▩▩▩▩▩▩▩ WEEK 8 ▩▩▩▩▩▩▩

Reframing my Proposal

After submitting my RPO for formative feedback, I revisited it and found myself lost. Firstly, my introduction seemed lackluster and failed to effectively convey my research. It left room for ambiguity and did not provide a clear roadmap for my study. Additionally, I struggled to establish a coherent connection between my research method and objectives. Furthermore, as I scrutinized my RPO, I noticed that I had a surplus of 150 words. This extra word allowance presented an opportunity for me to expand on my key points.

I've developed a comprehensive checklist to guide the effective utilization of language in articulating my research. In the introduction, I aim to provide a clear problem statement and emphasize the significance of my proposed solution:

When summarizing existing research, my goal is to bolster my understanding and categorize it into three core pillars. The research objectives encompass two primary elements: promoting algorithm awareness and knowledge acquisition. I also underscore the role of speculative design in prompting critical reflection rather than furnishing immediate solutions. In the 'Approaches and Methods' section, my intent is to rationalize the choice of a critical journal approach as the most suitable method for my speculative prototyping. By ticking off this checklist, I'm confident that my research proposal will be both clear and concise.

Challenges with Ideation

The ideation process has proven to be quite challenging. Creating a speculative prototype entails checking numerous boxes to ensure its true speculative nature. To effectively stimulate critical thinking, I delved into a plethora of references that touch upon human psychology and behavior. Simultaneously, I've been striving to gain a comprehensive understanding of the current perspectives on algorithm curation.

However, I've also encountered some limitations stemming from the computational nature of the atelier. Many of my ideas demand a strong grasp of coding, and the format I've opted for feels more like a personal preference rather than a strategic choice for the prototype. My motivation is to leverage this opportunity to familiarize myself with TouchDesigner, as I see its substantial potential for future interdisciplinary collaboration. The challenge I'm currently facing is that becoming proficient with TouchDesigner requires a significant investment of time.

In the past, my approach to 'making' in computational projects significantly differed from my approach in other projects. This difference can be due to my lack of experience with non-traditional computing software. My initial plan, which was to generate one idea for each of the listed dilemmas, hasn't yielded the desired results. Consequently, I've opted for a shift in strategy. Instead, I'll allow ideas to emerge serendipitously as I become more acquainted with TouchDesigner. During this learning process, I also aspire to craft small experiments, thereby enriching the quality of my catalogue of making.

A list of abandoned ideas


Learning TouchDesigner &
Small Experiments

In all of my experiments, I've consistently employed the same approach. When I engage with tutorials, my focus is on identifying and documenting the parameters that directly impact the final result. I've discovered that this method is more efficient than gradually comprehending the function of each parameter within every operator.

For each of the experiments listed below, I will meticulously record my key takeaways and learning points.

A glimpse into my saved tutorials


Experiment 4 – Image Instancing

In this experiment, I delved into the concept of instancing, a technique that involves generating multiple copies of an object using a single render call. This computational approach not only conserves GPU and CPU resources but also contributes to smoother frame rates.

My exploration extended to a comprehensive understanding of the parameters within the noise function, encompassing both CHOP and TOP variants. Additionally, I found the feedback OP to be pivotal in creating a feedback effect, one that could be influenced by the blur and ramp OP.

Through experimentation with the parameters of the blur and feedback OP, I achieved markedly diverse outcomes. By adjusting these parameters, I could make the feedback appear more fluid and even produce a captivating kaleidoscope effect in the background. The noise CHOP function played a crucial role in functioning as a control mechanism, allowing for adjustments to the speed, quantity, and path of the images.

I noticed that noise is used a lot for abstract visuals, but personally, I would prefer a more controlled rather than random input. So I’ll keep a look out for what can achieve that.

Side by side view of using Image as texture


Experiment 5 – Image Instancing with Particles

This experiment represents an extension of the previous one, where I explored the application of instancing as a texture overlay on an image. I was intrigued by the idea of symbolizing data manipulation in this way, so I decided to give it a try.

