There are rain filters available for the editing program I use, Adobe After Effects, but they cost money. I decided to write a computer program myself to generate rain footage. This took a few days to get right. Multiple "sheets" of rain, each at a different distance and with its raindrops falling at a different speed, are superimposed over a transparent backdrop so that the mix - a PNG sequence - can be laid over footage without needing any (harmful) chroma-keying. The result is pretty good.
Now the lightning. AfterEffects has a lightning effect, but I wanted as much of the channel to be my own work as possible, so again I wrote a custom program. This wasn't as successful as the rain; I should have spent longer on it. However, the result is quite simple and endearing, as in a certain 1980s cartoon, so it's not all bad.
The thunder sounds were most fun to do. Of course, real thunder audio can be easily obtained on the Internet but that didn't seem right. Again, I wanted it to be my own work. I remembered that I had read somewhere that a good way to fake thunder was to record the sound of paper being crumpled and then greatly lower the pitch. After some experimentation, I found that -35 semitones gave a good result. This is how most of the thunder heard on the channel was created. A small amount was also created by rolling a wheelie-bin around and, again, down-pitching it.
Lightning sounds were made by a variety of means. I set up a few VST synths in Fruity Loops to emit bursts of noise, then manipulated the result in Sound Forge. I also used a range of sources, including early '90s hip-hop songs, as random audio "stuff" that was massively distorted, speeded up, re-pitched, chorused, etc.
The sound of rainfall was achieved by simply creating fuzz (see above) and removing everything but the highest frequencies.
Lastly, wind sounds. Ideally these would have been made with a hardware synthesiser, me twiddling the filter cut-off over a white noise generator. However, neither of my hardware synths (the Ensoniq Fizmo and the Elektron SidStation) could generate pure white noise, so I had to use software synths (such as Krakli's Gargoyle VST) to generate noise that went through automated LFOs. I made the sounds more erratic by manually pitch-bending them in Sound Forge.
The opening sequence originally started with an animation of the 9/11 attacks - the bolt of lightning which now opens each video was originally a "response" to the plane colliding with the North Tower. I removed this prelude because it felt disrespectful to use the atrocity in this way. The rest of the sequence stayed as planned: the channel title "wipes" over the screen like the titles in a 1930s horror film, then another thunderbolt heralds the episode title. Fade into the video, and the sound of rolling thunder gradually dies down over the next thirty seconds.
The closing sequence was intended as a gentle way to let the viewer detach from whatever had been said during the video, and perhaps contemplate it on their own. For this reason it is much longer than necessary - 25 seconds. There is also a 10-second lead-in, with wind sounds slowly rising, signalling to the viewer that the video is about to finish. During the closing sequence, several contact details are shown.
Both the intro and outro have been modified here and there along the way, but I will not detail these changes as they are so trivial that I don't remember them any more.
In June 2014, with the centenary of WW1 looming, I decided to make a series about it. This was an opportunity to use my Fizmo on the channel for the first time. It was also the first time that I wrote a program to automatically generate an animation for each episode in a video series - in this case the title card for each episode, rendered in an "old typewriter" font and being typed up on-screen, complete with sound effect. The syncing of the sound effect was particularly onerous to code, because I had not yet thought of employing Chrome to do these sorts of things (see 2017).
In August, the channel reached 1000 subscribers. I made a celebratory video. In this video, a vocoder appeared for the first time on the channel - a very brief vocalisation put through the Fizmo's on-board vocoder.
In October, I began a new series: "reply" videos. This required custom intro and outro sequences, and in these I introduced the 1920s German Expressionist aesthetic which has since become a mainstay of the channel. It is achieved by completely desaturating the image, increasing the contrast, and then overlaying solid blue with the "Hue" blending mode, with solid white overlaid with the "Overlay" blend mode. Then lumpy film grain is overlaid on top of that. Later I began using a similar treatment on the main footage of these videos - a solid of RGB(0,108,255) overlaid using "Soft Light" blend mode.
The original recording set-up (the FZ200 camera on its own!) was maintained for 10 months, until I got a microphone at the end of October: the Blue Yeti. This hugely improved the audio quality of my videos.
neoreactionary testosterone (Nov 20) (PB animation of blood shards)
Also in November, I began another new series: Skype recordings. This again required custom intro and outro sequences, done in the same style as the reply videos. But the Skype Recordings are interesting for a different reason. Since I expected to release a lot of Skype recordings, I decided to quicken the production process by automating it, to a great extent. I wrote a fairly elaborate software processor which takes the finessed audio recording and builds a video around it, complete with the pre-filmed intro, a custom title card, a looped animation throughout the conversation, and a pre-filmed outro. It also automatically creates the video thumbnail.
Finally, in December, I bought some LittleBits modules. These are essentially synth blocks that can be joined together in various ways to create any number of different custom configurations to generate sound. One module is a white noise generator and another is a resonant low-pass filter. These two, along with power and a speaker module, are enough to create wind sound effects, which was my main desire. However, I also invested in the Arduino module (which can be programmed) and used it to write a custom random LFO to modulate the filter independently.
