paperdove.com

Dihak Storytelling

Sat, 14 Feb 2026 00:00:00 GMT

The website for the Dihak Storytelling course.

Cube-as-a-Service

Thu, 12 Feb 2026 00:00:00 GMT

Cube-as-a-Service (BETA) is a personal experiment that was built with the help of AI coding tools. You can upload your cover artwork images, add effects and adjust lighting to create a 3d preview. The original inspiration comes from Bookup.

I approach the AI with both curiosity and skepticism. I am beginning to find it useful in the narrow context of building software tools for myself and others especially in the area that big tech companies have zero or negative interests. I am also aware of the underlying issues surrounding LLMs. I am trying but struggling to navigate this new landscape just like many others are. Where do you draw the lines? Is there a way to make peace with this technology? Do small tools like this change your perception?

Posting Web

Sun, 18 Jan 2026 00:00:00 GMT

Posting Web improves upon Posting app that I created in 2024 by now running on a remote server 24/7 and adding the web interface for better usability.

The goal is the same as the last version — help users spend as little time as possible on social media platforms while still maintaining their online presence. A user may be an artist who needs to promote their work without too much stress or a small business owner who has a regular posting schedule but don’t want to check in every time.

When I upload posts directly on each platform, I tend to get distracted by viral posts recommended by algorithm and waste time scrolling without realizing. These companies spend millions of dollars to keep you hooked in so it is a game I cannot win. I also become self-conscious on how well my post is doing. I now use my app to schedule ahead without interacting directly with the platforms and it is especially helpful to publish posts while I sleep. This obviously doesn’t solve any inherent issues of social media, but has helped me regulate my own use. It is also very relieving not to be exposed to all the clickbaits and AI slop.

My Instagram usage stat, totalling 3 minutes in the past week

I also added a few helpful features. The app protects privacy by stripping metadata of any attached media files. It removes any sensitive data such as GPS location or device used, etc. before uploading to each platform server. It also checks for file size and image dimensions to meet each platform requirements, thus, saving user’s time on preparing them. The UI is simple and minimal to keep the time spent on scheduling as little as possible. If I’m spending a lot of time on my app instead of on social platforms, that’s still a waste of time. I simply write a message, attach a few images, hit the schedule button and forget.

Currently, the app is only available behind VPN for a small closed group of people around me.

I set out to solve one problem and I am quite satisfied with the result, but in doing so, I neglected another problem that is the use of AI. For this web version of the posting app, I got a lot of help from AI coding agent. Last summer, I tried Cursor for the first time and although it was quite impressive, it also made mistakes that made it quite frustrating to use. I gave another try by using Claude Code this time and was surprised at how much better it became. I am still skeptical about the overall AI landscape. There are so many things to be concerned about — environmental costs, economic collapse, weaponization of the tech, consolidation of political power, no compensation to creative people, just to name a few. So, it leaves me a bitter taste in my mouth even after I found a good use of the LLM coding agent in this project. I am still struggling with how I should feel about all of it. I have zero reason to defend any of the big AI companies. All I can say is that in this very narrow context of the project, it was helpful.

Visual Tactility

Sat, 27 Dec 2025 00:00:00 GMT

I taught a 10-week course called Visual Tactility at Design School in Seoul with a small group of students with minimal design background.

The course emphasized the direct experience from the primary source. This feels more important than ever when we spend most of our days looking at a flat screen watching short-forms or talking to AI with no real engagement with reality.

experience to language/concept to image/text/video/etc

We explored design possibilities that engage our sense of touch, from creating tactile boxes inspired by Moholy-Nagy at Bauhaus to producing objects that connect users and the world as an extension of our body. Tactility is multi-sensory, which may include all five (or eight) senses. When we consider not only the sense on our skin, but also weight, pressure, temperature, spatial and atmospheric expriences, there emerge interesting possibilities to create a rich experience. As Kenya Hara puts it, we can “awaken” the senses.

From top to bottom; continuum between direct (primary) exprience and visual (secondary) experience; From idea to handicraft to digital and the space between them; the parametric space of visual expression

The final project prompted students to consider tactile experience of the 2-dimensional graphics they create as well as the physical object itself. Below are a few examples of the student work:

Posting

Fri, 02 Aug 2024 00:00:00 GMT

As an artist and designer, I have to use various social media platforms to share and promote my work. It takes a lot of energy to maintain several accounts not only because it takes time to open up each app or websites, write text and attach files but also because these platforms are carefully and intentionally designed to entice users into spending (or wasting) as much time as possible. If you are like me, this is not news. Often times, I open an app to upload a new post, and before I know it, I have already spent a half hour mindlessly scrolling. Worse, they force-feed you things you are not interested in only because it benefits their bottom line. They are full of controversy, anger, frustration, jealousy, FOMO interspersed with a quick dopamine hit. The only thing these platforms care about is that you spend more time in their apps. For example, Instagram still doesn’t let you include a clickable URL because they don’t want anyone to leave the app.

You also have to play by their rules. Their algorithm decides which posts get boosted and that has a real consequence to artists, designers and small businesses. If you are only connecting with your close friends, then this may not be a big issue but if you have to use them for your business, it’s an emotionally draining experience most of the time. As much as furstrating the experience is, I cannot avoid using it. We invest our time in planning, curating, posting and managing these accounts hoping to get exposure and some returns. Then, suddenly, they change their recommendation algorithm or even remove your account. I have a bitter experience of my Instagram account with over ten thousand followers removed without any clear reason and that ruined my online network and freelance business. Over the years, I learned that they own everything and it’s no use trying to build my own space on these platforms or optimize my work for them when I have no control.

It’s a game that I can never win but I don’t have much choice but playing it. So, at least to feel a little better, I have been trying to be passive in my social media use. I have tried muting words and turning off notifications. I removed the apps from my phone and only accessed using the web interface. I tried to only look at a few posts quickly and leave the apps.

