Showing posts with label Simulation. Show all posts
Showing posts with label Simulation. Show all posts

Friday, September 7, 2012

Soft Feet + {

I did a little more work on the Beast's Soft Feet and thought I'd post the result. The goal was to make the nail extending and retracting more true to a realistic cat paw.

  • The model needed to be changed to adapt to the improved rig. 
  • The nail needed to pivot around the toe joint accurately. 
  • The surrounding skin needed to react to the nail.

This was my reference for the nail pivot:

The nail pivot was a simple matter of adding a bone or 2 to the end of the nail to use to pivot the nail. The trick was getting the skin to wrap around the nail as it extends and retracts. For this I decided nCloth was the best bet to get a realistic effect.


Skin Cinch {

I didn't want to simulate the entire mesh, since I wanted the skin around the nail to have really accurate collision (high settings). So I created a cinch around the nail. The cinch is an nCloth with dynamic constraints for maintaining transforms and attracting toward the nail. The nail is a passive collider for the skin cinch, so it never penetrates the nail itself. I then use the cinch as an influence on the actual mesh. This avoids simulating the entire mesh and focuses the simulation on the area that matters.

Comment below to ask questions or give feedback! Hope this helps!

-K

Tuesday, July 17, 2012

"Gauntlet" {


An old woodsman possesses an ancient gauntlet with the power to summon armor and weapons at will. This weapon is put to the test as he discovers that his gauntlet is not the only ancient magic in these woods.

Created by Jeff Belgum, Katlan Merrill, and Ben Smith Sound design by Jesse Rope Score by Anthony Ferraro and Zach Johnston

Finally it's done! Here is my thesis short film, entitled "Gauntlet". I had the pleasure of working with two other talented friends of mine, Jeff Belgum and Ben Smith. We all put a couple years of our lives into this project, and many late nights. We wanted to push the envelope of student films with stunning visuals and a dynamic action sequence. Even if we produced nothing, it would have been worth the first-hand knowledge of going through the process of making an animated film from start to finish.

Role {

My roles in the creation of this film were primarily in the pre-production section. I was the Lead Designer, Lead Modeler, and Lead Rigger, so I designed, modeled, and rigged the two characters, the "Hero" and the "Beast", in addition to helping with story, layout, simulations, hair, rendering, and animating. I also managed and automated parts of the pipeline to speed up workflow with MEL scripts, vrscene files, and program integration.

Beast {


The Beast proved to be the most complex and time-consuming rig. I got a bit creative with the control scheme, which made my rig unique and intuitive. The back controls are all exposed well above the actual model, allowing the animator to always select them from any camera angle. The custom control spline shapes were created in Illustrator, and each describe the action they perform. My goal with the rigs was to create the most automated, easily animated rigs possible, while maintaining a good amount of control. This led me to the decision, for example, to make the tail fully simulated, requiring no animation at all to look smooth and realistic.

I created my own muscle system for deformation at particular joints, like the shoulders, hips, and biceps. I had experimented with maya muscle, and due to the buggy and uncontrollable nature of them, decided to make my own. It is a simple system, taking 2 points as attachments, and based on the distance between them, bulges or stretches the NURB muscle object using user-created SDKs. Complete with a basic MEL interface, I was able to get decent deformation at all points with full customization and no bugs.

Hero {


The Hero rig gave me the challenge of incorporating Motion Capture into the pipeline. Using Maya's HIK auto-rigging features, I was able to get a decent rig that was Mocap compatible with relative ease. Given the time, I would have created my own rig and used HIK as an input. HIK is horrible to animate with and I don't recommend it. The challenge in this rig came with the armor (modeled by Ben Smith). I didn't have the processing power or time to simulate each piece of armor. This route is also bug-prone and takes a good amount of time to perfect. I instead decided to get creative and make the armor conform to the pose of the Hero. With a variety of tools, including SDKs, I was able to make the armor slide, pivot, and interlock to match all extreme poses. I was sure to add some controls to certain troublesome regions in case of any unforeseen tweaking that might be needed on a per-shot basis. All in all, the armor worked out quite well. Given more time, I would have added a bit of constrained jiggle simulation on top of them just to sell the effect.

Please comment if you'd like me to break down any part of the pipeline! I'd love to hear your input and questions!


-K

Monday, June 25, 2012

Soft Feet {



I'm currently working on adding details to a quadruped rig from my senior thesis film "Gauntlet". I didn't get as much time as I wanted rigging him, so I decided I'd enhance a few parts. One part that bothered me about this character was his feet's interaction with the ground. So I set out to fix it by adding deformation to the foot to simulate the pads squishing on the ground. The animator can get the foot relatively realistic and with minimal clipping, but having a touch of deformation sells the impact.

The simple approach would have been to just create some NURB deformers and give the animator control of the squash and stretch of the foot. However this would add even more work to the poor animator who is tasked with a walk cycle of a quadruped. So I decided an automated/simulated solution was ideal in this scenario.

Now I needed to avoid the messiness and expense of making the entire body into an nCloth. The body mesh is quite high poly, and I only wanted to simulate the feet. Thus, I decided to only simulate the pads of the foot and use those as influences on the main mesh. The pads are extremely low poly, so they simulate almost real-time. And they also more correctly describe the foot of a lion (my inspiration for the feet). The one trick here is to make sure you check on "Use Components" in the skinCluster of the main mesh being influenced. This means it will use the components of the influence object and not just its transform.

Here is a quick video showing the deformation effect:




And here's one on a more bumpy surface (it is simulated after all):


So, to replicate, follow these steps: