The past few weeks I watched a lot of tutorials and did quite a few tests as well and this week I started to combine most of the interaction and test the behavior of the whole system. However I quickly noticed the problem that making few interactions work alone is a totally different thing that making everything work together as TouchDesigner use different module (TOP, SOP, DAT,…) to make things and therefore some of the module don’t work directly together and require to change big chunk of the network to achieve the goal. Also, the most important point is to make the installation real time and it requires the system to keep up with a high frame rate but adding interaction on top of each other reduces the speed of the system (from 60 to 35-40 currently and the videos, textures are not embedded). I will need to clean the network to make the debugging easier but also to try to optimize it in order to go back to 60 frames per second. Also, I’m currently looking into changing the beginning for a totally different method (using mostly TOP instead of SOP which uses GPU power instead of CPU in order to speed up the system).

Currently working:
Detection of a person at a defined distance that dissolves the images/videos into particule
Creation of the silhouette of each person detected
Attachment of a bubble to a person ( one only, but I tested two and it works if I duplicate part of the modules)
Deformation of the bubble controlled by music
Deformation of the bubble controlled by hand (not perfect, need debugging)

Next step:
Code/Network optimization
Change the beginning method to use GPU instead of CPU
Debugging of the deformation of the bubble controlled by hands

Next milestone:
Creation of the interaction particules/hand and its effect on the bubble shape
Improvement of the look of the bubble (texture, feedback) and particules

Additional milestones:
Make everything work for up to 2 persons at the same time