Finally the BlinkAdapt ShockCube project has come to a relative close, well the company and product presentation was conducted today and the official play test is next week. Our presentation was pretty strong with a good mix of hard financial fact and visual dressing, also the promo film went down pretty well with the audience.

Anyway here it is, still a work in progress and will be updated with new footage from the focus group/play tests next week. You can also watch it in HD by clicking the HD link in the video (it will take you to the official vimeo page).

The common approach with custom touch table design is to modify the popular PS3eye cameras by removing the IR filter and swapping it for a natural daylight filter thus turning it into a basic IR camera. However for reasons unknown (we have been told our table is too large) this solution didn’t really yield a camera anywhere near responsive enough for what we required, so this forced us to look elsewhere. One offering by PeauProductions seemed to be the best option, an expertly modified PS3eye camera with external lenses and a number of other fixes for the various problems that come with modifying the camera yourself. Chiefly among which was the well know focusing issue where the camera would lose it’s ability to focus when zoomed out, a huge problem for us with a 1m² surface to cover.

Though not a cheap solution (£180 after tax) it was our only real option for a reliable touch system.

Right: Our first unsuccessful attempt. Left: Shiny new expert version

Due to the unforeseen factor of how the compliant layer (silicon) would effect the final image we have had to rethink our game aesthetic on a fundamental level. As you may have seen in previous posts the necessary silicon layer has produced a somewhat textured surface, this in combination with the projector we have available to us has forced a rethink away from the light and subtle designs we have currently. The reasons being that light colours tend to accentuate silicon texture meaning a clean white background is no longer suitable and small details are somewhat lost by the projector so these will have to be traded for bolder shapes.

We are currently testing a combination of backgrounds and design features to get the best out of the equipment we have and should have something more appropriate soon. Below are some features of the first design draft which we are now revising.

First Design Draft

Finding an ab toner that contains a circuit board easy enough for us to hack turned out to be a little hit and miss, fortunately the second ab toner we purchased was useable. The circuit board for the ab toner works with simple buttons to trigger the shocks, which has enabled us to attach relays to these buttons. When we activate a relay with the arduino board (Controlled via Flash) it switches the button and thus increases the voltage.

So the basic process is

Flash > Arduino > Relay > Ab toner

The ab toners purchased run off mains power, we have however hooked them up to battery packs to grant portability to the system and will hopefully make the shocks a little safer for ShockCube players. Another added safety measure we have incorporated are large emergency power buttons, which when activated shut power off to the ab toner completely. This is achieved mechanically with buttons that simply break the connection between ab toner battery pack.

Finally we now have a responsive and accurate compliant layer with minimal image distortion. The last part is fairly import to the overall quality of the final projected image, something we learnt the hard way from our previous surface attempt. This was covered in our last post and as you can see the silicon produced a streaking effect that when dry distorted our projection quite badly. So we stripped and cleaned the surface back down to bare perspex and applied a new silicon layer but this time using a more absorbent cotton roller. When dry this produced a much more uniform texture that we could live with, however using a cotton roller with sticky silicon resulted in a few stray fibers but the intense light tends to hide them.

White Vellum is our projection layer of choice due to a good balance of clarity and thickness, in the end Rosco Grey was just too thick to produce decent blobs. The vellum also came in a handy A0 size which almost matched the width of our table minus a few centimeters. With the projection surface finally down we sealed off the sides with the last IR ribbon section and used electrical tape to block out all interfering IR light. Furthermore we were at last able to mount the camera and mirror into fixed positions beneath the screen due to the surface layer being finished.

As things now stand we have a working touch table minus screen calibration, which will be our next task.

Our table as it stands now (projecting white)

Projector, IR Camera and Mirror all fixed into their final positions

The IR Camera can capture the whole surface at a distance of 1 meter

So the second attempt to create a successful compliant surface. The method used this time was again silicon based but instead home brewed from an intoxicating mix of builders silicon and white spirit. These 2 ingredients need to be carefully mixed in a 1/1 ratio then sponge rollered onto the perspex in multiple coats (3/4 recommended with around 10 hours drying time in between each coat) A very handy video tutorial by Dean Segovis actually called for a much faster drying spirit “zylol” but the closest UK equivalent we could find was White Spirit.

Silicon material tests

Coming out of the last experiment with a somewhat negative attitude we decided against jumping straight in and coating the entire 1m² area. Instead 3 separate mini tests were conducted using various surface layer materials that have been used in the touch table community. Rosco Grey, Vellum tracing paper and normal tracing paper were given 3 coats of the custom silicon mix, which took around 2 days to complete. Then on the third day using a small scrap of the newly silicon coated Rosco Grey, success! Then again with the tracing paper, success! Full “blob” detection with minimal finger pressure on a combined perspex, compliant and surfaced layered sandwich.

