Functionality Preserving Shape Style Transfer (SIGGRAPH ASIA 2016)

Functionality Preserving Shape Style Transfer (SIGGRAPH ASIA 2016)

living spaces are often populated with
sets of objects in a common style unfortunately while large databases of
man-made objects already exist sets of functionally diverse shapes in a
particular style are often not available we present an algorithm for synthesizing
shapes by transferring style between man-made objects with different
structure and functionality users specify the desired style via an
exemplar shape such as the orange dresser here and our method
automatically transfers its style to functionally different target objects
such as the cabinet coffee table side table and desk and similarly transfers the style of the
loveseat to the sofa and armchair and the style of the sugar pot to the rest of the coffee set our algorithm is motivated by observations
about human perception of style hinted at by art history literature same style objects frequently have similarly shaped elements as well as similar dominant
curves following these observations, we transfer
the exemplar style to a target shape via a sequence of element level
operations including element substitution curve based deformation element removal and addition unconstrained element edits can easily
break the functionality of the target shape for example transferring the style
of the examplar table to the target chair without constraints yields a
dysfunctional chair thus we search for compatible editing
operations constrained to preserve the target shape’s functionality to this end we introduce a cross-
structural element compatibility metric that estimates the impact of each
operation on the edited shape the metric is learned from training data
automatically generated from scenes containing coordinated, same style shapes
downloaded from online shape databases our compatibility metric is used in an
optimization procedure which searches for element level operations that
progressively increase style similarity to the exemplar while strictly maintaining
target functionality the operations we perform are designed
to preserve the shape connectivity specifically, substitution and addition
operations require alignment of slots or attachment areas between the newly added-in and pre-existing elements the alignment step seeks to both preserve output functionality and minimally change element shape our curve based element deformation
operations modify the feature and contour curves on the output model to be
more similar to their exemplar counterparts since style-function separation is
necessarily subjective to account for personal preferences we produce multiple alternative
solutions to the user presented in a gallery of results ranked according to
style similarity to the exemplar the degree of style similarity between
the output and the exemplar is affected by the structure of the original target
model users may provide a specific target
shape for style transfer or have our algorithm automatically recommend the most compatible target given a database of shapes in a specified class we evaluate our method by synthesizing
more than a hundred new shapes covering four broad categories of everyday
objects furniture, lamps, cutlery, and coffee and tea sets we validate the key properties of our method via three user studies the first user study evaluates style similarity between our output shapes and the exemplars we asked participants to
compare style similarity between an exemplar model and our output generated
from it, against other alternatives such as shapes found in style-coordinated
scenes that included the exemplar and were manually created by modelers participants perceived our synthesized
shapes as at least as similar style-wise to the exemplars as these manually created
shapes the second study evaluated the functionality of the output shapes against other alternatives such as the original ground truth target shapes our outputs are deemed functional at nearly the same rate as the ground truth shapes lastly we compared our compatibility
metric against prior work by Laga et al. participants prefer the correspondences computed using our compatibility metric at dramatically
higher rates we now show more style transfer results
based on our method thank you for your attention

3 thoughts on “Functionality Preserving Shape Style Transfer (SIGGRAPH ASIA 2016)

  1. Very amazing

    edit however I cannot listen to the narrator I can see his sweating virginal face sweating into the microphone kill me now Jesus Christ

Leave a Reply

Your email address will not be published. Required fields are marked *