI is alright. At least it has a dark mode. But no matter how many times you tell it not to give you update popups, it still gives you update popups every time it launches.

In this GUI, you have a list of uncommitted changes. Or as they call it, “Pending changes”, because there is no such thing as a commit. There are only changesets. When you “commit”, you also push. You have to push when you commit. These push-commits are called changesets.

Well, this list of pending changes is constantly getting filled with unchanged files. Files you never touched, but they’re there in the list, as if they’ll be part of the changeset. You can click on them and see the diff. Nothing. No difference. They’ve even added a button you can press to remove all the changes that don’t contain any changes. Incredible. This is true innovation. With two clicks of your mouse you can revert all your non-existent changes to how they were before you didn’t change them. Exactly as they already are now.

And if you don’t do that it can actually cause problems. One time, I was going to do the equivalent of a pull, and it said that the lead programmer had made changes to the master soundbank. Which is completely stupid, he would never have.

Luckily, this was a small changeset, and I knew to look out for this. If I didn’t, it could’ve easily done a bad auto-merge with some old version of a file nobody noticed that Plastic had erroneously added to the changeset. Et voila, mysterious bugs!

The Catch 22 Lockup

I tried to checkout another branch once, and all the changes it made to the files in order to get me to that branch, it thought I had done and was waiting for me to commit them. 884 pending changes I never made.

If I tried to revert these changes, it threw an error saying it didn’t know what to revert to.
If I tried to change branch back to where I was, it threw an error saying I had uncommitted changes.
If I tried to temporarily shelf the changes, it threw an error.
If I tried to commit the changes, it threw an error.

I literally had no way to recover, and all I could do was download the project all over, and set it up again. And because it is a Unity project, that took multiple days. I had heard Unity was slow, but holy shit.

This is it trying to open the project for the first time. That number means 13+ hours. This also happens when you change the build target. Download the project, try to open it, leave that over night, change the build target, and leave it over night again.

This is it literally just quitting. It sometimes takes 5 to 6 minutes to just quit the program. Usually it takes just a few seconds, but on occasion it just sits there for multiple minutes.

Selective merging

After doing that, and finally starting to get back to work, it was time to merge my changes with the progressively named “main” branch. I needed to merge all the stuff I had added and removed, but I didn’t want to touch the files that the rest of the team had been working on. Good thing there’s absolutely no way to do that.

With git, all you need to do is checkout the branch you’re wanting to merge into, and from there, checkout the specific files you want to bring over from your branch.

In PlasticSCM, however they have two options: Merge, and Cherry Pick.

Merging means you have to merge everything, and if the program thinks it can auto-merge your files, it will. Even if you go to the merge options and tell it you want to do it manually. After merging everything, the merged things go in the pending changes list. If you then try to undo the changes to specific files, it won’t let you because you can’t undo just some of a merge. You have to undo all of it at once. WHY?? JUST LET ME DO THAT, IF IT IS EXACTLY WHAT I WANT TO DO.

The other option, Cherry Pick, doesn’t let me cherry pick anything, it just randomly happens to merge only the exact files I don’t want it to merge, with no actual way of selecting what files I want to merge. It just picks the two files I don’t want to merge, and I can’t do anything about it. All the files I’ve added it just leaves behind. It’s as if it only can see the changed files, not the added files. Maybe because those files weren’t there in the master branch before my branch started? No clue, but either way, it leaves the cherry picking option useless.

My solution: Copy the files I want to merge to another folder, checkout the branch I want to merge into, and copy the files back into the repository. All the while praying I don’t get stuck in the same catch 22 lockup I was earlier when changing branches.

Diff and merge tool

When the merge tool has you resolving conflicts, it seems to copy and rename the files it needs to diff into a temp folder before launching the merge tool. This means that when you look at the incredibly unhelpful file titles of “Source”, “Destination” and “Base”, you will have no help to understand what files are what, because they’re all named “D:/Something/PlasticSCM/temp/F79QPFG.temp”.

Even better is this thing:

Which has me resorting to the command line version of Plastic, which is fully capable of giving me the diff. What makes the GUI so much more handicapped is a mystery.

Goodbye Babylon

The project is nearing its end now, though, and I can’t wait to go back to git. I’ve been wanting to try out Sublime Merge, because even as a vim user of many years, Sublime Text has recently impressed me.

Twitter's fractal Rube Goldberg machine of bloat

Tue, 31 Jan 2023 05:47:11 +0100

In my previous post about software bloat I ended on a somewhat unsatisfying note.

Is there an actual solution?

I have no clue. Refuse to write and use shit software, I guess.

However, we might be seeing another, much brighter light at the end of this tunnel unfold before our very eyes: Twitter!

Of all places, I had never expected a code-bloat reduction revolution to come from silicon valley. Especially not a web company. All it took was a controversial acquisition from the Elon Musk of our time: Elon Musk.

If you haven’t been keeping up with these news, Elon is trying to do a few things:

Optimize for unregretted user hours
Bring free speech back to the platform
Actually become profitable for once

The latter requires not having thousands of unnecessary engineers on payroll, like they do now, even after his initial downscaling. Doing that requires simplifying the codebase massively, because with a bloated codebase, you need the bloated workforce that created it. Untangling this mess is incredibly difficult, because nobody knows what’s going on in the codebase. Nobody knows what anything does other than their in their exact department. Twitter is a many-million-line behemoth and it doesn’t need to be.

Luckily, Elon seems to be conversing with actually skilled engineers about these problems, such as George Hotz and Jonathan Blow.

