Pressure Project 2 – The Flipper
Posted: March 3, 2026 Filed under: Uncategorized Leave a comment »Description
The Flipper is a TouchDesigner patch that uses an audio input to create video, and uses a video input to create audio. When used in a network this “cell-block” acts independently by creating entirely new audio and video, instead of just modifying the audio and video it receives. Its modularity lies in its ability to provide other users in the cell-block network with new sources of audio and video, that are themselves generated from other audio and video over the network.
Collective Documentation
Pending
Individual Documentation
Overview of my cell-block’s network. This is connected to three inputs and outputs on the outside of the container, which connect to other cell-blocks on the network. While there’s a lot on screen, it breaks down into a few simple sections.
This portion of the network takes in audio from over the patch through the in_audio CHOP. The envelope, math, and audioparaeq objects slow the stream of data and boost high frequencies, respectively. This then is turned into a spectrogram and is sent directly to a chopto TOP.
This portion processes that audio spectrum into a new visual. Starting in the bottom left, I use a series of TOPs to create a flow-like visual, which is then composited with the spectrum. This new visual is colored using a series of ramps and a look up TOP. The ramps are cycled through using either an LFO or an input from in_osc over the network. An example of the visuals this produces is below.
Lastly, this portion of the patch processes video received over the network from the in_video TOP (or in this case, a camera input) into audio. While I didn’t get quite as interesting an audio output as I wanted, I still think I was effective in transforming video to audio. The video that is received gets sent directly to a topto CHOP, which reads RGB values over the X and Y planes of the video. the following objects then reduce the amount of data, and turns those waves into a stereo audio signal by the merge CHOP. This wave is given an envelope by the math objects (I attempted to control this with another osc input but failed to make it work) and is sent out over the network. An example of audio is included below.
Reflection
Since I knew I wanted to flip the audio and video signals inside my patch, the independence of the cell-block was semi-inherent the entire time I was working on it. In order to ensure it was connectable with others, however, I needed to ensure that whatever the patch did was interesting enough, while still clearly using audio and video to influence the opposite output, so that it didn’t just seem like I was generating something entirely new.
I made choices about what to include and exclude primarily by trying to figure out what I could accomplish that was reasonably within my ability, but still interesting. For example, I’ve worked with spectrogram imagery in the past, so I knew I would be able to incorporate that the easiest. On the opposite end of that, I attempted to integrate FM synthesis into the audio part of my patch to get some more interesting sounds. However, with my inexperience in TouchDesigner, I found it really difficult to make FM work, so I chose to exclude it.
One thing that surprised me was how even if cell-blocks didn’t work “perfectly” together, they still were able to have some sort of interaction, even having unexplainable interactions. I was also a bit surprised how underutilized the OSC data we were sending was. I know I was personally having difficulty in doing something interesting with the OSC signals, but it was interesting that it was a widespread problem. I think this might come from the fact that the other signals we were working with were both very tangible. Since the OSC input was just a number, I think we were a bit less motivated to find an interesting way to use it, as opposed to the audio and video, which we could immediately do interesting things with.
I think we didn’t have quite enough time to experiment with combining our cell blocks in different ways for a lot of emergent behavior to emerge. But one that I enjoyed seeing was how the visuals would layer together through 2, 3 or 4 cell-blocks. I thought that all of the cell blocks were interesting on their own, but the most interesting visuals were created through the combination of several together. This relates to Halprin’s cell-block framework through the idea that we can each create our own module that does its own thing, but the most exciting behaviors only emerge once we begin to combine the different cell-blocks, and experiment with how they feed into each other.
Download Patch
Pressure Project 1: The Musical Spiral
Posted: February 13, 2026 Filed under: Uncategorized Leave a comment »Description
The musical spiral is a self-generating patch that randomly generates shapes at different sizes and positions, and spins them in a random direction for a random-length cycle. When these shapes cross a line, they (are supposed) to trigger a random musical note.
Documentation
Before starting to code my patch, I did a quick sketch for my idea of what I generally wanted the patch to do to help me save time later. While I had to change and add a bit outside of this, this essentially became the outline of what my code would look like.
The overview of my patch, upon entering the scene, the random numbers for duration of the cycle and the direction of the spin are generated, since they’ll be applied to all of the shapes. When the cycle ends, the spinning shape user actor sends a trigger to the Jump++ actor, going to a duplicate scene, which jumps back to the first scene.
Inside my “spinning shape” actor, the final result of my original user actor sketch. The bottom 2/3rds of the screen contains the actors randomizing the attributes of the shapes actor. The top 3rd deals with spinning the shape clockwise or counter-clockwise (decided by the “flip coin router” user actor) for a cycle of random length with a random delay from shape-to-shape.
Inside my “hitbox trigger” user actor. This actor takes each shape (which has been sent to its own virtual stage) and looks for when it makes contact with a small white rectangle I sent to every virtual stage in the “hitboxes” user actor. When it makes contact, it was supposed to trigger a random sound in the “sound player” actor.