Continuing that effort, I created a tool called Posting, with which I schedule and automate social media postings on multiple platforms at once via terminal interface. As much as the platforms take all the power, there is a tiny opening called a free developer API or Application Programming Interface. It is a way to directly connect to the platforms without going through their user interface. I am using each platform’s free API to automate postings in advance and reduce my screen time.

By scheduling in advance and not having to deal with whatever the algorithms decide to shove in my face, I have been feeling better. By delegating scheduling and posting tasks to my app, I have been less distracted by social media during the day. I now tend to forget that I have uploaded a new post and thus, check the apps less often as a result. I can still check in from time to time to see other people’s posts or reply to a comment when I want to.

I know it is a false sense of control as the free API access can be revoked or removed at any time. Twitter recently scaled the free API down to almost zero other than writing posts. Meta API, which I used for the first time for this project, is just frustrating to work with. It has incomplete, wrong or missing details as it doesn’t serve their interests to give any power to third-party developers who use their data. In contrast, Mastodon API was a breeze to use. Their source of income is solely on user supporters. They don’t serve ads. By design, they cannot control the entire network. Bluesky that doesn’t seem to have many users at this time, is also taking a decentralized approach and the API was open to use.

I am hoping more people will move to decentralized platforms like Mastodon in the near future, which, although it has its own problems, I believe is a much better option than centralized ones such as Meta or Twitter. Just look at the track records. As a user and a developer, I hope for a future where I can have more control over my digital life and work. Until then, and until the platforms cut off my access to these free APIs, I will enjoy a tiny bit of control over my social media use with my app.

Related: I was reading Radical Technologies and Filterworld before and during the program development. Highly recommended.

Rename Screenshot

Mon, 22 Jul 2024 00:00:00 GMT

Rename and organize Mac screenshots by their contents with the help of AI. This tool watches for any new screenshots, renames it to describe its content, moves it to one of the pre-defined categorical directories.

The default Mac screenshot file names are not formatted well in my opinion. It does not sort nicely due to mixed use of single and double digits. The date string in the middle is not easy to read. And once you start collecting many, it becomes difficult and cumbersome to find what you need and organize them.

Below is a demo video showing how newly added screenshots are processed, renamed and moved to a user-defined folder:

The program watches any new screenshots added to your Desktop. When it detects one, it sends the image to your chosen API provider (either local Ollama or ChatGPT) to get a new filename suggestion. Once the response is received, the screenshot is renamed and moved to the pre-defined directory ~/Desktop/Screenshots. If there are categories set, it will be moved to a corresponding sub-directory. For example, any code screenshots can be saved to code directory. The original screenshot is kept as a backup, which you can later review and delete. It will continue to run in the background monitoring any new screenshots or you can just run it for existing images only.

The credit goes to Charlie Holtz for the original project and inspiration. It was such an interesting idea so I thought I’d make my own version. You can download Rename Screenshot on the Github repo. It only works as a CLI app at the moment.

Supercut

Fri, 19 Jul 2024 00:00:00 GMT

Supercut is a tool that creates a montage video with Google Gemini large language model. You can ask it to search video contents (ex. “Find sections of street signages”) and it will respond with timestamps. Then, corresponding video clips are extracted and stitched together to create a montage video.

The Gemini model responds to a request with timestamps from the input video. Supercut then creates video clips according to the timestamps and stitch them together to create a montage, automating the process for quick and easy review. In case of getting unsatisfying results from the full automation, Supercut also provides several sub-commands to give user more control over the process such as uploading, encoding, timestamping, video clip generation and video concatenation.

I uploaded a 13-minute long public domain documentary A Bronx Morning (1931) found in the Library of Congress collection and asked to find street signages. Supercut created the following montage video:

Another example. I uploaded 6 animated films from the silent film era found from the same collection, and extracted “text sound effects.” Some timestamps were unrelated and had to be removed manually before creating the montage:

In addition to the examples above, I used the batch processing feature of Supercut to process multiple movie trailers to extract typography shots to get a quick sense of what styles are being used. Supercut saves each video clip before making a montage, so if there are any unwanted clips, I can manually remove or bring them into an external software for further editing.

I also used prompts such as “find scenes where red is the dominant color” to generate a montage of shots that had interesting use of the color. The answer was not always as good as I hoped but I still liked how it presented the materials in a way that I’ve never seen before or been able to assemble in my head. These are the things that I would probably have never tried if it weren’t for such tools because of the amount of manual work involved without a guarantee that I’d get something interesting at the end. Now, I can save time with the initial gathering and generation of the materials. This is where AI can be useful in my creative process at the moment. I can also imagine researchers and historians using a similar approach when dealing with a large amount of historical video recordings.

While making the tool, I found the responses from Gemini were not accurate in some cases. It’s hard for me to tell whether it is because of the model architecture or training data or both. A reason I can think of is that it only looks at video at 1 frame per second and thus, missing any action happening in-between the interval. Also, because of that, it does not understand the continuity of movement. It also gave wrong answers on questions about color or composition especially on longer videos.

As long as we don’t trust its answers blindly as facts (bad idea) or attempt to create a finished film with just one click (another bad idea), I believe it can still be a useful research tool or even a creative tool for quick iterative experiments.

The executable code is available to download from the Github repo and it is currently only available as a CLI app.

Sequenced

Thu, 11 May 2023 00:00:00 GMT

Sequenced is an exploration of generative motion compositions. Line segments are all connected in a loop and move within the constraints of a simple 2-dimensional grid structure. Each segment is assigned a sequence of motion data to follow. Some data is reused, while others are generated in real-time. Together, they act like a musical score but for the visuals. It falls between predictive sequences of rhythmical motion and endless random permutations.

The project is part of my ongoing exploration into programmed motion. It runs in real-time inside browsers. It is created with my own open source tool Ssam.js with help from other libraries.

Sequenced is available to mint on the Solana blockchain.

I took inspirations from music - how it uses repetitions to create impressions and how variations and improvisations bring surprise and vibrancy. Digitally created motion is precise and has regularity but it can go out of control quickly with mathematically applied randomness. It is a balancing act to find an interesting middle ground between being too random and thus boring and being too precise and also boring.

