Wednesday, March 12, 2014
London
Its been quite some time since I posted last. After my wonderful, almost 4 years at R&H, I took some time off, learned how to surf and just enjoy the summer. A little work showed up here and there, but now I am happy to say I am at Double Negative in London!!! So far it has been about a month and it has been great. I will be in the FX department here, working in Houdini and Mantra. More to come on the Lego project which is sort of my time off project I have been kicking around. Can't wait to put all the new things I am learning to use on it!
Friday, May 3, 2013
Lego Mini Container Truck Project : FX Test
Started back up on working on an FX test pass for the mct truck and wanted to share a little bit of my progress.
In the last couple of weeks, I had to switch a few things around. When I first assembled the truck, I did it all in sop level, all under one geo node. Probably not the most effect way of doing it, as when I started implementing the simulation pulling the correct transforms where getting a little dicey. As DOPs works better with obj level animation, what I ended up doing is reconstructing this section in obj level and then feeding each piece individuality to DOPs. This made the animation and velocity calcs much cleaner. All in all I combined animating the entire truck, with animating individual blocks and then turning on and off the RBD keyframe active on those individual blocks, so the simulation takes over at the point where the blocks are not intersecting each other. I rendered out a rough flipbook to better illustrate. Have a look, I think its starting to come together
I've got some time now due to the rough spot in the industry at the moment. Other then keeping busy with this and other self projects, I will also be enjoying some of the Fun-employment. Stay tuned, and hopefully something falls through.
In the last couple of weeks, I had to switch a few things around. When I first assembled the truck, I did it all in sop level, all under one geo node. Probably not the most effect way of doing it, as when I started implementing the simulation pulling the correct transforms where getting a little dicey. As DOPs works better with obj level animation, what I ended up doing is reconstructing this section in obj level and then feeding each piece individuality to DOPs. This made the animation and velocity calcs much cleaner. All in all I combined animating the entire truck, with animating individual blocks and then turning on and off the RBD keyframe active on those individual blocks, so the simulation takes over at the point where the blocks are not intersecting each other. I rendered out a rough flipbook to better illustrate. Have a look, I think its starting to come together
I've got some time now due to the rough spot in the industry at the moment. Other then keeping busy with this and other self projects, I will also be enjoying some of the Fun-employment. Stay tuned, and hopefully something falls through.
Saturday, February 9, 2013
Collision Workflow
As some of the blocks are getting more and more complex, I've decided to go with a different workflow for creating the collision geometry. After diving into some RBD Solver refreshment, I recalled the Laser Scan option. Knowing this is a heavy calculation when done on simulation time, I wanted to find a way to use something like the Laser Scan attribute, to create the parent geometry for the collisions. So after some digging I discovered the isoOffset SOP which will create geometry based on an object sign distance flied.
By calculating something similar to the dimensions of the original block it gave me a better representation of itself. I then wrote and read the geometry back in, scaled it a little, and then ray-soped it to get it closer to the original. The last step was to poly-reduce the vert count, this gave me an optimal amount of verts with out losing to much contour. The result is something which resembles a triangular mesh, but overall holds more surface area. I think it will be a better collision body then what I first tested, have a look.
Gonna start testing a few of these in a smaller scale simulations, to determine sim times. More to come.
By calculating something similar to the dimensions of the original block it gave me a better representation of itself. I then wrote and read the geometry back in, scaled it a little, and then ray-soped it to get it closer to the original. The last step was to poly-reduce the vert count, this gave me an optimal amount of verts with out losing to much contour. The result is something which resembles a triangular mesh, but overall holds more surface area. I think it will be a better collision body then what I first tested, have a look.
Gonna start testing a few of these in a smaller scale simulations, to determine sim times. More to come.
Subscribe to:
Posts (Atom)