There is a natural crossing between the code reduction they’re doing, and the goal to bring back free speech. They seek to undo the huge mess of automated censorship and suppression models/algorithms that exist in Twitter’s codebase. For more about this, you can read Dave Rubin’s thread from when he had visited Elon and the Twitter gang. [Nitter link] It’s worth the read, because it is interesting for a whole bunch of reasons other than software bloat.

I asked Elon what I could share and he said, “anything that’s true.”
— @RubinReport, 6:51 PM · Jan 26, 2023

Basically, over the last couple of weeks they had gotten a vague sense that something was wrong. The numbers just didn’t seem quite right. That’s how complex this system is now, that things were good for a bit, and now they’re not, and nobody knows why. Dave Ruben came over to talk with them about this, because his account was affected by this downturn in numbers. Then, when digging around in the codebase after multiple weeks of day-and-night work by engineers at the top of their game, they discovered additional ways that the system was suppressing people on the platform. They didn’t know this was happening. They discovered it.

Accounts aren’t just hit with labels that are obvious to insiders. They now found more “secret” labels which are causing shadowbans. My account was hit with all three; “Recent abuse strike,” “Recent misinformation strike”, “Recent suspension strike.”
— @RubinReport, 7:03 PM · Jan 26, 2023

It’s unclear so far what these strikes actually do, but for sure they suppress views and recommendations, they are trying to figure out to what extent.
— @RubinReport, 7:06 PM · Jan 26, 2023

Do you have any idea how much of a problem it is when you dig around working ‘round the clock for multiple weeks to suddenly discover unwanted functionality in your codebase? It’s a billion line dumpster fire. Elon describes Twitter as “a fractal Rube Goldberg machine”. For every problem they solve, they discover more problems. And I’m sure he’s right. It’s what I’ve come to expect from big software companies at this point.

So, about that unsatisfying end note. This one will be more satisfying.

If Elon succeeds in his mission here, we will have the first big public example of how to do code cleanup at a massive scale. I really hope he does, or at least gets the total rewrite he’s talking about. And I really hope he survives the stress of seeing it through, so he can share what he learned along the way. Getting to mars, solving AI alignment and all that is nothing compared to trying to un-need 20 million lines of Scala.

Ok, marginally more satisfying at least.

Professionals should use 32-bit float recording

Sat, 04 Jun 2022 04:11:57 +0200

TL;DR: In choosing a solution to clipping, 32-bit float is simply superior to limiters.

What is 32-bit float recording?

Digital Audio

Standard digital PCM audio is 2-dimensional. One dimension is amplitude, the other is time. When you have different amplitudes over time, that is a waveform. Both dimensions need to be stored in the computer using numbers with limited resolution. The resolution of the time dimension is the sample rate, measured in kHz. The resolution of the amplitude dimension is the bit depth, measured in bits. Every sample is a number, stored in a unit that is the size of the bit depth. These samples occur at the interval set by the sample rate, so commonly 44,100 or 48,000 times per second.

(It should be mentioned that with dithering, the effective resolution of any bit depth is preceptively “infinite”. This comes with the trade-off that you’ll be introducing noise at the lowest possible volume level you can store in that bit depth. Meaning that increasing bit depth doesn’t actually increase resolution, but instead it decreases the necessary level of dithering noise. However, because the electronic and environmental noise in our recordings are way beyond dithering noise, you can ignore this.)

24-bit integer audio

In the common 24-bit audio file, these samples take up 24 bits. What is key about the 24-bit audio file is that it is an integer number. The only thing a 24-bit integer can store is whole numbers between 0 and 16,777,215. Signed integers sacrifice one bit to say if the number is negative or not, so they can go from −8,388,608 to 8,388,607. The number of unique numbers a 24-bit integer can store remains the same either way. 16,777,216 unique numbers. In PCM audio, these numbers represent the range between -144 (ish) dBFS, and 0 dBFS. If you try to go above 0 dBFS, it clips. When you only have 24 bits, and they’re all 1, then you can’t add another 1 to make louder. Which means you have to avoid clipping at all costs.

That is why we use limiters. Instead of letting the waveform cut off at 0 dBFS, it warps the dynamics to avoid clipping. This is an imperfect way to work around the limitations of storing audio using 24-bit integers.

32-bit float audio

32-bit float is fundamentally different. Not just because it is 32-bit and can therefore hold a vastly larger set of unique numbers, meaning a higher resolution. But because it can store decimal place numbers, in addition to extremely large numbers. That’s what “float” means. Without getting into the technical details, they’re awesome. The smallest 32-bit float number, other than 0, is 0.00000000000000000000000000000000000001175. There are 1,065,353,215 unique float values between 0 and 1. Although not linearly distributed, this is way more than the 16,777,216 total unique values of 24-bit.

Will you hear this increase in resolution with your ears? In most cases, no.
Will it technically be better? Yes.
Is that all? That’s it? Oh, hell no.

The way 32-bit float audio is implemented means that it can store values above 0 dBFS. That’s beyond clipping. With the right recording hardware, it is possible to record audio that is “clipping”, but not really, because the information is still there, available for recovery. This means you don’t have to use a limiter and warp the dynamics of a recording to avoid clipping. No more messed up dynamics, no more clipping, and higher amplitude resolution.

What is the downside of using 32-bit float audio? It requires 8 more bits pr sample, compared to 24-bit. However, audio is so cheap on storage anyway, that a 33% increase in size is completely ignorable.

With these benefits, you might imagine audio professionals like us would be scrambling to make use of this great development. No more need for limiters to catch the occasional loud laugh or loud impact sound! However, no. Watch or read anything about a 32-bit float capable recorder from a location sound mixer, and you’ll hear nothing but hesitancy and people brushing it off as a “neat fallback for noobs and kids who don’t know what they’re doing”. That’s why this article is not just some info about 32-bit float and why it’s cool. It is to show that it is actually something professionals should be excited about. Not gawk at as though we’re too good for high quality audio encoding.