Just like musical score alone is not music, coded instructions are not motion. In music, everything flows over time. Motion-based work, unlike still images, creates a temporal rhythm in-between images. This is something that cannot be described on paper or with numbers in code. It can only be experienced when it is played over time or when the code is executed.

Rules guide visual rhythm, overlapping and shifting to create playful moments of predictability and chaos. You glimpse patterns, then they disappear. Strangely, I find it calming to watch the infinitely generated overlapping visual rhythm it creates.

Selected Exhibitions

Processing Community Day 2024 at Porto, Porto, Portugal, 2024
Taipei Hacker House, Taipei, Taiwan, 2023

Click each image below to see live version of the minted tokens.

Press c to show code and press 1 or 2 to change responsiveness. You can also add optional URL parameters: [w] and [h]: set width and height in pixels. (ex. w=1200&h=600); [pr]: pixel ratio. set it to 2 for hi-DPI screens; [ar]: maximum aspect ratio; [bpm]: change speed (ex. bpm=72)

Knowledge

Mon, 20 Mar 2023 00:00:00 GMT

What does it mean to “know” something? Just because we do not see it does not mean it is not there. It does not mean it is not happening. Many things take time and energy to happen and to be understood.

Look at how long it takes for rocks to become sand and how ephemeral we are in the passage of time. Drops in a waterfall evaporate within a few seconds. But they blend, collide, and hit a rock or two before they disappear. These are the moments that I cherish, that I find meaning from, even if no one will remember them. After all, it is small drops that turn rock into sand. It is grains of sand that change the surface of the Earth.

Knowledge is a generative NFT project that was launched on fxhash, and lives on the Tezos blockchain.

Click each image below to see live version of the minted tokens.

You can add optional URL parameters: [w] and [h]: set width and height in pixels. (ex. w=1200&h=600); [loop]: play as a loop. you can add a number to randomize. (ex. loop=1234); [pr]: pixel ratio. set it to 2 for hi-DPI screens.

Ssam.js

Wed, 01 Mar 2023 00:00:00 GMT

Ssam.js (쌈 as in Korean dish) wraps your HTML5 Canvas sketches and provides helpful features such as animation props and file exports. It is inspired by canvas-sketch. It is an open source project for anyone to use. I’ve been using it to create all my web sketches and genrative artworks since 2023.

Some people ask why I create my own library or how this is better than others. First and foremost, I create tools for myself. I create tools to learn and understand the materials I use, to get new ideas and to have a better control over my art-making process. Creating tools small or big is an important part of my process. There are already better or similar tools out there but they will never give me joy of making them myself, making mistakes along the way and learning something valuable. Working with my own tools allows me to feel comfortable while leaving room for experimentation and exploration.

Features

TypeScript: It can work both in JavaScript or TypeScript projects.
ESM: You can use it alongside other ESM packages.
Vitejs: Vitejs is modern and fast. You can extend the config to add functionality. Ssam also has a few of its own plugins. Multiple sketch modes: It supports vanilla Canvas 2D API, WebGL context, or you can use with other Canvas libraries (Three.js, OGL, Two.js, Pts.js, etc.) as long as they support an existing canvas. p5.js is not supported (yet).
Hot Reload: Code updates will be instantly applied without a full page refresh and wthout losing the sketch states.
Git snapshot: Commit your code to git and export a frame with the same hash for an easy archiving and retrieval.
Timelapse documentation: Automatically export an image when you save your code to create a visual documentation of your project over time. Animation loop: The playhead prop repeats 0..1 and makes it easy to create a seamless animation loop. Other props such as time, frame, deltaTime and loopCount are provided as well. You can also adjust frame rate for playing and exporting independently.
Sketch settings: Many essential settings reduce boilerplate code in your sketch - animation duration, playback frame rate, filename, etc.
Sketch props: Use props for each mode such as context, width, height to help your coding.
File exports: Export canvas as image, animated GIF, PNG sequence or MP4/WebM video at various frame rates.

Duet

Sun, 21 Aug 2022 00:00:00 GMT

Duet is a long-form generative artwork that was released in August 2022 on the fx(hash) platform.

It explores generative + computational motion and creates an almost infinite sequence of visual compositions. Two groups of shapes transform, merge, and split randomly. What can be disorienting starts to make sense as every element is animated from one to the other. This transition animation is what connects random pieces and creates a relationship between them. Each element is mathematically interpolated, with its motion defined by an equation.

Although there are key moments, it is hard to summarize time-based work as a static image. As a constantly evolving generative motion piece, Duet invites viewers to form their impression by feeling the rhythm over time.

Click each image to view live version.

Introduction

Duet explores generative + computational motion and creates an almost infinite sequence of visual compositions. Two groups of shapes transform, merge, and split randomly. The properties of these visual elements are generated in real-time inside a viewer’s browser. They have nothing in common other than the rules initialized according to a transaction hash randomly assigned at the time of minting a token.

What can be disorienting starts to make sense as every element is animated from one to the other. This transition animation is what connects random pieces and creates a relationship between two unrelated numbers, forms, and compositions. Each element is mathematically interpolated. Interpolation is the process of finding in-between values. With what equation is used to interpolate determines the quality or characteristics of motion, and at what rate the values are sampled determines the smoothness of motion.

Background

With recent projects, my main motive has been to explore and experiment with ways to represent time and motion with digital media mainly through computer programming. With the Structures series, I used my own software, Modular Design System One, to work with rigid grid structures. The colorful shaded rectangular blocks that look 3-dimensional are fixed in place, and essentially, act like big blown-up pixels. In this limited environment, only the colors change and create an illusion of motion. With my last long-form generative piece Time Intertwined, I explored two inter-connected or intertwined timelines to create an infinite number of fixed-length loops. Duet continues on that idea.

