Assignment 1: The Documentation

Data Characteristics

This data sculpture is a visualisation of the data of the July 2008 portion of my diary. My data sets look at the characteristics of each diary entry within this timeframe. The attributes of the data sets I have used are Purpose (one of Pondering, Recounting or Venting), average number of words per paragraph, average number of words per sentence and average number of sentences per paragraph. What I have found through analysis of this data is that:

1. I write to ponder a lot more often than I do to recount or vent.
2. My writing pattern changes depending on purpose:
a. When venting, everything becomes shorter. I use fewer words per sentence, fewer words per paragraph, and subsequently fewer sentences per paragraph.
b. Paragraphs tend to be longer when I recount. The reason for this being that I tend to forget to form paragraphs as I am too caught up in transcribing dialogue.
c. Paragraphs also tend to be long for pondering as I question various events.

Research & Data Sources

For this project, I had to create my own data sets using my diary as a resource. To do this, I had to first digitise all 28 entries, then I used the online text analysis tool [topicalizer] to gather a whole lot of data about the entries. I had to sift through it all and picked the attributes I saw as most relevant. Then I went through categorising the entries according to Purpose, to form my second data set.

Data sets found here: The Data Sets and Quick Visualisations.pdf

Design Rationale

Trends and patterns aside, there were two main points I wanted to convey through my sculpture. Firstly, that all three purposes fit together as a whole to form a part of my diary, and a part of myself. Secondly, that my writing pattern changes depending on purpose (the main trend I found).

An important data characteristic I wanted to preserve in the sculpture was the sequentiality of diary entries, whilst it was not absolutely necessary to preserve the actual dates. This offers further insight, or more context and meaning, as this shows the order of the entries (with their purposes) and also where they stand in relation to each other, allowing the user to gain an understanding of the events that might have taken place. For example, they might see that there was a time in July where I spent several entries on end just pondering, or maybe there was a long time where I was not angry enough to vent. It allows them to build a back-story to what they're seeing.

My final form was inspired by the Venn diagram, used in statistics where circles (groups) intersect each other. I took the form of the overlap between three evenly arrayed circles and used these as the templates for the three interlocking stacks.

Data Mapping Technique

The idea is that each piece would represent a single diary entry, and that piece would be shaped by the data attributes I have mentioned above. This would make each entry unique.

I had to test out many different methods of using the figures in the dataset in order to find the best combination of curves, curvature, angles, often changing placement of mid-points and intersecting points. I had to test out the same method using the both the smallest and largest figures, since often a method that worked for one would not work for the other.

The parameters I settled on for determining the shape is shown below.

In the middle, where the pieces all overlap and fit together, the thickness is determined by the average number of sentences per paragraph. I liked the idea of some “hidden” information that is revealed as you pull the sculpture apart.

The original idea is that each curve would be just extruded straight, creating a staggered profile for each stack, emphasising the differences between the lengths and widths of each piece, and their corresponding data attributes. Upon modelling this in RhinoCAD, however, I found that this did not achieve the effect I desired. It looked too incoherent, and it was difficult to compare stacks and pieces.

Using the same curves and spacing between them, I modified the design slightly. Rather than straight extruding, I chose to loft between two curves - the one of the current entry and that of the one before it. As each block is now shaped by two curves, there is a much smoother transition from one piece to another, making the sculpture more coherent as a whole. This approach also proved much better in showing the change between one entry and the ones before and after it, as users can now physically see the difference through the degree of slope between the top and bottom, left and right.

Whilst I could have made the three extruded blocks solid, I decided that it would have been more meaningful to have each "entry" piece separate and unique. In this way, there are a number of different ways users can choose to "read" the sculpture. They can choose to look at each entry individually; each purpose individually; compare two purposes together; compare the whole sculpture together as a whole; or they can choose any combination of entries to compare.

Labels were designed to be fully integrated into the sculpture. I had a single additional block (which fits in with the rest of the sculpture) with the text engraved into it to describe how the dimensions correspond to the data attributes. I also had a top plate to name the subject of the visualisation and label the three categories. My name was also designed to be engraved on one of the bottom pieces.