In September, I decided to begin a new series of deliberately introspective monologues, called the Monday Moan. Like the Skype recordings, these would be audio recordings processed into videos automatically by a new piece of software which was derived from the SR processor.
In November, I began a new format for long videos: the multi-parter. This required the creation of "book-end" sequences for the start and end of intermediate segments. For these I created a new, incredibly "whooshy" wind sound effect using a second-hand hardware synth I had just bought on eBay, a Korg EX800. (Unfortunately this synth, being 32 years old, broke only a fortnight after arriving!)
Later in November, my channel reached 5,000 subscribers. I felt it was appropriate to make a celebratory video. For this, software was written to extract subscriber data from YouTube and generate an animation sequence to precisely reflect the growth trend visually.
In February I began planning a new aesthetic for the channel. It was going to be very ambitious, involving a model set, slow-motion filming, etc. Eventually I realised I simply had too much other work to do, that was far more pressing, and these new title sequences (which would have created a unified look across all the strands of the channel) was an indulgence. However, while planning it, I bought a new camera, the Sony RX100. I chose it because it could do slow-motion filming, but soon realised that its image quality was simply incredible, so decided to use it for all filming (from June onwards).
In April the channel reached 10,000 subscribers, and it was time for another celebratory video. This one was more technically elaborate than the 5,000 video, featuring a flickering line that crossed the screen, showing the subscriber growth in a particularly beautiful way.
Nothing else of note happened until September, when I created a new intro for the regular videos, this time involving some basic 3D animation (done with the antiquated Daz Bryce) as well as my usual tools. However, for some reason I didn't start using this new intro until 6 months later, in February 2017.
In October I prepared for my trip to America - the Dangerous Haggis Tour. Since this was my first real visit to America, I was determined to make as many videos as possible, documenting my exploration. Obviously I was going to need a portable device on which to create videos, but I didn't have a laptop. I should probably have taken the leap and spent enough for a decent laptop, but, afraid of splurging cash, I instead chose a cheap little tablet. It was very portable and very cute, but would be unable to run Adobe After Effects. I decided to "go lo-fi" and make videos in a semi-automated way. A program was written to combine all the footage files in a specified folder, generate a custom title sequence, and combine these elements into a finished video. This software took a few weeks to write (and was improved in little ways throughout my time in America) but worked very effectively with minimal computing power.
For the Dangerous Haggis Tour, I also invested in a new microphone: the Zoom H2N. Originally intending it only for outdoor recording, I now use it for everything.
Once settled abroad, and still feeling quite depressed and listless, I treated myself to a new hardware synth: the Studiologic Sledge. It is a strange beast, digitally emulating analogue yet never sounding anything other than 1980s digital. It works along analogue lines, so is very easy to work with, but always brings that delightful 1980s texture to everything. It is much more straightforward than the Fizmo, but, I confess, less interesting - probably because it is so straightforward.
For some time, I had wanted to improve the code for generating lightning bolts. Each frame was being drawn randomly, and was therefore completely different from the one before it. Thus, there was no "flow", as there should be with electricity. I took some time to write new code which works in an entirely different way. It involves "pulses" that travel along a straight line, at different speeds and with different impact levels. In each frame, at every point along the line, the cumulative impact of all present pulses dictates the divergence from the centre. This results in a beautiful evolving lightning bolt which can be rendered at any speed and for any duration.
Ever since my first speech in Rotterdam in September 2016, I had wanted to have a special intro for the speech videos. I came up with the idea of showing the country (where the speech took place) being "drawn" by a lightning bolt. This would be simple to code, if I had the outline of the country. So I wrote code for discerning the outline of a solid shape, as a long sequence of coordinates. This was very tedious to code, and to be honest the program still doesn't work perfectly and sometimes needs manual "help" - but it works. I combined this with the evolving lightning code and the Chrome sound manipulation code to produce what is now seen at the start of every speech video, from Oslo onwards.
Something that had always aggravated me about filming videos, whether with the FZ200 or the RX100, was the 30-minute recording limit on cameras. It meant I could never fully relax, always having to be aware that the camera might stop recording. This is especially bad with the RX100 because it doesn't even bleep when it stops, and several times I have kept talking for 5 or so minutes before realising. Even without these mishaps, it seemed a silly limitation to be tolerating, given that making videos is my job. Thus, I decided to look around for a device that would circumvent this limit. I found the Hauppauge Rocket. At first I was delighted with it, but soon realised that the Rocket brought with it technical problems of its own which complicate my work. I will replace it with something better in due course.
In July I treated myself to a proper vocoder - the Electro-Harmonix v256. I haven't really used it yet for anything, but doubtless it will be useful in future.
Finally, I wrote a little program to automate the final stage of making multi-part videos. Each part is rendered out as an mp4, and this program stitches them together. It also counts up the duration and outputs a list of where each part will begin in the finished video - a list I then paste into the video's description on YouTube.