A few examples from Structures and Time Intertwined

Computational Motion

I am conscious of the medium and the tools I use to create work. I am interested in how and with what we create digital, time-based work. How much of what we create is enabled or limited by the tools we use? Why did I decide to work with computer code instead of paper or Adobe softwares? Is it because everyone was doing it? Is it because there are things that can only be done this way? We all know that a computer is a powerful tool that can do almost anything as long as we can provide the proper instructions. It is also a meta-tool that creates other tools. Early computer art pioneers had to either create their own tools or collaborate with those who could create for them. Compared to those old days, it has now become a lot more accessible and easier to work with the computer. But at the same time, it is also true that most of the time we use the same softwares and with a pre-defined finite set of menu items, for example, in After Effects for digital animation.

After Effects is great in its own right, but these days I am exploring alternative ways to create motion with my own software that I program. These tools may be simple and rudimentary compared to commercial softwares but they are flexible and allow me to do things that cannot be done otherwise. When I create software with code myself, I have direct access to all the underlying data and how they are combined to create motion, and through that process, I can make unexpected discoveries and develop my own visual expression. I feel there is still more we can do when we program digital motion ourselves. Below are a few ideas that I explored in Duet.

Timelines

In my previous project, Time Intertwined, I explored two intertwined timelines. I have not yet found a good way to explain the idea but one way to think about it is the Earth rotating on its own axis as well as revolving around the Sun. Essentially, there are two connected timelines — one goes for a day and the other for a whole year. It isn’t just the duration of these timelines, but each has a different effect — changing day and night vs. seasons. In Time Intertwined, there are the structural timeline and the surficial timeline. The motion in each is linked just like in the example of the Earth and the Sun, but unlike the Earth-Sun relationship, it can be controlled independently from the other, which results in an almost infinite number of combinations, and creates a fixed-length loop or a cycle. In Duet, I expand this concept to have an infinite-length timeline.

Time, Not Frames

In a typical motion-based work, we need to define frame rate or frames per second so that the continuous time can be converted into a discreet set of images. In animation, it is the number of drawings per second. All the key timing and spacing rely on it. In a live-action film, it is how often time is sampled. We still need the frame rate in digital animation because that is how often the screen updates and refreshes the image. 24 or 30 fps has been good enough but with today’s technology, we are reaching the level that is almost indistinguishable from reality, or at least equal to or better than the analog counterpart. To me, there is something transcendental when I can no longer recognize the technology and get immersed in the work. Of course, it is still an illusion after all, but a very convincing one.

A typical computer screen supports up to 60 fps, but more recent monitors support 120fps or even higher. To adapt to different frame rates, I decided to use delta time that adapts to a screen’s refresh rate. This is nothing new to a game developer, by the way. I still use the concept of keyframes or key moments but every parameter of visual elements such as forms, motion paths, and colors are mathematically interpolated with an equation in real-time and the equation can spit out a value at any point in time no matter what frame rate is being used.

Duet #157 looping from 0 to 6.

Duet #6 responding to different window sizes.

Responsiveness

In Duet, everything exists as code or computer instructions. The code gets executed only when the work is presented within a browser. As mentioned above, time is sampled in real-time instead of using the pre-defined frame rate. It also adapts to any screen resolution. This is different from pre-rendered digital videos where you have to adhere to any existing standards. Once the video is rendered, it can only have a fixed resolution, frame rate, and duration. Another point related to the real-time rendering is that the total file size of the Duet code is only 93 kilobytes. That is smaller than a typical single image file.

Visual Rhythm

Duet does not have any audio, but I took inspiration from music. I wanted to create a generative artwork about the visual rhythm that is not only seen but also felt over time. Rhythm comes from repetition. By default, Duet does not repeat the same animation twice. It looks like it is looping but it is not. If you want, however, it can go into the fixed-length loop mode, too, by pressing the “n” key. Every composition state is generated in real time according to a pre-determined random sequence of numbers. When you refresh the browser, it will play the same sequence of motion again from the beginning, but as long as you keep the window open, it will continue to generate a new composition. There is never a definitive ending to this piece. What is repeating is the rhythm that only comes alive when the work is played over time.

Generative Motion

As a generative motion piece, Duet endlessly generates a new composition within the boundary of a pre-defined set of parameters such as position, color, line thickness, etc. This set of parameters makes up the overall structure of the work and it is what I, the artist, decided to include, or not include, in the work. But the way these parameters are combined to create the final outcome is not in my control, and that is where the generative aspect comes into play. And because the motion sequence is never-ending, there is always something new to discover.

While working, I thought about how we experience an artwork, especially a time-based one. When we look at static artwork such as a painting, we still have to take time, move our eyes around, see from distance or up close, and take in different parts of the image at a time. In motion-based work, the perception of time is more obvious. And because the images keep changing, it’s hard to define or summarize a work as a single image. The essence of Duet is its motion and rhythm.

Among all the random composition variations, I found myself attached to a certain section and wanted the ability to only reproduce it without having to play all the way from the beginning. As a browser-based artwork, it was also important to have the ability to share interesting moments with others. Duet provides a way to navigate to any section by using the URL parameters, which will pre-run the generative functions up to that point.

Closing

Everything in Duet looks seamless and smooth thanks to the computational power of precise calculation. I have a personal desire to keep everything in perfect order and I have to fight not to go all the way because it can look too stable and not interesting anymore. There’s an irony here — I love to work with geometric forms and grids for their simplicity, orderliness, and proportional beauty, but at the same time, I’m deliberately trying to avoid being locked in them. The beauty is somewhere in-between. After all, what drives my work is motion which is about change.

I also have to keep reminding myself that nothing in life is perfect. In fact, what I found while working on Duet was that it was not the case at all. The seemingly perfect look is the ideal that I was pursuing, but in reality, I had to make a lot of adjustments to make it look seamless. Technology is never perfect, and even though I am directly manipulating computer code without any intermediary interface, I still have to rely on a lot of infrastructures to make it work. For example, I had to consider how browsers behaved differently. There is never a perfect solution. Instead, I have to embrace the fact that not everything is in my control just like I do not know what random composition Duet will generate next.

Duet received an overwhelming response from collectors and minted out quickly except for the six reserved tokens for future use myself. Currently, the 250 minted iterations are owned by 157 collector wallets on the Tezos blockchain. I am motivated by all the support I got. I will continue to use this newly developing decentralized space to share and release my future digital artworks.

