TracyChen

Feb 4

HK paper draw a line in one dimension → 3d model pose: Actually, I have seen some software like that, but haven’t really had the chance to use one.

Silhouette → 3d pose: Probably can be used in some transformation of 2d animation to 3d animation, especially just for blocking poses.

Physik: move model according to the path of root/pivot point, great for composition/ storyboarding?? Like for pre-production of animations or movies.

Reconstruction of motion capture: FABRIK, MGIK

Motion capture dancer’s motion and to generate new dance→ The way they mapped everything is impressive but just dance to express motion might be too subjective?

Feb 6

Disney – Precomputed Environment Semantics for Contact rich character animation: have been implemented greatly into games, less time cost on creating animation transition with giving players more sense of freedom?

AIST paper – Controlled optimization: ease the work of artist a lot that system can simulate/give a more natural motion animation, but it definitely needs a big database

Weight painting spline interface: a good tool for the artist to use tho, but like Muhammud said, it was not any sense that to show the time cost of artist weight-painting in the traditional way comparing to he using the new tool (how familiar he is to the tool, etc)

Differential blending for expressive sketch base posing: more control is given compared to the paper shown last class, much better weight painting and skinning

La Faim(Hunger), interesting!

Skin-cage, have used a lot back to college using Advanced Skeleton. It actually got decent result and artists are given a lot of control during rigging and skinning, but definitely some unclear process going under that requires iterations and much more polishing work.

Feb 11

ToonSynth presentation by Amy Wu: Interesting that this is a quite new paper, so according to Amy, the researchers are trying to map animation from hand-drawn 2d animation to skeleton then they can reuse the same motion into different kinds of style of animations. It sounds really useful to me this research, but as Amy mentioned there might be some limits on it so… Hope it got better further development in the future.

Jaw animation presentation by Kang: Actual human jaw motion is actually quite complex than what nowadays animation does. The paper simulates the human jaw movement with a “3d envelope” to create the rig so they can achieve a different kind of jaw movements that the pivot point of the jaw is actually moving all the time in real life. It’s quite interesting, they are kind of customized different jaw rig to a different person. It might be too many efforts if they try to make a specific rig for every character. But if they can sort of come out with a universal-wise skeleton will increase the accessibility of it as an application in the real world.

Language-Driven 3D scene Processing, presented by Iris: Efficiency in scene modeling, for quick scene generation and edit. A rapid prototype of getting a general sense of what the environment will look like.

About this paper: Romero, Javier, Dimitrios Tzionas, and Michael J. Black. “Embodied hands: Modeling and capturing hands and bodies together.” ACM Transactions on Graphics (TOG) 36, no. 6 (2017): 245.

Body scan and hand scan, if you can capture enough data of human body/ hands so, in the end, you can have your low-res model to a high-res model

They did clothing scan in 2017 and have a way to separate the cloth from the body

Their goal is to generate a 3d model with pose through 3d image, their way of retrieving poses from their database is actually pretty good. So you can kind of generate 3d scene of real life with less effort and can be a great use in the game industry. I think it could be implemented in one of my classmate’s semester project, which is about streaming improvisation performance on twitch.

Feb 20

2002 robust treatment of collisions, contact, and friction for cloth animation – spring-mass cloth

Disney paper, adjusting the parameter to give different cloth simulations different kind of feeling, they use Maya nCloth and Disney simulator. Is it just a test about what’s the best values of parameters that those two simulators can get?

2011 Sensitive Couture of Interactive Garment Editing and Modeling

Goal: to make simulator fast enough to do iteration at the same time, UI

Cloth pattern and sort them together to fit the character.

ARCSim, more wrinkle more triangle, re-mesh, can be used for animation cuz a bit slow

Spring-mass, re-meshing algorithm, re-generate meshes

Near-exhaustive Precomputation of secondary cloth effect

Store cloth state, simulate them, find similar state, jump to it.