Answering the hesitancy

“I don’t need it, because I’m good at my job, and professionals don’t clip their audio.”

Don’t flatter yourself. You trigger the limiter every now and then. Though, by this logic, you surely turn off said limiter on your recorder too, right? Just in case you won’t need it? After all, professionals don’t clip, so what would you need a limiter for?

Our job is to deliver the best possible quality sound. Not as low quality as possible before post production or producers complain and ask us to do better. So get out ahead of the curve. Use the highest quality format available. When a bad workaround like a limiter is replaced with an actual solution to the problem of clipping, use it whenever possible.

“Camera people don’t constantly shoot in raw.”

32-bit float is not raw. It’s vastly, vastly superior. This is a terrible analogy.

Audio is cheap, video is expensive. Audio is 2D (amplitude across time), video is highly multi-dimensional (R-amplitude, G-amplitude and B-amplitude, across X and Y, across time). Audio is always captured uncompressed, video is nearly never.

Cameras still blow out when shooting in raw. If you try to clip using a decent 32-bit float recorder, you’ll hit the limit of the amplifier or microphone first. If that was how raw worked, it would be the equivalent of giving the camera sensor enough dynamic range to capture something so bright that the lens would break. While also putting an end to all banding issues. And all that while barely making a lick of difference to the file size, or how slow it is to edit! In fact, editing with it is faster than 24-bit, because 32-bit lines up better with current CPU’s word lengths. If that was how good raw recording was, you’d never turn it off! However, that’s not even close to how raw works. So they don’t. It is how 32-bit float works, however. So use it.

“My recorder/mixer can’t do it, so there is clearly no demand for it in the industry.”

There were no good train routes before the tracks were laid, so people clearly didn’t want trains. And besides, when did we start thinking it was a good idea to only upgrade when it is demanded of us? In my book, if you were slower than “the industry”, and had to be asked to upgrade, you were too slow.

“Why change the workflow, when it already works just fine? “

The simple answer is: There is no workflow change to be made. There is no adjustment to be made, no software to learn, no workflow to change. Everything just works exactly like it used to, except now, you can instantly recover peaks above 0 dBFS.

“Plugins and software don’t ‘work with’ float anyway!”

Yes they do. Adobe Premiere Pro, DaVinci Resolve and Final Cut Pro all work just fine with float. As does any DAW that isn’t complete trash. Any decently modern DAW like Reaper also does all of its internal processing in 64 bit float. Plugins all mostly work using float too. Try pushing a way-too-loud signal through an EQ. It isn’t digitally clipped when coming out the other side. You can still turn down and recover the signal. Et voilà, you just revealed that the plugin is using some form of float.

Reasons not to use 32-bit float

“On my recorder, it increases the noise floor.”

Fair. On some recorders, like the SoundDevices MixPre II series, using 32-bit float brings up your noise floor a bit, making it a sidegrade instead of a simple upgrade. This means that it has a downside in some cases, and that you should consider whether to use 32-bit or the limiter.

“The post people are stuck using OMF.”

OMF somehow doesn’t “support” 32-bit float. A timeline interchange format like this shouldn’t need to “support” any format, because it shouldn’t need to know what format any of the files are. It should just tell you where in the timeline they are placed, and leave the files alone. OMF doesn’t because it is a retarded format made by Avid. Don’t use it. Use OTIO, XML or AAF.

“I’m running an all-digital all-wireless system, which is 24-bit.”

Fair enough. In this situation, you won’t have any use for 32-bit float at all. Let us all pray that the Audio Ltd. A20 system will be capable of transmitting 32-bit float.

“Ok, great but my recorder actually doesn’t do 32-bit float.”

The current line of pro-level recorders were released before this recent wave of 32-bit float recorders came out. So they can’t record in 32-bit float yet.

Sound Devices have this to say on their website:

The A/D converters used on the Scorpio, 888, and 833 are 32-bit resolution parts, but this is unrelated to 32-bit float files. The ability to record in 32-bit float is a future possibility.

I have no idea what this means. The first part seems to imply that these 32-bit ADCs are not built to do float recording, but the second part seems to say that it only takes a firmware update to enable the functionality. And I don’t know which is true.

I imagine they could leverage their great limiters, and program something that perfectly amplifies in the digital stage to compensate for the limiter’s attenuation. Returning in the digital stage, the dynamics that were lost in the analog limiter stage. Maybe. This depends on the power of the ASICs they use, and the bandwidth of the interface between the digital components and the limiters. This is all speculation and daydreaming on my end, of course.

In closing, 32-bit is good. You should use it if you can.

RSS: Savior of The Internet

Tue, 10 May 2022 00:11:13 +0200

What is RSS?

It’s like podcasts — but for reading.
— NetNewsWire

It stands for Really Simple Syndication (Or RSF Site Summary, if you’re a nerd). Basically, it lets you gather updates from all sorts of websites and Internet-channels into one program/feed. Any blog, news site or Internet-channel you want to follow, just grab the RSS link, put it in your RSS reader of choice, and you get all of it in one simple feed. No need to visit a bunch of different sites to see what you care about. It’s all there in one place.

YouTube channels have RSS feeds. This is PewDiePie’s RSS feed. NRK has a bunch of RSS feeds you can follow. Any decent blog has RSS feeds. For instance, the singularity is nearer has an RSS feed.

And of course, I have RSS feeds for my website. One for the whole site, and one for each tag.