Click each image below to see live version of the minted tokens.

Selected Exhibitions

Bideotikan Digital Art Festival, Bilbao, Spain, 2024

Check the project page for instructions on keyboard interactions and URL parameter settings.

Time Intertwined

Mon, 23 May 2022 00:00:00 GMT

Click image to view live version

Time Intertwined is a generative NFT project that was launched on fxhash, and lives on the Tezos blockchain. There are 256 limited editions, each with unique properties.

Time Intertwined is my exploration into time and motion. Two linear timelines - structural and surficial - are intertwined. By weaving these timelines at different points in time, almost infinite variations can be made. The changing relationship between the two creates dynamic and unique compositions and challenges the visual perception.

Selected Exhibitions

Bideotikan Digital Art Festival, Bilbao, Spain, 2024
Digital Art Mile, Basel Art Center Space 31, Basel, Switzerland, 2024 (more link )
Motion Evolving, Beijing, China, 2023
Unblock Gaudi, The Museum Angewandte Kunst, Frankfurt, Germany, 2022 (more link )

Featured

Creative Applications, 2022

Click each image below to see live version of the minted tokens.

Check the project page for instructions on keyboard interactions and URL parameter settings.

Nodes

Wed, 09 Mar 2022 00:00:00 GMT

A series of experiments exploring motion and connections. I created a network of nodes with a custom growth simulation algorithm that is similar to Diffusion-Limited Aggregation. Each node is created sequentially and remembers its parents and children. By animating the nodes over generations, we get an interesting insight into the makeup of the network.

Growth One

Fri, 04 Mar 2022 00:00:00 GMT

Growth One is a generative NFT project that was launched on fxhash, and lives on the Tezos blockchain.

This project visualizes a growth simulation of a virtual micro organism that follows simple rules. The organism starts with a few seed nodes, and it grows when the seed nodes and search nodes accept each other when they meet certain conditions. The organism loosely resembles structures found in nature such as coral reef or fungus. Do we embrace the beauty of its emerging structure or reject it because of pathogenic associations?

Token #5 with random seed nodes placement; Token #7 with horizontal placement; Token #10 with circular placement

This project visualizes a growth simulation of a virtual micro organism that follows simple rules. The growth algorithm is somewhat based on Diffusion-Limited Aggregation or DLA. I was introduced to the algorithm some years ago by one of the Coding Train videos, and liked its organic quality and the simplicity of code implementation. Thinking about ideas for a generative NFT, I naturally thought about this algorithm. For this project, though, I did not reference any existing code because I wanted to create my own custom version with a different way to grow the virtual organism.

Seed Nodes

Where I spent most time was building the growth simulation method. The organism starts with a few seed nodes, the number of which is randomly assigned for each token. Then, there is the seed layout, the value of which is used for a horizontal, circular or random placement of the seed nodes.

Examples of inner growth; Token #9 and #13

Search & Accept

After the seed nodes are placed, a search node is added one at a time. The seed nodes want to grow themselves and the search nodes want to be added to an existing structure. But instead of the existing nodes accepting any and all new search nodes that come close, I added a few conditions to make the growth process more varied and interesting. If we only check for the minimum distance as an acceptable condition, the inner nodes will not get a chance to grow. there’s a threshold so sometimes search nodes go past the closest nodes to attach to others. This condition resulted in some interesting variations especially with circular layouts because the structure can grow both outwards and inwards.

As the organism starts to grow, it loosely resembles structures found in nature such as coral reef or fungus. While creating it, I encountered many interesting and beautiful emerging visual structures, but at the same time, I couldn’t help but associating it with pathogens growing out of control.

Finding the right balance of these parameters, or fxhash features as they call them, was not easy. The simulation can be unexpected and some results were not good. Sometimes, there was no growth at all. It required a lot of manual tweaking. Although the simulation itself is automated, finding the right range, right amount of variations is where each artist is going to use their own judgement. I ran well over a thousand randomized testings to make sure the system will stand any random hash values fed into it. But honestly, I am not entirely sure if I have caught every subjectively bad case.

Real-time Growth Simulation

Early on, for debugging purpose, I was switching back and forth between running the whole simulation in a loop and running the same simulation in real-time to see what is really happening. This real-time simulation usually takes anywhere between 5 to 30 minutes. It is intentionally very tedious and inefficient process due to my decision to only add one node at a time and use custom search/accept rules, and I enjoyed watching the organism evolve. Every step of the growing process reveals something interesting about the organism. The building process is never finished as it will continue to grow if it is given more time and space. As I really enjoyed watching this real-time growth over a long period of time, I decided to include it in the token using a URL parameter.

Edges

The nodes are connected by an edge, which is basically a line connecting one node to another. When the nodes are connected with lines, it reveals the relationship, passage of time, the process of growing. The aspect of showing the growth simulation as sequential and that it relies on previous state of the structure was important to me. So, colors are used to convey the growth process by shifting hue and saturation per generation. I like that the colors are not only used as decorative elements, but also to reveal a story about the growth as we see all over in nature - geologic formation, shellfish, trees, etc. It was not my goal to simulate the reality, but it is always an inspiration, and in this case, the colors of the coral reef was fascinating to me with its stark contrast and uncanny combinations of colors.

Colors are different based on seeds and generations. From left: Token #4, #17, and #23

NFT Features

Generative NFTs on the fxhash platform can have fxhash features, which show the characteristics of each token from the whole series and tell us how rare each variation is compared to all the others. In this particular project, I did not really go after the rarity value. The initial randomized parameters do affect the result a lot, but it takes search nodes to travel at least a few hundreds of times to be accepted by the structure and complete the growth. And for each search and accept iteration, the parameters are updated further. So, the stated rarity in the feature descriptions cannot fully convey or represent the interesting quality of the end result.

Reflections

This was my first generative NFT project so the whole process was an experiment for me. I have experience with code-based, parametric work mainly in the context of graphic design and typography, but releasing a generative system on an immutable blockchain was quite a different experience.