Simulate the cloth and collecting the data

precomputaition->real-time

Simulated Yarn, slow but interesting

Cloth in Game is mostly constrain-based

Feb 27

-An Advection-Reflection Solver for Detail-preserving Fluid Simulation

-Perceptual Evaluation of fluid Simulation Methods

Eulerian:

MP marker-particle, particles for rendering usage

LS Level Set

Mix:

FLIP Fluid implicit Particle

APIC Affine Particle-in-cell

Lagrangian:

WCSPH Weakly Compressible SPH

IISPH Implicit Incompressible SPH

SPH

A classic of smoke control: Fluid Control Using the Adjoint Method, really interesting

Control the shape of the fluid/smoke,

Semi-lagrangian: advection

Remind me of a VFX artist called Randy Cano, who has done a lot of smoke/fluid/deformable body control, kind of shaping the particles into the shape of people/animal and drop? Them for different effect or impact

High map/depth map to fluid simulation, can get different angles of the “wave”. Not purely visual but still interactable with objects.

Mar 27

We talked about the St Venant Kirchhoff model and the video shows a lot of fracture simulation. It reminds me of Voronoi fracture that Houdini and C4D usually use to do fractures. I kind of wonder what’s the difference between this two method? I guess Voronoi is kind of pre-fractured the object and just let set active after it is set to active? And SVK is more real-time simulating in a more close way of what’s happening in a deformable body?

Active Animations of Reduced Deformable Models with Environment Interactions:

– The deformable body that’s alive is a really interesting point for animating the deformable body, which can save a lot of efforts for artists to animate creatures like starfish or some tofu who accidentally has its own mind. With forces inside the mesh and drag forces from outside? Most of the results are quite interesting and convincing to me.

Stable Neo-Hookean model bones and with triangle mesh as skin- convincing results can be used in skinning for animation? Volume conservation changes also affect a lot of the final result.

Projective Dynamics: putting constraints on strain limiting and volume conservation, interesting!

Apr 1

Fluxed Animation Boundary Method- presented by Hong Zhu

Sample the velocity of the material, not the boundary to get the details of splash/spray of the waves/water

Not gravity but using the force field, small forces to drag particle.

Kind of faking the interaction of the boat and the water, just get the motion/vel/pos of the boat and then combine the boat animation and water simulation together.

Wave-based sound synthesis paper

Rendering sound while doing the simulating

Finite-Difference Time-Domain Method: Updating gird pressure (air is fluid).

Parallel-in-Time synthesis: chopped up the accelerations into disjoint time chunks with overlap time added at the tending.

Using sound shaders for different things (just like materials for different objects), and use their framework to do the sound rendering while they have the full access to the data of the simulation.

It’s really cool, but I just wonder does it have anything to do with different simulation methods?

Apr 3

Example-based Synthesis of Stylish face Animation:

Interesting sounds like what many machine learning studies have the same purpose on this sort of recreating art style on video or paintings.

Facial Retargeting with Automatic Range of Motion Alignment:

From face motion capture converting values to the Blendshape “scalar”, then apply the scalars to the Blendshape. Different ways of interpolation(linear/RBFS/…) for the “scalar”.

Enriching Facial Blendhsape Rigs w/ Physical Simulation:

A good way of giving a feeling of physics do things on the skin, but the bouncing part of the skin seem a little bit too much. But maybe it just needs more adjustment on the painting portion or parameters. Yes and as Figo said, there should be other forces like inner stress from the inside of the mouth (muscles?), so for the lips, they should have different performances.

Apr 8

Meekyoung SIGGRAPH2017: Deformable body, interesting.

Computational Bodybuilding: Automatically-based Modeling of Human Bodies:

How they control the portion of fat and body? Is there any way to get the proportion of a fat person’s muscle ratio comparing to a slim person.

Apr 10

Character Simulation and Control

Boston Dynamics

Pitch Controller, Hopping Height Controller, Velocity Controller

1991 On the Run (Kangaroos)

1994 Atlanta in Motion(Runner)

2007 SIMBICON

Feedback error learning to bring the stiffness down