Just go to any tag listing (Such as Technology), and look for this icon:

RSS is an old standard. A protocol. It is NOT a service or website. There is no RSS Premium. Anyone can put an RSS feed on their website, anyone can read that RSS feed. Nobody can stop the author, nobody can stop the reader. It’s freedom. And it’s completely site-independent. You can mix and match RSS feeds from wherever you want, however you want. You get to choose how they’re categorized, organized and displayed.

Why use RSS?

You know all this talk of censorship and/or content moderation and social media algorithms? Conservatives saying they’re getting censored, leftists saying they’re not being protected enough. People being addicted to likes and interaction counters. Facebook manipulating what you see to manipulate your mood. You’ve seen the debates, articles and documentaries.

Well, the solution has been right under our noses this whole time. Just use RSS and be happy. As a publisher, it’s on your own website, so YouTube won’t delete your account, or your videos. As a reader, you get to actually see everything people share, unfiltered. You won’t be randomly unsubscribed, or randomly miss notifications. It is all in the user’s control, and that is sadly becoming rare nowadays.

So, to help keep the Internet free, and user-controlled, use RSS.

Better “content”

RSS is more than just freedom, philosophical superiority and convenience, however. The content¹ you’ll find when browsing the web with an eye out for RSS feeds, is more likely to be of way higher quality than the stuff you’ll passively consume on social media. It’s the difference between being force fed spoons of refined sugar, versus organic grass-fed local game you hunted down yourself.

“But… but… scroooolling on TikTok feels so goooood”

I’m sure it does, but I’m also sure you know you shouldn’t be doing it. Especially when you snap out of it, wondering what happened to the previous 4 hours of your life.

You could start by having a look at my recommendations. After that, browse the web, follow links between random people’s websites, explore, and you’ll discover interesting stuff. Many people (such as Zoraster) maintain a list of other good websites to check out. Those are great places to look. Infact, if you find something really cool, send me an email. I’d love to hear about it.

How to use it

Download an RSS reader.

There are lots of great RSS readers to choose from, but here are a few good ones:

Android
- Feeder
- RSS Reader
Apple stuff
- NetNewsWire
Linux / Termux on Android
- newsboat (My favorite)
Windows
- No idea, I just use the Ubuntu terminal and newsboat again

Your quest

Grab one of the readers from above.
Find an RSS icon on my site.
Right click it, copy link, put it in your reader.
…
Profit, and save the web.

I don’t like the term “content” because most of the stuff called “content”, especially in the media industry, has absolutely zero actual content in it. It is only made to make you think about a brand. Vapid, empty, shameful spam, made and consumed with zero respect for the human soul. ↩︎

Replacing Adobe

Sun, 08 May 2022 21:25:08 +0200

Here’s a bunch of reasons why I’m really starting to hate Photoshop and Adobe’s software generally.

The obvious reasons is that it uses a parasitic monthly subscription business model. Everyone and their moms already hate Adobe for this, but the actual quality of Adobe’s software is tanking. I say this as someone who has grown up using Photoshop since I was 11 years old.

They killed my boy. I shall avenge you, CS5.5. The greatest there ever was.

Photoshop

My once beloved Photoshop has taken an absolute nosedive over the last decade. They added an ugly and slow start screen that wastes my time trying to load previews of old projects. Get out of my way. Just let me get to work. Please stop trying to make me watch your stupid tutorials. Stop auto-updating everything so I lose my plugins. Stop pushing your stupid vendor lock-in cloud storage on me. I will not store anything in your cloud.

In Photoshop 2022, the normal save menu is now replaced by some fucking “Save to the cloud” menu, where a gray button in the corner lets you save “On your computer”.

They are pulling a Microsoft, and replacing fully functional and fast menus with slower, bloated versions, that are missing basic functionality. Just like the new Settings menu in Windows. Control Panel is much better, more comprehensive, and faster, but they’re still pushing their awful replacement.

There is now a modern “export” button, and a “Save for Web (Legacy)” button. The modern one is extremely slow to load, and isn’t even 10% as useful as the original, being completely lacking in the ability to save and customize the compression of GIFs. The “Save as” button used to let you save as any file type. Now they’re pushing you to use some modern, slow and unintuitive export menu, or press something called “Save a copy”, which brings up the actual save menu. They say their hand was forced by Apple, but I don’t believe there was no way around the API change, and they certainly didn’t need to break the Windows version in the process.

And to top it all off, Photoshop feels like unfinished beta software when trying to work with 32 bit float images.

Which is my segue to talk about GIMP!

This is one of the main reasons I’ve had to pull out GIMP, because while it has a bunch of usability problems, such as not letting you select multiple layers, at least it almost fully works with 32-bit float images. This became very relevant to me on a fairly major (by my standards, at least) VFX project. Photoshop was absolutely shitting itself trying to make a 32-bit float clean plate of a scene, and so GIMP was the only viable option. It worked great in comparison. And I can compare, because I’ve been using both for many years. Photoshop on Windows and GIMP on Linux.

There are many other things GIMP is better at too, though.
The gradient tool leaves the line you drew on screen, so you can move and manipulate it afterwards, and it lets you change all the color stops along it. You have much greater control of gradients and how they’re rendered generally in GIMP.
The pencil tool (which is a brush tool without anti-aliasing) displays as snapped to pixels. In Photoshop, it floats along with the cursor, being able to be put in-between pixels, even though it can’t draw that way.
Marquee selections can be transformed after making them.
Unique shortcut keys for each tool. If you press R, it selects the rectangular select tool. Every time. Consistently. Unlike Photoshop, where every time you press M, it selects a different version of the marquee tool.
Brushes take effect outside of the image bounds. In Photoshop, a brush stroke just gets abruptly cut off at the edge of the image. This is a consequence of GIMP’s individualized layer-size system. And a vastly superior selection of blending modes, and much better RGB curves, and so much more.