Normally, when working with such systems, I generate many variations with randomized parameters, and pick and choose the ones that I feel most interesting and engaging. I take a role of a creator as well as a curator. But with a generative NFT, at least on fxhash, I cannot pre-curate the result as the whole system relies on a transaction hash which is randomly assigned at the time of minting. So, I had to make sure that any variation that can be generated will look consistent as a whole, and at the same time, distinctive enough from each other. And of course, as a visual artifact, it should look appealing as well. Finding this balance was harder than I first thought. I had to smooth out my choice of parameters to make sure they all look good enough on average. It’s just a different way of working instead of hand-picking a few excellent variations. This also opened the door for me to explore further the parameter space that I didn’t know existed.

I am now asking myself how I should design my program in future projects - whether the hash should only affect the initial conditions or dictate the result. In this type of simulation-based work, the hash and the pre-defined features tell only a half of the story because many iterations of the simulation can result in unexpected and surprising places which cannot be examined by only looking at the initially defined features.

Also, the probability distribution I set in my code was not exactly reflected in the minted tokens. you have to toss a coin many times to get the expected probability distribution, but my project had the maximum edition of 128, and much less actually minted, and the distribution was not how I expected it to be. If it were fully minted and had a larger edition count, I think it would have made a difference. I think I also had a skewed perspective as I was testing thousands of times before minting, and more interesting iterations were registered in my head. That to me was a disappointment. After all, I put a lot of time and effort to make sure there will be a lot of unique variations. It is unfortunate that many of them will never see the light of day.

This project didn’t get minted as many times as I had hoped but the whole experience of learning and creating was very valuable for me. I very much enjoyed the whole process. If anything, NFT gave me motivation to push forward, finish and release the piece out in the world. If it weren’t for that, I would have just tinkered with it for some time and moved on. It could have been one of many incomplete code files in my computer. I would like to continue to experiment with this new method of creating and releasing my work.

Click each image below to see live version of the minted tokens.

Check the project page for instructions on keyboard interactions and URL parameter settings.

Neon

Mon, 14 Feb 2022 00:00:00 GMT

Click each image to view live version

Neon is a series of programmed motion loops released on the Tezos blockchain.

Structures

Fri, 21 Jan 2022 12:00:00 GMT

Structures is a series of programmed motion loops released on the Tezos blockchain. With this series, I am exploring rigid structures and their variations that I can create based on simple rules. Motion is essential in all of my work, Here, the only motion is colors changing. Any directional movement is an optical illusion. By using 1 dimensional change in 2 dimensional space, I can create many variations.

These loops are all created using my custom software, Modular Design System One. I originally developed the software to experiment with typography but it evolved into a tool that can handle abstract compositions as well. And that’s how I usually end up with software-based work. The code is malleable and it usually leads me to explore directions that I have not thought of in the beginning.

Click each image below to see the video on the Objkt website.

Modular Design System One

Tue, 04 Jan 2022 00:00:00 GMT

Modular Design System One (MDS1) is a web app built with JavaScript technologies (HTML5 Canvas, p5.js, react.js and material-ui) that enables me to quickly create design compositions.

I always wanted a more direct way of creating digital animation. The traditional approach is you define keyframes first, render, wait, and go back to more changes. While this approach has its own merit and gives you fine control, I wanted to “feel” the motion more directly while creating. With my software, I can make changes to the composition and typography while it is being played. There is also chance encounters with different color palettes. I have also created a grid-based font specifically designed for this project. It has variable font-like characteristics. Everything is controlled programmatically.

It is a work-in-progress and currently, only for my personal use. Check out examples of motion-typopgraphy compositions created with this software.

Below are process images and some of the early examples:

New Day

Sat, 01 Jan 2022 00:00:00 GMT

A motion poster that celebrates a new year. Created with Modular Design System One, a web-based custom software created with JavaScript.

This Will Be Over

Thu, 23 Dec 2021 00:00:00 GMT

A motion poster created with Modular Design System One, a web-based custom software that I made with JavaScript.

Time To Rise And Shine

Thu, 23 Dec 2021 00:00:00 GMT

A motion poster created with Modular Design System One, a web-based custom software that I made with JavaScript. You can purchase the poster as a limited edition digital NFT.

NYTimes Huawei

Tue, 02 Nov 2021 00:00:00 GMT

Animated illustration for the New York Times story, The Strange Saga of Huawei.

NYTimes Word Lists

Wed, 06 Oct 2021 00:00:00 GMT

Animated illustration for the New York Times story, How Word Lists Help — or Hurt — Crossword Puzzles.

I had a generous lead time for this assignment, so I thought I would develop a proper design system as the crossword puzzle itself is based on a solid grid system. I had a lot of fun developing this software which can take various numbers of columns and rows, word lists and different types of transition animation.

NYTimes Youtube Misinformation

Tue, 05 Oct 2021 00:00:00 GMT

Animated illustration for the New York Times story, Youtube’s Ban on Misinformation.

NYTimes Tech Profit

Wed, 10 Feb 2021 00:00:00 GMT

Animated illustration for the New York Times story, Profits Steer Our Digital Future.

Sayhye.net Website

Fri, 08 Jan 2021 00:00:00 GMT

Website design & development for sayhye.net

Developed with Gatsby.js for performance and easy data management. React-draggable package was used for the experimental yet functioning drag interface. It also works well with multi-touch mobile devices.

NYTimes Facebook Qanon

Mon, 21 Sep 2020 00:00:00 GMT

Animated illustration for the New York Times story, How Facebook Can Slow QAnon for Real.

NYTimes Internet Fracture

Sun, 23 Aug 2020 00:00:00 GMT

Animated illustration for the New York Times story, Trump’s Attacks on TikTok and WeChat Could Further Fracture the Internet.

I had a lot of fun creating this piece as I continued to experiment with Box2D physics simulation continuing from my previous assignment. This time, I added more interaction to literally express the fractured shapes.

NYTimes Apple WWDC

Tue, 23 Jun 2020 00:00:00 GMT

Animated illustration for the New York Times story, Apple Announces New Privacy Features.

