In the current version of the gallery, I have built a networking system which will allow all of the players to be in the same virtual space simultaneously. At the moment, the system only works over a locally created network, but in the future I will expand it so that we can bring in remote artists to work with the students.

These are the VR headsets that were used in the exhibition. Because the exhibition wasn't very long, I didn't experience any issues with battery life. In the future, I may implement chargers into the stands for the headsets. The challenge here is the headsets don't have any downtime so they won't have time to charge. In reality, the only solution is to have a rotating set of headsets so some are charging while others are used.

These images are renders of the gallery. These have much higher visual fidelity than the recording on the previous page because these were rendered in octane. Each one of these frames took around 2 minutes to render. You can imagine how much of a challnage it is to render 60 of these images each second and still make them realistic. This is where baking comes in. The lighting so the shadows and bounce lighting (global illumination) are all rendered beforehand, baked into the textures and applied to the model in unity. This means the Oculus isn't doing any light calculations. Unfortunately this means that I cannot change the lighting at runtime and interactable objects don't have realtime lighting.

This clip is a recording directly off the Oculus VR headset. This is what the viewers of the gallery saw on the exhibition night. I curated the work for this exhibition, but in the future I wan't the students to be involved. The current version of the gallery allows students to pick up prints and hang them on the wall using their virtual hands.

As well as this machine needing to comply to strict safety requirements, be low cost, fit around the current machines in the workshop, and many more constraints, it needs to be easy to assemble. This is because the machine is going to be assembled in sessions I run with students. This also presents the challenge of accuracy. This is a precision piece of equipment where many parts need to be aligned with sub millimetre precision. To allow the students to assemble the machine and to make it accurate, I have designed everything to be adjustable. If you can't make it precise, make it adjustable. The whole machine is constructed using t-slot aluminium extrusions so the parallelism of each beam can be finely tuned after the students have assembled it.

Here are some of the other iterations of the machine. The very early designs for this machine were done in Cinema 4D. I now create all of the iterations in Fusion 360. Throughout this project I have learnt an incredible amount about design and engineering even though i haven't built the machine. I have been learning how to use CAD as a prototyping tool. This allows me to only spend money and time on the most highly refined design that has had every nut and bolt modelled and considered.

In my quest to transfer a linear actuation over a long flexible strand, I experimented with hydraulics. In this test I tried to use two actuators connected together to transfer the motion from one to the next. I experience many issues with this that made it clear it wouldn't be feasible.

Here are some earlier iterations of the mechanism. Here I was experimenting with TPU to create live pivots in the mechanism. The whole mechanism couldn't be printed from TPU because it needed some rigid parts. To print just the pivots from TPU and the rest from PLA I manually edited the GCODE to add pauses at the right moment for a material change. I struggled with adhesion between the two materials but tuning of the temperatures, speeds and flow produced good results.

This is an even earlier iteration. The pivot system in this version is fundamentally different to subsequent versions. It uses a double pivot system. This adds complexity and resulted in very un-linear motion of the arm.

The main challenge in this mechanism is to turn a single linear motion into the pivoting of many arms placed in a circle. Another avenue i explored was helical gears. If you place two helical gears at right angles, turning one will still turn the other. The issue with this design was how much force each pivot requires. Any pressure on one of the gears made it impossible to turn the central one. I suspect moving to machined metal gears would have helped.

This earlier version of the mechanism was much easier to assemble but had high resistance in its movement. That is what led to the use of bearings in each pivot. What this did show was the success of the pivot design and its ability to open and close the tessellation.

To attach the paper to the mechanism, a thin strip of double-sided tape was applied to each arm of the mechanism. This worked but over time the tape lost grip and it was also very hard to stick each pivot in to the paper. For the next iteration the paper will be mechanically fixed to the mechanism. The arms also should be fixed to the paper first then the core should be fixed to the arms as a separate sub-assembly.

To create highly accurate folds matching the desired pattern, the paper is scored using a vinyl cutter. This machine is usually used for cutting out vinyl lettering but I have re-purposed it to create scores. As i move to larger and larger score patterns, I had issues with the tracking on this machine which I eventually resolved by hand programming the move commands.

To allow me to experiment more with the tessellation without having to fold each test, I programmed a simulation algorithm. It is programmed in C# using unity. After adjusting the parameters and coming to a design I am happy with, I can export the score pattern and 3D model.

This animation shows the design for the core which supports the larger paper tessellation. In order to better use a sheet of acrylic, the arms are separated from the core. This adds complexity to the assembly process but the reduction in material wastage make up for it.

From my experience using double-sided tape to adhere the mechanism to the paper, I designed these paper mount brackets. As the paper is quite thick, the holes easily hold the weight. The weight is also very evenly distributed among the many arms. This issue was it was very slow to thread all the holes and screw in each bolt.

As well as a wall mounted version, I made a much larger suspended version. Making a tessellation this large presented many challenges. It needed to be scored in two half's then joined together. It also needed to be made out of a thicker paper so it could support its own weight without ripping.

This is the fully laser-cut core that holds the paper in the desired shape. It also allows for the mounting of a heatsink and high-powered LED. The bent stainless steel pipe is threaded into the heatsink and is mounted to a custom base containing the LED driver and allows for precise levelling.

These clips show the pumping self-priming and remote sensing system.

Here are a selection of renders created through the long proccess of getting to the final render.

Here is a pre-vis animation of the animation applied to the vegitaion.

This is the original painting, A View of Tivoli Cascade by Jean Joseph Xavier Bidauld.

For this shot, I built a transparent water tank from an old Ikea storage box. I filled it with water and added dye to make it look blue. I built an underwater scene, experimented with the lighting and photographed a car being dropped in.

For this shot, I designed and 3D printed a cable car pylon and then built a mountain from black and white sand. I built a secondary mauntain, photographed it and composed all the elements together to create the illusion of a mountain range.

This is how i achieved those images on the previos page. It was great fun experimenting with the differnt lighting conditions and seeing how each completely changed the image.

These images are from the two studio shoots I used my laser cut parisian-inspired models in.

This is the first model I built for this project and what really sparked my love for build an image completely from scratch.

I created these long exposure images using an LED light stick i built and programmed. The conecpt is pretty simple. I wrote some Python code to horizontally scroll through the pixels of an image and display on an LED strip. Now by moving the led strip in a long exposure you can re-create the original image.

As well as starting to design the embedded software to allow remote updates and app control, I designed many PCB's to facilitate the hundreds of LED's.

Here is my experimental blow moulding setup. The purpose is to work out how hard it is to create a uniform dome to be used as the diffuser.

I tested a few different possible diffuser materials. This one was the best as it created a uniform gradient even when the LED's were far apart.

These are some of the CAD iterations created in Fusion 360.

These clips show the random field generation algorithm. This allows the game to be a completely endless runner where you heard sheep from one field to the next.

This project is what really got me familiar with Unity and Object Oriented Programming in C#. These clips show some of the custom UI behaviours I created.

I would take them out to the park, crash them, fix them and repeat. Each time making improvements.

These images show the led grid and the chains of TLC5940s.