However, GIMP is not perfect, and is lacking a bunch of Photoshop’s nicer features. Such as: Content Aware, non-destructive layer effects, adjustment layers, Save for Web’s GIF compression options, an actual GIF workflow at all… (To do something similar to what Photoshop offers, I suppose you’d need to use Trout’s GIF Optimizer.)

There is another competitor that does have those things, however, and that is Affinity Photo. A commercial software, but very affordable, and they actually let you buy the program.

I’ve been using Affinity Photo since v2.0 came out in November of 2022. It is very good. Definitely faster and more comfy than modern Photoshop. It has its own equivalents for Content Aware, layer effects and adjustment layers. Infact, Affinity’s layers effects and effect layers are way better, because they lean much more heavily into the non-destructive workflow. It actually feels more like After Effects sometimes, where I’ve just got a bunch of live effects stacked on layers and adjustment layers.

For its speed it does make some sacrifices though. Many things seem to render using an estimative heuristic, rather than what the final “render” would use. So unless you’re zoomed in to 100% or more, I never feel like I can trust the final image to look exactly like what is on my screen. The noise effect, for instance, will render a different noise pattern depending on your zoom level. And about 60% of the time, when you paste an image, the 3 columns of pixels at the right-side edge of the image, will be glitched over to the left side of the image. These things will hopefully be ironed out over time, but as of now it is not as robust as GIMP or Photoshop. It is my new daily driver.

After Effects

If you work in VFX, you should have replaced After Effects with Fusion, Natron, and/or Nuke a long time ago. They are vastly superior for compositing and VFX. And they don’t require that one god-awful single-threaded plugin implementation of OpenColorIO to do proper colorspace manipulation. Of the three, I found Fusion the least user friendly, but they’re all massive improvements. Natron is a clone of Nuke, so it is also more comfortable to work with than Fusion in my opinion. Fusion has weird names for things, makes a lot of small things really annoying to do, and requires a stupid USB-dongle to be plugged in while using it. Or else it will silently sneak in random frames of noise into your renders, leaving you to wonder whether it’s a bug, or if your GPU is failing. Took me hours of research to figure it out myself. They could’ve chosen to give you a pop-up saying “Please re-insert the USB dongle”, but no. They chose violence. DRM-violence. Apparently it is possible to buy it using a product key, but I couldn’t do that in my region.

Cavalry is the only alternative I know of for motion graphics and that sort of animation work. It is sadly another fucking subscription-based parasite, but at least they give you a free version thats main limitation is a max resolution of 1920x1080, and withholding random features like the 2D physics sim, API access and render management. I’ve tried it out a bit, but the their GUI is kind of janky, and the user friendlyness is massivaly hampered by their focus on responseness and exporting to SVG and web and responsive stuff. This means you can get it to do a lot more cool procedural stuff while keeping everything vector-based, and responsive, but the layers and other things just don’t work how you would reasonably expect from every other program. I’m excited to see where it goes in the future, but it certainly has issues at the moment.

I have also been wondering if TouchDesigner is a viable alternative, because it does let you render videos. It is perpetually licensed and a lot faster than every alternative, being that it was created for real-time graphics. It might be cool for compositing, but I’ve not had the chance to try it significantly for hand keyframing.

Illustrator

Inkscape has recently started to not suck ass. I’d use it.

Affinity Designer by Serif is also promising. Again, I’ve been using v2.0 since November 2022, and it is very good. Way faster than Illustrator. In fact, it was my saving grace when I had to work with an architectural project that had hundreds of thousands of shapes in it. Inkscape just crashed, Illustrator was completely unusable, and Affinity Designer was just medium slow, which was enough for me to be able to hide some layers and make it manageable. Designer is still lacking a few features, such as image tracing. Inkscape has you covered there, though.

It really is too bad that Affinity software doesn’t run on Linux.

Lightroom

The new Lightroom is terrifyingly big-buttoned and bloated-looking, and forces you to upload every single image you want to edit onto the retarded cloud system to make you pay more. And the old one is extremely slow.

If you’re a bit technically minded, replacing Lightroom with RawTherapee is easy. It has most of what you need from Lightroom, but a lot of it is exposed in a more raw, math-y way. Instead of a slider labeled “magic bad stuff removal”, you’ll get 4 sliders that can do the same thing, but labeled something like “Delta decorrelation”, “Cyanic phase interlock amount”, “μ-rotation” and “Linear epsilon sync density”. (Note: Not real sliders) Which is just fine by me. It is faster, produces equally good images, has a fantastic blackframe calibration feature, and I also greatly prefer RawTherapee’s export menu and workflow.

I also don’t think it suffers from the gross Fuji worm-noise problem.

What RawTherapee doesn’t do is organization in the same way that Lightroom does. I really liked how Lightroom would import my pictures into a folder for each year, with folders for individual days inside that. Now, I just do that using ExifTool by Phil Harvey. The command looks like this:

exiftool -r -P -progress "-Directory<DateTimeOriginal" -d D:\Pictures\%Y\%Y-%m-%d\ \Input\Directory\

(Ugly backwards slashes, becuase I do this on Windows. It’s gross, I know.)

That’ll move the files from the \Input\Directory\, and organize them neatly on my harddrive. ExifTool is also way, way, way more powerful than just a batch organizer, but if you want to look deeper into that rabbit hole, go to their website.