I sampled the Apple logo to determine the size of each cube, and added sine-based animation on top. Created with Processing programming language.

NYTimes Apple Censorship

Wed, 17 Jun 2020 00:00:00 GMT

Animated illustration for the New York Times story, The Internet’s Most Censored Space. This work was selected in The Year in Illustration 2020 by the New York Times art directors.

As most of my NYT assignments have been, I only had a half-day to create this motion illustration from start to finish, but I wanted to try a different method of creating motion visually and conceptually. I used Processing programming lanuage and Box2D physics simulation library to create this piece. There is no keyframing and I recorded the mouse movement in real-time and later, edited in After Effects. By using the physics simulation, I was able to visually express the big ecosystem of App Store through abstract representation.

Featured

New York Times: The Year In Illustration, 2021

Happy Holidays

Thu, 12 Dec 2019 00:00:00 GMT

We wanted to take a traditional holiday card format to digital mobile experience. The original card was printed with Axidraw plotter using glitter pens. We loved the texture and how the light was reflected on the surface. The digital screen is already bright and sharp but it feels very different from ink on paper. Using motion and orientation sensors on mobile devices along with animation was how we decided to translate that physical experience to the web. Now, users can also customize the message and share with others all from their browser.

Visit Project

Collaboration with @justsayhye.

p5js Canvas Saver

Fri, 05 Jul 2019 00:00:00 GMT

This app is developed with Electronjs to export frames and video files out of p5js sketch. As a desktop app running a local server, you can render out unlimited number of frames as well as webm/mp4 videos all whthin the single app.

Github repo

Chillin’

Wed, 12 Jun 2019 00:00:00 GMT

Interactive web poster. You can write your message and save as image. Best viewed on mobile with motion sensor enabled.

https://exp.paperdove.com/chillin

Interview

p5js Website Showcase, 2019

Deepressed

Sun, 09 Jun 2019 00:00:00 GMT

An interactive web poster using motion sensor.

https://exp.paperdove.com/deepressed

Rabbit Hole

Mon, 03 Jun 2019 00:00:00 GMT

An interactive web poster inspired by Alice’s Adventures in Wonderland. Best viewed on mobile with motion sensor enabled.

https://exp.paperdove.com/rabbithole

왜/Why

Thu, 30 May 2019 00:00:00 GMT

An interactive web poster using a variable font concept. To maximize the expressiveness of the letter form, I created a custom software that controls a variable letter with a set of parameters. This web poster is the result of the experiment.

https://exp.paperdove.com/why

Selected Exhibitions

Motion Evolving, Beijing, China, 2023

Thick or Thin

Sun, 26 May 2019 00:00:00 GMT

Upside Down Inside Out

Sat, 25 May 2019 00:00:00 GMT

You can purchase the poster as a limited edition digital NFT.

Don’t Look Back

Thu, 23 May 2019 00:00:00 GMT

You can purchase the poster as a limited edition digital NFT.

Selected Exhibitions

Motion Evolving, Beijing, China, 2023

Right Here Right Now

Sun, 19 May 2019 00:00:00 GMT

You can purchase the poster as a limited edition digital NFT.

Midtown Mystery

Wed, 15 May 2019 00:00:00 GMT

Auto Mate Expo Sure

Tue, 14 May 2019 00:00:00 GMT

AIGA Frontier Reader

Wed, 01 May 2019 00:00:00 GMT

The typography + computer programming education research conducted with Prof. Jiwon Lee from Kookmin Univ. was published in AIGA Frontier Reader.

Power

Fri, 26 Apr 2019 00:00:00 GMT

You can purchase the poster as a limited edition digital NFT.

Boing Boring

Wed, 24 Apr 2019 00:00:00 GMT

Selected Exhibitions

Motion Evolving, Beijing, China, 2023

Los Angeles California

Tue, 12 Mar 2019 00:00:00 GMT

I always miss LA.

Hangeul Workshop

Fri, 19 Oct 2018 00:00:00 GMT

A two-day Hangeul Workshop was held at Maryland Institute College of Art. The workshop was designed for students who may or may not know Korean. The main goal was to introduce Hangeul as a modular design system.

Taught by Dae In Chung and Minsun Eo.

Seoul

Mon, 28 May 2018 00:00:00 GMT

This poster is my visual response to visiting Seoul in the Summer of 2018. It was very humid and hot. I was soaked wet with sweat everyday. I wanted to find some personal space and relax but everywhere I went, it was filled with people. And a lot of talking. We were talking about tensions and changes between South, North Korea and the United States. And the Summit. I felt energies, hopes and confusions. The city was constantly moving.

디자인캠프

Tue, 25 Jul 2017 00:00:00 GMT

2017년 디자인캠프는 7월 25일부터 한 주간 서울 필동문화예술공간 예술통에서 열렸다. 이혜진과 정대인은 9명의 멘티들과 ‘가내소공업’이라는 주제로 소형 플로터를 사용한 디자인 워크샵을 진행했다.

대량생산과 자동화에 이어 인공지능 등 세상의 변화를 겪고 있는 지금, 기계가 대체할 수 없는 디자이너의 역할은 무엇일까? 우리는 사람의 손길이 들어간 작업을 시도해보려고 한다. 작품의 질, 노동의 즐거움, 과정을 중시했던 윌리엄 모리스의 공예운동을 돌아보고, 기존의 작업방식에서 조금 벗어나 익숙하지 않고 비효율적인 다양한 방식을 이용한 새로운 탐구를 해본다. 우리는 꼭 효율성, 자동화, 시간/비용 절감, 생산성만을 최고의 목표로 삼아야 할까? 이를 통해 잃어버리는 것은 무엇인가?

작업목표

기계 도구를 활용하면서도 기계로 재현할 수 없는 디자이너의 손길이 닿은 작업을 해보자.
플로터라는 도구를 이용하여 디자인의 계획부터 생산까지 통합된 결과물을 만들어 본다.
익숙한 작업방식에서 벗어나 탐구하고 실험한다.
다양한 재료와 아날로그적인 기법을 이용해서 재료, 촉감 등의 미묘한 차이를 느껴본다.
사람들과 함께 논의를 통해서 작업을 발전시켜 나간다.