Random selection of 18 short samples of single notes. Chromatic from C3-F4.
How I checked if one of the spinning shapes was inside the same area as the hitbox, sending a trigger when they “made contact.”
Sound playback user actor.
A sample of how the final version of the patch behaved. The white line (actually smaller than the hitbox) was left on screen to provide a reference for when the sound was supposed to trigger (despite it not working that way due to the high load of the patch).
Reflection
One of the best ways I managed the 5-hour time constraint was to make the sketch of my idea as seen earlier in this post. By working backwards from my initial idea to solve the problem the best I could on paper, I gave myself a framework to easily build off of later when problems or changing ideas arose. It also meant that I had a general idea of all the different parts of the patch I would need to build before I actually started working on it. This also guided what I would include/exclude in the patch.
While my patch didn’t end up working the way I wanted it to (sounds were supposed to trigger immediately when the shapes crossed the line, unlike what is seen in the above video) I was very surprised how this didn’t “ruin” the experience, and how it even created a more interesting one. With the collision of the shapes and the white line being decorrelated from the sounds, the class became seemingly became more curious about what was actually going on, especially when the sounds would appear to trigger with the collision after all. I was also interested to see the ways people “bootstrapped” meaning on to this patch. For example, Chad had noticed that in one of the scenes, the shapes were arranged in a question mark sort of shape, leading him to ask about the “meaning” of the arrangement and properties of the shapes, despite them being entirely random.
During the performance of the patch, I unlocked the three achievements concerning holding the class’s attention for 30 seconds. I did not make someone laugh, or make a noise of some sort, as I think the more “abstract” nature of my patch seemed to focus the room once it started.
AI Expert Audit – The Elder Scrolls
Posted: February 5, 2026 Filed under: Uncategorized Leave a comment »Source Material
Since for my topic, I chose to pick a game which already has a wealth of in-universe literature written, my primary source was a pdf of every book that exists in the series, found at LibraryofCodexes.com. I also chose to upload a small document I found giving a general timeline of the series and its history, as well as a short video covering the history of the world.
I chose this topic as over the course of the last 10 years, I’ve likely played up to (or over) 1000 hours of these games over 3 different games. Even more so, I’ve listened to countless hours of videos doing deep-dives on the world’s lore as background videos while working or driving. I think the reason I find myself so drawn to it is the relationship between world-building and experience in RPGs. As I learn more about the world, the characters I play can have more thought out backgrounds and motivations, improving my experience, which makes me want to learn more about the universe. I was also interested to see how the AI would handle sources not about the game itself, but rather about a range of topics that exist *inside* the game.
The AI-Generated Materials
Podcast
Prompt: Cover a broad history, honing focus on the conflict between men and elves
Infographic
Prompt: Make an infographic about the Oblivion Crisis and how the High Elves capitalized on it.
Mind Map
Audit
1. Accuracy Check
Overall, the AI got a lot right about the historical origins and monumental events in the world of the game. There are some topics that are somewhat confusing that I was surprised it got mostly right. It didn’t get much wrong, but it did make a few strange or even incorrect over-generalizations. For example, in the podcast it said that the difference between the two types of “gods” in this world is “the core takeaway for how magic works”, which it is not. Even weirder, it got the actual origin of magic in the games correct later on.
2. Usefulness for Learning
I do think that these sources would be incredibly useful for someone with no prior knowledge of this series to easily learn about the world they exist in. The podcast does a good job at simplifying the most important events for understanding what’s happening and the motivations of different factions. However there are a lot of nuanced ideas that it completely misses, which could be due to the length being set to normal. The mind map does a really good job at connecting important ideas of the universe together. However, it also places too much importance on certain topics, such as a handful of weapons, only one of which has any real importance to the larger plot. Lastly I thought that the infographic did a nice job at laying out the events that I prompted it to, but there were a few spelling errors.
3. Aesthetics of AI
One of the strangest things I encountered doing this was the ways that the AI would try to make itself sound more human during the podcast. For instance, it would stutter, become exasperated at certain abstract topics, and even make references to memes not found in the sources. The AI definitely has a certain voice to it. I don’t know how to exactly describe it, but in the podcast it seems to talk like everything it mentions is the most important thing ever, and the other AI “voice” always seems to be surprised at what the other one is saying. I actually thought that the AI did a pretty good job at emphasizing the same things a human expert would. However it somewhat glosses over the actions of the player characters during the games, which I think a person would focus a bit more on.
4. Trust and Limitations
From this, I would probably warn a person against trusting the importance the AI might place on certain topics as well as the connections it makes between topics in generated educational materials. It also seems to avoid any sort of speculative ideas whatsoever, which I found odd since there were books in the sources which do theorize on certain unknown events or topics. I’d say the AI seems the most reliable in taking the information you give it and organizing it into easily consumable chunks. However, this only seems to be at a surface level, and when it tries to draw conclusions about topics, it tends to fall flat or make incorrect assumptions. I think in this case, you’d be better off just watching a video someone has already made on the games.