Affinity Photo also has a raw editing module, like Photoshop’s Camera Raw. It has some things RawTherapee doesn’t, and vice versa. It’s notably lacking blackframe calibration/compensation.

Premiere Pro

Usually, all you need is the ability to assemble clips and export them. For this, Kdenlive is enough. They’ve made great strides in speed and stability recently. However, it is still missing fundamental editing tools, like ripple delete, and feels somewhat awkward when coming from a commercial editing suite. If you give it time, and adjust your workflow to its quirks, though, you will have a very capable editior on your hands.

If you’re a professional, maybe try Vegas Pro (I know. I feel weird saying it, but it seems to have gotten better, and you can buy it without a subscription) or DaVinci Resolve. Resolve is the gold standard of color grading software, and as of late it has gotten pretty good for general video editing too. Easily beating Premiere Pro in responsiveness, stability and intuitiveness. It also integrates well with the previously mentioned compositing software Fusion. They’ve even got live collaboration within projects that is almost Google Docs-esque, but local and private.

Currently, I see two problems with Resolve.

Its interface can’t be customized.
It uses an absolutely retarded “database” system instead of letting you save normal project files anywhere you’d like. This is not optional. You have to use this dumb database system. This database system seems to be integral to the collaboration feature, but if you don’t care about that, then you should be allowed to disable it.

Audition

Haha, really?

Adobe Fonts

In my opinion, this is the only good thing about Creative Cloud. If you ignore how difficult it is to locate the actual font files on your computer so you can use them outside of Adobe software, the idea of a “Spotify, but for fonts” is really nice. And it’s even better because the price doesn’t start at $2500 pr year. Fonts are expensive, and I’m always looking for new ones, so being able to pick and choose, knowing the licensing is all handled for me is great.

So, when you lose that, the only alternative is: Buy the fonts. Lets be honest. You use 4 of the 68 fonts you picked out from Adobe Fonts, anyway. The rest you grabbed off of DaFont. However, when you start looking for fonts, you’ll quickly find that the basic set of Futura PT costs 300€. The basic set of Neue Haas Grotesk Display costs 329€. And don’t even get me started on the paneuropean version of Neue Helvetica. I mean HOLY $HIT. And all of them have frankly garbage licensing terms that are unacceptably annoying.

Therefore, I will gradually compile a list here of font foundries with acceptable terms and pricing, sorted by how much I like them:

United States Graphics Company has extremely tasteful monospace and technical fonts with the best licensing terms I’ve ever seen. Massive respect.
Fontspring’s Worry-Free license gains massive respect from me as well, and covers some genuinely great fonts too!
HVNTER is so good I almost don’t want to tell you about him. Very simple and permissive licensing system.
Züli has very few, but quite good over-stylized fonts. License is extremely simple: “No Limitations on Commercial Use”.
Omar Sacca is a bit of a legend. He’s made things you’ve seen. Some very cool heavily stylized fonts here, with fantastic terms.
David Jonathan Ross has some good fonts, with simple predictable pricing, aside from the web licensing which requires you track monthly “unique” visitors (Which is annoying, unethical, probably illegal in the EU not to mention technically impossible)
Indian Type Foundry has a few good fonts, with ok licensing, but charging for webfonts based on pageviews.
Lost Type has a couple ok fonts, with good licensing terms.
Artemy Lebedev Studio has a couple of decent fonts, and good licensing terms.
Uncarving Nation has plenty of great stylized fonts. However while the licensing is simple, it is restrictive.

How it all goes horribly wrong

Sun, 24 Apr 2022 14:06:43 +0200

The “Stack”

I see kids in uni taking notes using Google Drive, or Docs or whatever they call it. They use Google Chrome on their Windows laptop. Google Drive runs using a ton of JavaScript, and their servers run on all sorts of other stuff. Including whatever they need to do magic AI analysis of everything you write so they can rent out their model of your mind for targeted advertising.

Microsoft Windows: At least 45 million lines of code (Probably way more by now)
Google Chrome (Chromium): 25 million lines of code, spread across 36 programming languages
Machine learning probably done in Python: 1 million lines of code
Google’s monolithic repository of stuff is about 2 billion lines of code
Server OS, likely Linux: 12 million lines of code

This is a really small slice of all the stuff that goes into the software “stack” involved in typing stuff into Google Drive.

Google’s monolithic repository is about 85000 gigabytes of code. They employ 25000 engineers to work on and maintain it. They use 12.4 terawatt-hours of electricity pr year, which is more than most of the world’s countries use in a year.

And from all of this, you get the amazing ability to…

Write notes. Slowly.

In order to write some notes, that you won’t even open again after having written, you utilize all that. And you think you’re making things simpler by “Just using Google Drive”. Because the interface you see is white, and has rounded buttons, you think it’s simple. When in-fact, you are making things about as complicated as they could possibly be. And what you get out of it is still laggier than editing a text file in the 1980s.

Mo’ lines, mo’ problems

Throughout all this, the bugs will creep in.
More lines, more bugs.
More lines, more complexity.

Already, people have died in plane crashes caused by software bugs.

“Oh, but the kids nowadays are so good with technology, and they grow up with it, they’ll end up working on it, and fixing it”

No, the kids nowadays are good at using iPads, Snapchat and TikTok. If they ever learn to code, they learn to code in some bullshit language like python, C#, or very modern C++ with .NET, or even worse, whatever Duplo shit Unreal Engine comes with now. They will make their websites using Squarespace. And they’ll write their notes on Google Drive. And they will have absolutely no idea how any of it works. All while having convinced themselves that they’re doing it the right way, the simple way.

And then, when all the gray beards that created, maintained and understood all the underlying structures die off, and their wisdom is lost, we’re fucked.

