Download on Patreon
Why are cyberpunk HUD overlays so sexy? Pure tech essence satisfaction. I've been collecting inspiration for the last few years and since I had so many ideas ready, I licensed some vector artworks and then did some collaging so that I could focus on the animation process.

I wanted to make it feel like the GUI was loading and snapping into place and so I wondered if I could use the AutoFill 2 plugin in After Effects for the "Pattern" scenes. I did some tests and knew instantly it was a sandbox goldmine. I did a bunch of experiments to refine the technique and realized that the scenes with chunky block shapes needed a different approach compared to the scenes with thin delicate lines. So I used the Shape Based Flow mode (Speed: 5000) for the "Thin" scenes and the Slower Near Borders mode (Speed: 1250) for the "Wide" scenes. Also using a Border Strength of 10% was an important attribute to allow the visuals to flood quickly. From there I would apply the effect to each scene, lay down the Growth Points, and tweak until satisfied. From there I added the Tracery 2 plugin to get some connection visualizations automated. I love the look of these when they animate rapidly, but know that VJs will need to slow them down and so I made these scenes to be 120fps.

I had a bunch of ideas I wanted to explore using just basic keyframe animation in AE for the "Elements" scenes. A few of these scenes utilized the Repeater plugin to make some mirror grids. In the process of creating the animations, I had a happy accident where I masked an image against itself and it looked really interesting when I animated the position of the matte. This ended up being a very rich technique and so I did a bunch more in this style to better explore it. Then to make it pop even more, I used the Tint FX to have the base layer be red for when the mask was animating and then have it instantly switch to white for then the last position keyframe was reached. Then I wanted to stagger the paired layers (alpha track matte and base art) so that it visually felt like the GUI elements were triggering at different times. But because each comp had 100+ layers, it was tedious to apply manually and it was slowing down my creative process. So I fired up ChatGPT and created a series of JSX scripts which would automate this task: ISO_TrackMatteMulti.jsx, ISO_SelectAlternatingLayers.jsx, ISO_StaggerSelectedLayersInPairs.jsx. Also ISO_ScriptLauncherHotkeys.jsx was useful for quickly pipelining these scripts via different hotkeys. I also made these scenes to be 120fps so as to give VJs lots of flexibility when performing.

Another idea I had was to have a bunch of different types of crosshairs acting like they are targeting. I used the AE masking technique described above to give them some more life. I also did another variation where I used the Ripple FX to give it a different vibe. From there I used ChatGPT to create ISO_RandomizePositionWithinRange.jsx so that it could auto randomize their positions within a given range of values. Then staggered the layers.

I've always loved the look of the render buckets while V-Ray is rendering an image. So I used this as inspiration for creating my own version for the "Render Buckets" scenes in AE. This also paired with another idea that I've had for a long time and was excited to try out. Basically I wanted to utilize both the alpha channel and the black background so that when a VJ played back this clip, it would look as if they had render buckets processing the video underneath. Therefore I needed to create the render bucket GUI, black background matte, and alpha background matte so that these three things could work in tandem to pull off this illusion. This ended up being much tricker than I expected, mainly due to the scope. First I designed my own render bucket GUI while considering how it would animate to reveal the content underneath, and then laid it out into a grid of 512 buckets (x32 buckets wide by x16 buckets tall). The first difficulty was animating the buckets into different patterns. After a bunch of trial and error, I realized that the order of the layers could be used as a way to create and animate the patterns. Then I manually laid out the patterns in Photoshop and then imported the PSD into After Effects. But then another difficulty was that I needed to swap out each PSD layer for a comp that contained the render bucket animation. So I used ChatGPT to create ISO_PasteClipboardAndReplaceSelectedLayers.jsx so that I could automate it. Now I had the render bucket animating in the desired patterns, but the final difficulty was figuring out how to make the background go from black to transparent according to each render bucket. So I added a green background behind each of the 512 buckets, but each layer only became active when its related bucket was visible. Then applied a chromakey FX to remove all of the green colors from the whole comp and then added Spill Suppressor FX to remove any green pixel aliasing leftover. It’s bizarre to render out render buckets. Does that make this a meta render?

I have always wanted to create my own "Console Log" scene featuring scifi flavored technical text. But animating the scrolling text and terminal cursor in AE is not a fun process. So I was brainstorming some different techniques and I was just about to give up when I realized that I could instead ask ChatGPT to make a python script that would take a text file as input and use it to generate a fake session from within a real terminal app and then I could simply record the screen. Within the python script, I asked it to include features such as line delay, line burst size/chance, type speed, and differentiating when to do full line printing versus character-by-character printing. I customized the font size of the terminal app and the python script worked on the first try. Then I needed to create a bunch of surreal technical text. I started up a new instance of ChatGPT and asked it to assist me in generating technical log text that involved surveilling the audience of the VJ. I did some tongue-in-cheek variations on this theme, edited the text, and then did a screen capture of the Terminal window. Chefs kiss.

While rendering out all of these scenes from AE, I ran into an issue that I've experienced a few times in the past. Initially I was relying on the H264 codec, 3840x2160 at 120fps, via the AfterCodecs plugin since it automates the bitrate by allowing me to simply set the quality and speed attributes. But due to the high resolution and very high frame rate, for some reason this introduces undesirable artifacts when the video is played back. I can scrub directly to the frame in After Effects and it'll look perfect, but playing back the video in VLC results in artifacts that are reproducible. I think it might be that the video cannot be decoded fast enough due to the 120fps and so it glitches out. So instead I decided to render using the DXV-alpha codec since it's an intra-frame codec and therefore won't have this issue.

This reminded me of a separate finding where the DXV-alpha codec creates video clips which are roughly x2 smaller than the DXV codec. So I ran a series of tests comparing the DXV and DXV-alpha codecs. Interestingly the DXV-alpha always resulted in smaller files, sometimes up to x2 smaller. This is unexpected since the DXC codec contains RGB channels and the DXV-alpha codec contains RGBA channels, and so it's strange to me that adding an extra channel results in a smaller file size. The HAP-alpha codec creates video clips which are roughly x2 larger than the HAP codec. But the DXV codec is closed source and the developers aren't willing to spill the beans when I inquired. So def a welcome finding, but strange! Anyways, I ended up rendering out everything using the DXV-alpha codec. I was worried that I was going to end up with some super huge file sizes due to working at 3840x2160 at 120fps with alpha, but interestingly the solid color designs apparently result in small file sizes. Error: task failed successfully!

Released August 2026