Technical Challenges, Difficulties and Constraints

I encountered very little difficulties in creating the model, having had previous experience using Rhino. Once I got my head around issues such as evenly spacing out components, getting parts perfectly intersecting/overlapping and how to go about spacing out curves in accordance to my data, there were no real dramas, except for the tedium of doing each piece individually.

A much better way of doing it, as Gabe showed me after the model was completed, would have been by using Grasshopper, since all my pieces were parameter driven. The reason I did not use Grasshopper was because that plug-in was new to me and I did not fully understand it. Having watched a few more demonstrations since, I understand it enough to use this technique next time I need to do something like this.

The first “difficulty” I came across was compromising between two constraints – size and cost. My intention was to make the model as large as possible, within the printable area of the rapid prototyper. Since I had many parts that supposedly fit together, I did not want them to be too small and fragile. However, because of the disparity between figures in my data sets, I found I had both very thin pieces and very large solid blocks. We found that the cost of printing was a lot more than previously anticipated. Approximately $12.50 for a 3x3x 3cm cube. In the end, I kept the size of the model the same; I just went through hollowing out all the large pieces to save as much as I could.

Most of my problems occurred after the production of the model.

The first and most critical problem was the text labels not turning out from the printing. I had made the text as large as I could considering the (in some cases, limited) surface area, and as deep as I could considering the thickness of the pieces. Then I checked with Gabe who reckoned it should have been okay. In the end, though, it seemed not. So I really had no way of identifying the pieces other than consulting the 3D model to figure out the order of the pieces. Luckily the pieces were designed to be unique and could only fit into one spot. I’m glad I made all the joints different sizes.

Next, the joints I designed, for the pieces to “plug in” to each other, did not fit. I had inbuilt some leeway in the 3D model before I had it printed, but it was not enough. The printer added more thickness than I had anticipated. My only solution was to sand down the edges using a plastics file. This worked well for me.

It seemed that the RP added more thickness to some pieces, but not enough to others. Even though it was fine in the 3D model, some pieces were a whole several millimetres off. However, there was really no way around it. Some pieces fit perfectly the way they are, others were way off. I could not have known how much leeway to give. Due to this, the joints in the middle fit a little too tightly and do not allow the stacks to slide out as easily as anticipated. At least this way the sculpture holds itself together well.

This piece on the right is meant to sit flat against the table but on top of the stack on the left. It was way off…

Time was a major constraint. Had I had more of it, I would have been able to test out one or two pieces to get an idea of how to deal with it, and the text issue.

Breakages were another difficulty. Even after being baked, the pieces were pretty brittle. There were one or two pieces that did break and crumble a bit. I salvaged it as best I could by using superglue as modelling putty. After that I sealed all the pieces with a surface primer, which hardened it up a little and slowed the disintegration process.

The major issue was the lack of text. Without the labels, the whole thing is completely abstract with no discernible meaning whatsoever. I needed a way to convey at least some, if not all, of the information I set out to do. I tested out several solutions to the problem, which failed:

1. Stencilling with paint. Leaked around the edges.
2. Custom water-slide decals. Did not stick to the material, and clear coat eats into the ink making it unreadable.

3. Stencilling with pencil. Handwritten and too hard making them all consistent.

My final solution was to colour-code the stacks so that at the very least the Purpose categories can be distinguished. This was, even though there are no labels telling you the order the pieces go in, at least you can tell to which group it belonged, which narrows down the possibilities. It could be part of the fun of it, trying to figure out the order – like a jigsaw puzzle.

Despite the surface primer, the material did soak up quite a bit of the paint, making it a bit splotchy. I did several coats to get the colour as solid as I could. Then I gave it all a clear top coat to seal in the paint.

The last thing left to do was the labelling. I used Rhino-rendered curves and Photoshop to create a “legend” to contain all the information I lost in printing. I shaped it and coloured it in a way that is “visually consistent” with the rest of the sculpture. It was the best I could do to “design around the problem”.