How to use these methods to generate a design?

To answer this question, I start with a critical aspect: When reading is completely given to conditions and data, How will the text gradually lose its established reading system? Therefore, I tried to develop a rigorous process.

Normally, design focuses on adjusting layout until the text becomes readable. In this project, I stop adjusting the layout and move all design decisions to the rule-setting stage. Once the rules are defined, background data controls position, scale, and visibility. Reading is no longer guaranteed—it becomes a conditional outcome.

The reason that I need to define how to read first is because if I don’t define how the text is meant to be read first, the conditions can’t be evaluated.

---

Subject (baseline)

I chose Conditional Design Manifesto as my starting point. The text which is retrieved from online is regarded as a neutral linguistic material, not the content itself.

---

Data resource (site)

I chose this wooden surface as the base (background data) because I needed a physical site to ground the poster in reality.
The wood grain provides a wide range of background variations, allowing more possibilities when mapping parameters to the design.

---

Final outcomes

Iteration1 : Declarative Reading

---

Iteration2 : Non-linguistic Reading

---

Iteration3 : Non-hierarchical Reading

---

Iteration 4 : Indexical Reading

---

Transformation process

The image on the left shows how different text elements correspond to specific positions. The parameters of these positions are reflected in the applied conditions.

How to Hack

I reposition Cavalry from a tool for generating visual or motion effects into a system for setting and testing typographic conditions.

In this context, the image no longer functions as content, but as a source of conditions.

---

Data Source and Sampling Strategy

I choose a 5-second segment of film footage. and using it as a time-based data source. The narrative content is ignored; only numerical value changes are considered. In addition, sampling is taken at different time points within the clip.

---

The structure of generating design

Layer 1: Baseline (unconditioned plane) (Control Variables)

Layer 2: isolation (Single-condition intervention, 5–6 pieces per group)

Layer 3: Compound (Condition stacking, 6-9 pieces per group)

---

Layer 1: Baseline

These layouts align with three different variations.

---

Layer 2: isolation (Single-condition intervention, 5–6 pieces per group)

Rotation

Skew

Character spacing

Text box size

Word spacing

---

Layer 3: Compound (Condition stacking, 6-9 pieces per group)

In this layer, I used multiple variations as conditions.

---

Font size + skew

2. Font size + opacity

---

Knowledge Generated Through Iteration

1. I came to understand that the page itself is not neutral, but a conditioned structure.

2. What we usually perceive as a blank page is simply a condition that has been naturalized and hidden—a state in which parameters are present, but effectively set to zero.

3. When conditions can be explicitly set and adjusted, the role of the designer shifts from executing forms to defining and modulating conditions.

I choose Cavalry as a tool for me to develop in the beginning. It’s an application for motion graphic and data visualization. The most important characteristic of Cavalry is the parametric control system.

After that, in the first part — iteration 1, I selected a motion graphic design to replicate. During the process, I learned lots of tools and functions, one of them evoked my interest — the image sampler.

Image sampler is not merely an image-based visual effect. Instead, it functions as a converter that transforms images into numerical values readable by other parameters. The core output of the Image Sampler is not an image, but a data field. This data field is a numerical distribution composed of pixels.  For example, brightness / RGB / grayscale.

In the example below, the data is translated into shapes (squares), where areas with higher brightness appear, and areas with lower brightness do not.

Treating the Data Field as the Starting Plane

In conventional typography, the starting point is assumed to be a neutral blank page. However, a surface generated by the Image Sampler already contains conditions. This surface functions like an invisible landscape. The conditions only become visible when text is placed onto it.

This process led me to the following critical question: 

1. Is the so-called “neutral blank page” actually a set of naturalized and hidden conditions?

2. When these conditions become visible and operable, does the act of design itself change?

How to hack?

I reposition Cavalry from a tool for generating visual or motion effects into a system for setting and testing typographic conditions. In this context, the image no longer functions as content, but as a source of conditions.

The hack will be understood in three layers:

1.Expected output

Cavalry default: motion / effect / generative visuals

Hack: use it to create and test “initial conditions for typesetting” and “layout behavior”

2.Tool role

Default: Image → Visual Elements

Hack: Image → Data field → Typesetting conditions (condition translator)

3.Design process

Default: design the form first and then add the effect

Hack: Set conditions and relationships first, and the form is generated and read as a consequence.