In theory all that needs to be done now is scale up the silcon layer to our requirements (which is currently in progress) and we should have a rather large and working touch table. However as this project has taught us many times, theory and practice are two distinctly different concepts.

Silicon ripples drying from the first of 3 separate coats

Now that we have a stable frame the art of combining projection and compliant layer could be attempted. The job of a projection layer is fairly simple, to capture the projected light (image) from below inside a special material applied to the perspex instead of letting it pass through. For this surface we used a material know as “Rosco Grey”, which as the name suggests produces a slightly duller image but comes recommended by Jeff Han himself.

Sulky Solvy applied to Perspex

The Compliant layer however is a slightly more problematic concept. When you introduce a projection layer the effect of you finger on the perspex and thus the IR light refracting inside it is dampened. This results in a poor “blob” production and makes it difficult for the computer to recognise finger contact. So what has been dubbed as a “Compliant Surface” must be introduced underneath the projection layer to amplify finger contact by increasing the pressure with minimal user effort. Many compliant solutions have been dreamt up and the first one we attempted was called “Sulky Solvy”, a silicon based American product used in material stitching. So before the live test our set up was as follows…
<----- Rosco Grey ----->
<----- Sulky Solvy ----->
<----- Perspex ----->

It failed. No blobs, nothing. The failure was probably due to the size of our table as we are quickly realising through the online communities that this is one of the largest touch tables using the FTIR method ever attempted. With 2 weeks to go this is a minor disaster but there are a few more methods to try yet. However on a good note our new custom modified £150 PS3eye camera from the US works like a dream. More on that to come.

Bare Perspex, cleaned and ready

Silicon based Sulky Solvy applied

Only 3 IR strips were used for quick testing

After getting to grips with the Glue library we have been attempting to create our own Custom ab toner AS3 class to work with the games. To do this we access a number of the arduino digital pins and then switch the power to these on and off to switch the relays which then control the ab toner shocks. We have a number of timers which run all of the relays to increase the shocks at the correct levels for each specific game and player.

Below you can find a full document overview of all the arduino work for ShockCube:

Arduino Overview Documentation

After finally receiving the bulk of our required components from across the globe build time had come. After I drew up the initial table concepts last month we have been discussing the best possible method for constructing ShockCube and decided there were several conditions that needed to be met…

So after taking these requirements into consideration we sketched out an idea for a lightweight skeleton frame strengthened by struts to which external aesthetically pleasing panels could be attached. The beauty of this design is that it accommodates for our fairly limited wood working skills by hiding the frame and allowing us to experiment over the final weeks with various panels types (wood or metal). Then as a team of men we decided that it would definitely work and set off to B&Q. The wood used is standard construction wood cut to our previously discussed dimensions of 950mm x 900mm x 870mm, which should give 4 players ample space to operate with an audience.

ShockCube as you see it below took about 3 days to complete and is very much a work in progress. Construction has been halted for now as another specialist IR camera is in transit from the US after the moderate failure of the first one.

3 L brackets were used to secure each cube corner

IR ribbon sunk into the metal frame

Adjustable bracket for the projector

The access panel will be built around here

State of the ShockCube table now.

So today we finally received the most crucial component in our system, the infra red LED ribbon and paid £120 for the privilege. The ribbon is a specialist component from environmentallights.com which was modified from normal, coloured LED ribbons intended for shop display signs.
Basically this flexible ribbon of IR LED’s when wrapped around the edges of a perspex sheet and shone inwards creates an effect know as ‘total internal refraction’, which floods the internal space of the perspex with infra red light.

IR ribbon around the perspex

Obviously this light spectrum is invisible to human eyes, so instead we utilise an infra-red camera, which when directed at the perspex ignores all visible light and sees only a beautiful glowing shard of infra-red. Now due to the light refraction created inside the perspex any objects when pressed against it, such as the finger, visibly effects the refracted light creating a disturbance that can be tracked via computer software. These disturbances are then delivered to authoring programs like Adobe Flash as co-ordinates, forming the basis for a working touch table!

The IR disturbances can be tracked

Regarding the camera, our already burgeoning £800 budget didn’t allow for an off the shelf solution so we simply opted for modifying a PS3eye webcam. This solution is widely used for custom touch tables by removing the IR filter from the camera and replacing it with a visible light filter.

As for the perspex we had that laser cut to a 900mm x 950mm specification for and extortionate £100.

900mm x 950mm