We’re royally fucked.

You already struggle to stay on top of your projects when they cross 2000 lines of code. Even the professionals at Microsoft are absolutely fucking incompetent at writing software. And every day, we add more code. More layers. More “solutions”, to problems created by complexity. Oh, your server software is buggy and crashes all the time, because nobody knows how it works, or how to fix it? Here’s the solution: A program called Docker, that lets you run every program in a viritual machine, and just automatically reboot the whole thing when it crashes. We are 20 billion band-aid “solution” lines of code deep into this problem, and it shows no sign of stopping.

Not to mention that massive amounts of this code does stuff we don’t even want. Advertising, cookie popups, tracking, fingerprinting, update nags, DRM, adware, spyware, malware, content ID, social media post priority.
And so we develop “solutions” to those problems. Ad-blockers, cookie blockers, tracking and fingerprinting resistant browsers, pirate sites, browser extensions that give back control and customization on social media sites.

Too much code? Too much bad code? Solution: More code!

Is there an actual solution?

I have no clue. Refuse to write and use shit software, I guess.

The Joe Rogan Experience is not a podcast

Thu, 17 Jun 2021 00:00:00 +0000

It used to be, but it isn’t anymore.
I’ve seen many people talk about the Joe Rogan & Spotify deal as a “gREaT sTRiDe fOR POdcAsTIng!”, but it is exactly the opposite.

What is podcasting?

Podcasting is a standard of publication. You publish a collection of audio files via a public RSS (Or Atom) feed. That feed is the “cast”. Anyone can then read that RSS feed using an application like Podcast Addict, or any other podcatcher. It podcatches the podcast.

This is why podcasting is such a symbol of freedom. Anyone can publish whatever they want to their RSS feed, and anyone can catch that feed. Nobody needs to know what you listen to, or who is downloading their podcast. Nobody can stop the creator, nobody can stop the consumer. There is no middleman. There is no youtube.com that forces you to make an account to watch certain content. There is no 3rd party of underpaid Indian contractors sitting around deciding what content is acceptable and not based on poorly defined guidelines. There is no low-level staff that can complain about what episodes their service is hosting.

What has the JRE become?

It has become exactly what I described there. The JRE is now exclusive to Spotify. Spotify is hosting the content and intentionally only making it available in a format that doesn’t work with normal podcatchers. This is of course, so you have to use Spotify to listen to their exclusive podcasts. There is no public RSS feed, now that Spotify has shut down anon-podcast.scdn.co. Meaning, there is no cast. There is no podcast. It can’t be podcatched.

The opposite of a great stride

Using Spotify requires an account, not just to “"“prove””" you’re over 18, like with YouTube, but to listen to anything at all. Many episodes of the JRE are still missing after migrating to Spotify. Mysteriously, mostly the ones that some people at Spotify don’t like. Especially the ones featuring Alex Jones, a man who’s actual podcast was removed from Spotify.

You can’t remove a podcast from a podcatcher. That’s not how podcatchers work. However, Spotify isn’t a podcatcher. You can’t input an RSS feed into any Spotify client. The creators have to submit an application for Spotify to feature your podcast in their catalog. So whenever Spotify says anything about having podcasts, they’re lying right in front of your fucking face.

Worse than this, is that Spotify is going to heavily integrate tracking and advertising into their system. Which will be very targeted advertising. Much more targeted than the normal ad-read just being related to the podcast’s topic, as is the status quo for podcasts. This targeted advertising will be way more profitable and effective, because that’s the point of targeted and privacy invasive advertising. This means big money will be pushing long-form audio shows like the JRE to stop podcasting and start “Spot”casting (Eh? Like it? I came up with that myself. I know, give me the Nobel Prize in Literature.), so we can have even more privacy invasive and targeted ads. This will lead web audio shows in a direction far away from podcasting’s freedom, and right into the hands of a few large corporations. Again.

The future of online audio content still has the chance to remain free, or you can willingly let it slip into the control of a few corporations, and become dependent. This process has happened many times before, and it is about to happen again. This time, will you be doing anything to stop it? Or will you keep being a good little obedient and unthinking consoooomer?

Update 2022-02-05:

As of writing this, another 71 episodes of the JRE have been removed from Spotify, in addition to the ones they never published to begin with, totalling 113 episodes. For the full list of vaporized episodes, you can go to JRE Missing. Once again, nothing is sacred when you give up control of your content to random centralized corporations. All these episodes would still be up if he was actually podcasting. Maybe the video versions would’ve been removed from YouTube, but you could still find them in you favorite podcatcher. And then Spotify would still have some incentive to keep them up to compete with YouTube and actual podcatchers. These corporations don’t deserve you. They don’t deserve Joe. Don’t use Spotify, use actual podcasting and podcatchers.

How to anti-alias the threshold effect in images

Fri, 14 May 2021 00:00:00 +0000

I was looking into generating black-white images that were anti-aliased, but all I found were ways to definitely not do anti-aliased thresholding.

Among these non-solutions was an article that suggested using curves instead of threshold to just increase contrast a lot, and pretended that they had somehow created anti-aliased thresholding. What they actually made was a black and white photo with a lot of contrast, not even remotely similar to thresholding. And if you were to push it far enough to have the same contrast level as threshold would have, you would still end up with the same jagged edges and noise splotches as normal thresholding. People on forums were saying to blur it, which of course looks terrible, because you end up with all the same issues, but this time they glow.

Turns out the actual solution is pretty simple. You need to super-sample.

Scale up the image 8x or 16x
Apply the threshold effect.
Scale the image back down to 1/8th or 1/16th the size.