프로젝트

윌리엄 모리스는 “유용하거나 아름답다고 생각하는 것 외에는 당신의 집안에 들여 놓지 말아라,”고 했다. 내가 소유한 물건들 중에 가장 아끼는 것은 무엇이며, 쓸모없는 것은 무엇인지, 그 이유는 무엇인지 잠깐 생각해보자. 나에게, 그리고 다른 사람에게 쓸모있는 컨텐츠를 찾고, 그에 알맞는 아름다운 시각적 형태를 구성한다. 주어진 시간 내에 10개의 복제본을 생산한다.

AIGA Frontier Conference

Sat, 08 Oct 2016 00:00:00 GMT

Frontier is a conference concerned with alternative models for learning, radical pedagogy, utopianism, emerging technology and tools, global economics and design, virtual classrooms, invisible worlds, the growing relationship between design and tech, homogeneity in design, imagined scenarios, the role of critique in a post-internet culture, and the speculative future of graphic design education. — from the website

This presentation shares the process and result of a course titled Type+Media, which was offered in 2015. It was designed to teach typography and letterform design through computer programming. The presentation slides are available online.

LetterSeed 13 Journal

Fri, 15 Jul 2016 00:00:00 GMT

LetterSeed is the annual Journal of the Society of Korean Typography pusblished by Ahn Graphics. The research conducted with Prof. Jiwon Lee on typography and coding education was published in the LetterSeed 13: Technology and Typography.

Excerpts

The emergence of once innovative network and computers have become part of everyday media over the last 20 years. The graphic design field has also adapted to the change. Designers today can understand and use the characteristics of digital media with their own method. Typography, the field related to the shape of types and their arrangement, is expanding its domain of expression in a new way in a new media environment. Under these circumstances, it is necessary to adopt a different approach to typography, font design methodology, and education methodology. — p.37

Learning programming means the expansion of thought into unfamiliar domains, beyond mastering new tools. Many fields which have been regarded as the exclusive property of designers are being replaced with machines. The development of machines poses a challenge to designers. For designers, this is more than just a question of whether to learen programming, and it should encourage them to think about their roles once again. — p.52

Build The System

Mon, 15 Feb 2016 00:00:00 GMT

Designers often work with a system, whether it is a grid on a page or a typographical hierarchy. This poster is an exploration into building and revealing a system by gradually breaking away from it. In my own design practice, I start my process by building a system, usually by means of computer programming. Typography in this poster is first created with Processing programming language with a few simple rules. Changing a few parameters of the rules allows me to explore many forms it can create. But eventually, there comes a time when the rigid rules cannot apply any more, and that is when the instinct and manual interventions are needed.

Featured

The T Magazine, South Korea, 2017

Ekstasy Type Club

Wed, 01 Jul 2015 00:00:00 GMT

Ekstasy Type Club is an interactive typography application and installation. It explores the qualities of typography such as responsiveness and elasticity. Unlike type on paper, type on screen can interact with readers or users. A system with a few parameters generates a wide variety of letter forms. It constantly updates its shape and creates a unique spatial and temporal rhythm.

Technologies

Ekstasy Type Club is created with Processing programming language with the help of Geomerative library to get the outline data of the typefaces.

Selected Exhibitions

Type Directors Club 62 Exhibition, Cooper Union, New York, USA, 2016
The Society of Korean Typography Exhibition 10: Intersection, Seoul, South Korea, 2015

Selected Awards

Type Directors Club 62 Communication Design Award, New York, USA, 2016

Featured

Page Magazine, Germany, 2016
IdN Pick of the Month, 2015

Simple Investment Tracker

Fri, 01 Mar 2013 00:00:00 GMT

I wanted a simple way to track my investment. I don’t want to log in to my broker’s website every time I need to check the total balance. The search engine results distract my focus and show me too much information. I am not a day trader and never will be, but I want to make sure I’m heading in the right direction – that my investment is appreciating over the long-term, and also track the proper asset allocation once in a while.

Technologies

The software is developed with Processing, and it gets real-time and historical stock data from Yahoo.

New Year’s Greeting Cards

Wed, 02 Jan 2013 00:00:00 GMT

A traditional optical illusion technique meets coding, animation, and experimental typography. Thinking about the flood of new technologies in animation/design, and the vast possibilities of using these technologies together to create something new, we instead went back to an optical illusion technique that enabled us to create 8-frame animation cycle only using traditional mediums – paper and transparency film. This does not need any special devices/techniques to view animation other than the card itself.

Technologies

Adobe Illustrator, After Effects and Processing were used to aid the process, and quickly realize the ideas.

Credits

Design, animation and coding by Dae In Chung
Additional design by Hye Jin Lee

Hypervibes

Sun, 20 May 2007 00:00:00 GMT

Aluminum bars with resonators and motor driven fans. This is the vibraphone, a keyboard percussion instrument played with mallets and bows. With vibraphone sounds combined with abstract images, Hypervibes invites you to a 3 minute-long journey into your subconsciousness.

Animation by Dae In Chung
Music by Jay Garrett

Selected Screenings

Collectif Jeune Cinéma Jeune Public, Instants Chavirés, Montreuil, France, 2020
Mon Oeil Web Program, Centre Pompidou, Paris, France, 2019
Melbourne International Animation Festival, Melbourne, Australia, 2009
London International Film Festival, London, UK, 2009
Rome Film Festival, Rome, Italy, 2008
Oxford Film Festival, Oxford, MS, USA, 2008
Ann Arbor Film Festival, Ann Arbor, USA, 2008
CalArts Experimental Animation Showcase, Los Angeles, USA, 2008
Student Shorts Film Festival, Toronto, Canada, 2007
Experimental Film and Video Festival, Seoul, South Korea, 2007
Animation Block Party, New York, USA, 2007
Collectif Jeune Cinéma, Paris, France, 2007