I must admit that I faced some challenges in comprehending the intricacies of this experiment, as it involved a rather complex process. The approach included the use of 4 math TOPs to extract RGBA data from the image, followed by the reordering of this data into a geometric structure. It was confusing as the references for the values were used in seemingly specific ways that didn’t feel logical to me. The Red and Green values were plugged into translating the x and y axis, while the Blue value was plugged into scaling the Y axis. Does it not sound a little random?

I found limited control over this particular file, and my scope for adjustments was mostly confined to the noise and blur operators (OPs). Nevertheless, despite these limitations, I was able to achieve a distinct outcome compared to the tutorial. The tutorial's result had a more fluid quality, which didn't align with my vision. To introduce a more 'aggressive' aesthetic, I decided to increase the period of the noise and the filter size of the blur, resulting in the desired effect.

An overview of the nodes


Experiment 6 – Particle Flow

In this particular experiment, I was introduced to 'particlesGpu,' a shader-based particle system. I found this tutorial notably more straightforward and logical to follow.

The process involved using a 'ramp' and reordering the RGB values within the ramp. We then utilized 'point transform' to center it along the xyz axis. Subsequently, we generated particles atop this setup using 'particlesGpu,' followed by the application of a 'feedback TOP' and movement manipulation through the use of noise. The diverse outcomes were driven by variations in the parameters of 'particlesGpu' and noise. What piqued my interest was the incorporation of a keyboard in CHOP, enabling the creation of keybindings for altering the seed of the noise inputs. This added an element of interactivity that I found quite valuable.

At this juncture, my familiarity with the noise operators (OPs) was growing. A quick glance at the results allowed me to pinpoint the necessary tweaks to either enhance the frame rate or slow down the movements. I also noticed a significant improvement in my workflow, as I became well-versed in keyboard shortcuts, and the file setup process felt more streamlined.

Parameters of particlesGpu

Experiment 7 – Pixel Sorting

As I explored ways to manipulate images or devise filters to symbolize data extraction and manipulation, I encountered 'pixel sorting.' Although I'd encountered this concept numerous times while working with p5.js in the past, this was my first time implementing it in TouchDesigner.

This endeavor also marked my initial foray into the Python coding language and the utilization of external scripts. Setting up my VSCode required the installation of several extensions, though the tutorial didn't provide specific guidance on this aspect. I'm proud to say that after some research, I successfully added the module path, overcoming this initial challenge.

However, I confronted several difficulties with the code, which felt distinct from working with p5.js or HTML/CSS. I spent about an hour troubleshooting only to realize that my functions weren't recognized due to a missing indent. I also attempted to tackle part 2 of the tutorial but encountered insurmountable errors, leading me to spend over two hours before seeking assistance from Andreas via email.

Given the heavy reliance on scripts, I had limited parameters to experiment with in this project. To be honest, I didn't fully grasp the code's intricacies beyond the ability to alter the direction of pixel sorting. Part 2 delved into more advanced experimentation, which I struggled to reach.

This experiment was considerably more complex because the creator had optimized the code, but this also made it less beginner-friendly due to its brevity.

Up to this point, working on this experiment has been a source of frustration since my primary aim was to achieve the visuals outlined in part 2.

Heavy reliance on Code


Programming in Python

Experiment 8 – Particle Cloud & Pseudo Plexus

I must admit that this experiment ended up as a disappointment, even though it held immense potential, and I could envision creating something akin to the work of Refik Anadol (satire).

Several factors contributed to the failure of this experiment. Firstly, it placed an excessive burden on both my CPU and GPU due to a lack of optimization in the parameters. My frame rate consistently hovered in the 10-15 range, and troubleshooting proved challenging, primarily because of my limited knowledge about the operators being used. The project featured numerous points, lines, and nodes overlapping each other, making optimization a complex task. In my efforts to troubleshoot and optimize the process, I inadvertently crashed the file.

Despite the setbacks, I was genuinely captivated by the visual elements in this tutorial and felt a sense of disappointment that I couldn't achieve them. I could already envision each layer of the constant as an image, with the lines symbolizing algorithms extracting data to recommend another image. This experience underscored the crucial lesson of optimization, reminding me to revisit my catalogue of making and optimize all my files, in consideration of sparing others from similar frustrations.

9 frames per second...