As you can guess from how simple this is, it can be done in GIMP, Photoshop, Krita, really anything. If you are using Photoshop, I suggest using Preserve Details 2.0 when upscaling, so you’re maintaining as of the shape as possible. Maybe you could even use waifu2x.

You can do it using G’MIC, too, which would look something like this:

gmic input_image.jpg resize 1600%,1600%,1,3,3 adjust_colors 0,0,0,0,-100 threshold 127 normalize 0,255 resize 6.25%,6.25%,1,3,3 output output_image.jpg

Here’s a comparison of the different methods:

Standard threshold

Blur

8x supersampling using waifu2x

16x supersampling

Source image

How to set up Windows 10 almost without killing yourself

Thu, 11 Mar 2021 00:00:00 +0000

Windows 7 was amazing. I used to think “7 was way better than 8, but 10 is also ok.” However Microsoft’s behavior has increasingly been becoming really fucking disgusting. It’s so disgusting in-fact, that I’m having a hard time telling people to not buy a Mac.

Back in the Windows 7 days, you turned on your computer for the first time, set the language, keyboard layout and chose a username. After that, you’re good to install a better browser. Maybe you’d have to uninstall McAfee, but that’s about it.

Now, with Windows 10, the list of shit you have to do to set up a normal, non-surveilled, non-bloated, ad-free experience is so long it’s honestly idiotic. When I first saw the Microsoft Store wanting you to install “apps” (ew) to Windows I was already scared, but I didn’t think it would get this bad.

Not including normal things like picking language and keyboard layout, the list is:

Say no to expanded telemetry
Say no to location tracking
Say no to device tracking
Say no to both ad-tracking options
Say no to “online speech recognition”
Probably a couple other things I’m forgetting
Find a way to block internet access, so they can’t force you to make a Microsoft account instead of a normal account
Uninstall Office 365
Uninstall OneDrive
Uninstall Dropbox-MS-Store-Edition
Uninstall Spotify-MS-Store-Edition
Uninstall McAfee
Uninstall the games they pre-install for some reason
Uninstall Microsoft Mail
Download an actual decent browser to replace the executable turd known as Edge
If you downloaded Chrome, you have to click past the “Please don’t upgrade!!” screen
Find out you can’t install anything at all because of something called “S-Mode”

The list goes on, but first, we need to talk about “S-Mode”. This is absolutely disgusting, unethical, unacceptable behavior. In order to install anything using normal installers and avoid using the Microsoft Store or making a Microsoft account, you have to download an unlocker from the Microsoft Store, which requires a Microsoft account. A catch 22 that means no matter what, Microsoft gets to register your computer up against your Microsoft account. If anyone figures out a way of getting past this without buying Windows 10 Professional and overwriting the OEM OS with it, let me know.

Buy it on eBay, btw. It’s only a couple of bucks, and I’ve yet to get scammed after buying at least 5 licenses. Even if you have to try 3 sellers, you’re still saving hundreds of dollars.

Now, continuing the list.

Re-do everything you just did, because you’re overwriting the OS with a fresh version of Windows 10 Pro.
Finally install a better browser, like Brave, Chromium, Vivaldi or Firefox
Remove the Edge icons from everywhere
Remove the other trash software icons in the start menu
Go to settings and disable everything, including location access, camera access, everything-access for anything that doesn’t need it
Download whatever the latest sketchy telemetry-disabler tool is

These tools let you actually disable things like telemetry, tracking, calling-home, ads and Cortana. They change all the time, and they’re super sketchy. It used to be Anti-Beacon, but that’s paid software now. None seem open source, and when you search for them, all you get is SEO blog-sites that list off 10 of them, leaving you thinking all of them are viruses, or possibly worse than just letting Microsoft track you in the first place. Nevertheless, if you find a good one (Maybe ShutUp10 is good? Or maybe Windows10Debloater is better?) it’ll help you get rid of a lot of telemetry and other bullshit. I still can’t fucking believe they put ads and ad-profiling in the actual operating system.

Back to setting things up.

Either use registry editing to give you back the normal Windows Photo Viewer that doesn’t blur images when you zoom in, or install qView
While you’re in RegEdit, might as well set JPEGImportQuality to 100, so Windows won’t ruin your wallpaper quality
Install Everything Search by Void Tools, because Windows search is broken
Install AltSnap, because Windows still doesn’t doesn’t let you conveniently manage windows
Hide all the unnecessary search buttons on the bottom bar, because they don’t do anything you can’t do by just hitting the Windows key and typing what you want to search for
Install an actually decent media-player like MPC or VLC

And after spending so much of their resources making the experience markedly worse, you have to spend a lot of your time and brainpower undoing their hard work. Yet after all this, Windows is still far behind MacOS in a lot of places. Windows still doesn’t have a proper terminal, aside from an option of installing a really slow uncustomizable Ubuntu terminal through the Windows Store. Windows still has shitty audio, and no virtual aggregate sound card functionality, or low-latency audio option without using third-party drivers like ASIO. No ability to let multiple programs use a media source like a webcam at once. And apparently, they’re getting rid of the control panel and completely replacing it with the horrible settings menu, and rounding the corners of the windows?

At this point, I can’t recommend getting a Windows machine over a Mac unless you’re building it yourself, and will be installing Windows 10 Pro. Especially with Apple’s frankly fantastic track record on privacy and fight for encryption. The best option now is just to buy a laptop purely on build quality, and installing some flavor of Linux. Well, that has always been the best option, but it has never been the easiest option. Not that it has gotten much easier, just that installing Windows 10 properly is so damn difficult now. Seriously. Jump ship as fast as you can.