General Info

Workshop Leaders:  Alessandro Mintrone, Alessio Erioli

Software: Unity, Tensorflow [optional: possession of a 3D modelling software capable of exporting .OBJ or .FBX files] – Installation specifics will be given to the participants prior to the workshop beginning

Previous knowledge required: Basic digital 3D modelling skills

Delivery Mode: Remote (Zoom)

Maximum # of Attendees: 12

Schedule: 11:30-14:00, 15:00-18:30 CET.

Brief

This workshop explores iterative assemblages of parts driven by an implementation of the Wave Function Collapse algorithm, combined with machine learning techniques operating at the decisional level, focusing on a higher degree of control over the generation and deployment of articulated and heterogeneous spatial conditions. Participants will find out how the implementation of an AI system can lead to a cognitive understanding and deployment of emergent qualities in the assemblage. The workshop will explore ways to short-circuit the analysis and the design phases, training an agency that learns from experience and then develops an inherent sensibility.

Working in Unity, participants will develop procedural assemblages using the Wave Function Collapse algorithm, designing the geometry and connectivity properties of a collection of “tiles”. Utilizing machine learning techniques with Tensorflow, participants will then train a neural network to learn how the assemblage can be steered towards controlled target conditions. Participants will also learn how to keep track of the training process with Tensorboard, as well as bespoke visualization techniques in order to produce video and still images that showcase the spatial qualities of the outcomes.

Daily schedule

The schedule is split into 2 weekends (Saturday & Sunday). After the first weekend, an exercise will be assigned for the following weekend. The exercise should take no more than 2-3 hours to complete at its basic level.

Weekend 01

• workshop introduction: theoretical background, scope and goal
• introduction to Unity and WFC algorithm
• tiles design: geometry and connectivity
• tutored work: practice with WFC

• introduction to reinforcement learning
• training goals
• training a model with Tensorflow
• checking results with TensorBoard
• tutored work: training exercises

Day 03 – Saturday Feb 20

• pin-up review of results
• tutored work (project)
• visualization techniques in Unity

Day 04 – Sunday Feb 21

• tutored work (project)
• final presentation

Fees

EARLY BIRD

Educational*: 250 €

Professionals: 400 €

REGULAR

Educational*: 300 €

Professionals: 450 €

(*) Educational fee is eligible for students (secondary school to PhD) and graduates up to one year after the degree. To qualify (or inquiry for it) you should provide proof of status at this address: alessandro.mintrone5[at]gmail.com

Deadlines

• Early Bird: January 17
• Regular: February 10

You can freely cancel your application with a full refund before 23:59 CET -February 10th; past that moment no refund requests will be accepted.