even though TensorFlow and Fabric are conceptually similar things, they still do drastically different jobs.
I've created a simple TensorFlow binding for a project I'm building, here - https://github.com/FrozenKiwi/FabricTensorflow
The binding is extremely simple, it only allows for loading pre-trained models and evaluating them. The binding uses a home-grown extension of kl2edk that automates the building of KL extensions (similar to Kludge - I created it before Kludge and still find it markedly easier to use).
Feel free to use/abuse/wha-eva.
I'd like to share my implementation of the Centripetal Catmull-Rom algorithm.
drawDebug: display the curve segments and the control points.
parameterization: Define the knot parameterization of the curve, where 0=Uniform, 0.5=Centripetal, and 1=Chordal.
details: Define the smoothness for each segments.
nPoints: The number of output points interpolated on the curve.
isClosed: Define if the curve is closed for the computation.
inPoints: the control points that the curve will pass through.
While I initially needed a more simple solver for only 5 control points, I felt like digging into Blocks so that it could work with any number of control points. Thanks to @malbrecht for his "Break / Exit a ForLoop" trick!
I tried to keep it 100% nodal at the exception of a 2d to 1d array conversion node which I couldn't figure out how to do with Blocks yet.
I must say, for an ex-ICE user like me, Canvas has come a long way since it first released. The gap between what it is today, and what we could do in ICE really is getting smaller and smaller.
For more information about Centripetal Catmull-Rom Algorithm:
I joined the preset with 2 simple scenes.
As a softimage user, I've always been frustrated by the missing the "smooth" option from the
I'm not sure why the first time I looked at it, it seemed to be more complicated than that, but as I had some times this week end I gave it a go, and it was in the end pretty straight forward.
This is before
I'll see if I can make a version for the built-in ones and post it here.
We worked with our friend Tom Sporer to turn this production tool around in just a few days. Check out the article here
©Copyright 2017 Fabric Software Inc. All rights reserved. | Privacy