Data-Driven Computer Drawing Tools

El meetup va a hacerse en la oficina de LambdaClass (https://lambdaclass.com) que está a unas cuadras del obelisco, en Tucumán 840 Piso 5 Departamento C. Si llegan y está cerrado, manden un mensaje a https://t.me/papersweloveba

La charla la va a dar Emmanuel Iarussi (https://twitter.com/emmaiarussi).

Data-Driven Computer Drawing Tools

Humans excel at perceiving 3D objects from line drawings, one of the earliest forms of visual depiction. Therefore, freehand sketches are still the preferred way for artists and designers to express and communicate shape without needing to effectively materialize a real object. Unlike humans, computers struggle to interpret a 2D designs as a complex abstraction of our 3D world. Compared to photographs, dealing with sketches is a much more challenging task: line drawings lack key shape cues like shading and texture, projections are imprecise, and the shapes are often composed by several sketchy lines. Employing graphics, deep learning and interactive techniques, this talk is a walk through cutting-edge data-driven methods applied to novel computer design tools.

Papers:

• CoreCavity
  ** http://visualcomputing.ist.ac.at/publications/2018/CoreCavity/
  ** https://www.youtube.com/watch?v=UZ9hR9qLMAY
• FlexMaps
  ** http://vcg.isti.cnr.it/Publications/2018/MPIPMCB18/
  ** https://www.youtube.com/watch?v=73VvpKlrrQU
• SketchZooms
  ** https://emmanueliarussi.github.io/index_sz.html
  ** https://www.youtube.com/watch?v=kZSMaERkBs4

Recomendamos un poco de conocimiento previo en Deep Learning.
Pueden ver la charla de nuestro anterior orador Enzo Ferrante como introducción https://tv.vera.com.uy/video/54701

Si tienen más tiempo y ganas, Emmanuel recomienda el curso de Andrew Glassner. https://www.youtube.com/watch?v=r0Ogt-q956I