Submission B Prototype and Report: the evolution

We are the generation who started to conquer new millennium. Living in new millennium benefits us to experience various the most advanced technologies. Rapid evolutions of technologies are faster than ever. 

'The porch dog' project
Top: Reality
Bottom: Virtual Reality

At first it was just reality. Then virtual reality was invented to advance our living in various ways. Now it is time for augmented reality. Augmented reality is mixture of virtual reality and our physical reality. Augmented reality can be thought as hybrid reality between our physical reality and virtual reality. When virtual reality technology was introduced, not many people would expect how this technology can advances our living. Now virtual reality technology is generalized and used in various field. For instance, we experienced how ‘Second Life’ can be used for student in architectural field. For now augmented reality is new. It has not been refined and still developing. However it cannot be denied that augmented reality has enormous potential as virtual reality has been benefiting human existence.

Augmented Reality

Since the augmented reality is new I initially planned to make the most out of this technology within its limit. My main focus was practicality of this augmented reality in architectural field. I have been experimenting how this can be used practically. To experiment practical side, I planned to adopt this technology to present architectural model to people. My initial concern was the loading time, which is one of the biggest down side of this augmented reality application program called ‘Junaio’. To shorten the loading time of assigned artifact, I planned to use 2D artifact such as plan drawing. It was later found that there is no significant difference in loading time between 2D and 3D artifact (both of these were loaded within 10 seconds in 3G network). This competitive loading time of 3D artifact was achieved by ‘Junaio’, which restricted 3D file size to be within 1mb of size (500kb for md2 file and 500kb for texture). So I changed my plan from using 2D artifact to 3D artifact. When heavier file such as 3D animation or complex model need to be used, these can be uploaded on Internet as ‘Junaio’ can assign link with tracking image.

When touching the 3D artifact, link of web page was appeared.
So heavier file can be uploaded on Internet and assign link to 'Junaio'. 

It was found during my development, the stability of ‘Junaio’ is influenced by various factors such as the speed of Internet and type of device. I used both 3G-network Wi-Fi connection and ADSL Wi-Fi connection. There was no sign of problem when the device was connected to 3G network or ADSL network. However when the Internet speed dropped to 2G-network, it needed huge patience for loading time (It took more than half of a minute to load 450kb 3D model).

Type of device was also the key factor for ‘Junaio’ presentation stability. When the device has better camera, ‘Junaio’ tend to work better. Iphone 4g worked best during my development. Android tab was found to be the weakest device to work with for ‘Junaio’. Iphone 4g recognized every time but more than 50% of time Android tab had trouble recognizing tracking image.

Even though I used 3D model for tracking image during my development, I had to transform this into 2D by taking photo and assigned as tracking image. My initial proposal was to use 3D model as tracking image and 2D detail drawing as assigned artifact to model. Since loading time was not significantly different between 2D and 3D artifacts, I changed my plan to adopt 3D artifact, which need to be within 500kb limit to be used in ‘Junaio’ channel.

Difference between 2D tracking image and 3D tracking image. 

3D tracking model tend to be affected heavily by light,  which is causing shading effect.

After this stage I experimented on finding the interrelationship of camera and 3D tracking model on recognition. It was found to be very difficult to achieve acceptable level of recognition (I defined more 80% success rate as acceptable level of recognition) with 3D model. Reason was since it was 3D model, it doesn’t have flat even image for camera to recognize. 3D model was heavily affected by light. Light cased different shading effect on 3D tracking model through the day and night (flection and contour of 3D tracking model cause heavy shading effect). 

Different light influence creating different shade effect. 'Junaio' is having hard time on recognition.

'Junaio' recognize 2D tracking image better even with different camera angle.

Sometimes ‘Junaio’ recognize the 3D model but more than half of time during the day ‘Junaio’ could not recognize 3D tracking model. Therefore to achieve my goal of practicality in real life I changed my plan to use 2D plan as tracking image. 2D artifact does not have flection or contour to cause shadow. 2D artifact is flat and even. It was the best option for me to achieve my goal. My goal is becoming clearer by this stage. I want ‘Junaio’ to be fast and accurate. If not perfect, as perfect as it can achieve.

Simple 3D artifact exmaple
image from 'Brooklyn Digital Foundry'

Complex 3D artifact example

Since complicated detailed 3D model (for example sky scrapper with every surface detail) cannot be achieved within 500kb of size. I found simpler and meaningful model would be better to use for my case.

Simple 3D wall model is appeared on bathroom floor plan

Up to this stage I have been using 3D timber wall artifact to pop up on floor plan so people can understand how the 2D floor plan will become 3D reality when it is made. Fabrication of 3D model will be more refined at the end. For now I used simple timber wall 3D artifact to find the optimizing condition for my ultimate presentation.

2D artifact is appeared on 3D tracking model

At this stage various things have been changed during developing sessions. Initially I tried to achieve 2D artifact assigned to 3D tracking model. My initial idea was expressed in my previous post in my blog. For now after evolution of my initial proposal, my key goal has become more specific. I want ‘Junaio’ to act faster and accurate within its limit. To achieve this I changed 3D tracking model to 2D tracking image (better accuracy). I changed 2D assigned artifact to 3D artifact, as there was not much difference in loading time.

Experimenting the interrelationship between 2D tracking image and assigned 3D artifact

Below is the kind of situation that I wish to achieve at this stage.

‘Architect A is presenting floor plan to client. Client looking at floor plan then check it again with ‘Junaio’ to see how this floor plan becomes 3D reality. This is the situation I am planning to achieve.’

From now I will concentrate working on achieving the most informative and effective 3D artifact within the limit of ‘Junaio’.




Useful reference link:
http://www.brooklynfoundry.com/index.php?p=bdf3&id=1783&gal=2161