Old Planning Notes

I can't believe I forgot to post these. No wonder it seems like I didn't do any work for a really long time. Here are the finalized design plans for the plug-in, detailing the pipeline from start to finish.

Where to Go Next

We are now going to develop a "sketch-based interface for polygonal modeling of stylized characters using an armature pipeline". Oh man, now doesn't that important?

This is the closest paper to my project:

Sketch-based Virtual Human Modelling and Animation
Chen Mao, Sheng Feng Qin, and David Wright

However, I have decided to use an armature-based approach akin to real sculpting technique.

The overall contour of the human body is defined by two types of bones: long bones (eg - limb bones, spine, the femur, the radius, etc.) and flat bones (eg - protective bones, skull, pelvis, etc.). Using a 2D sketching interface, the artist draws an armature, using lines ending in circles for long bones and circles and cubes for flat bones.

The long bones are then converted into a Maya skeleton. The flat bones are converted into I-forgot-what-they-are-called-but-they-are-like-deformers-but-rather-keep-things-from-deforming. The bones are then used to automatically generate a mesh - for example, the circle defining the ribcage will result in the chest mesh being roughly the same shape.

Based on the resulting skeleton, the plug-in adds curve pairs (tetraCurves from the previous blog entries) which give artists recursive control over figure obesity and fitness. Artists can alter the entire figure or selected body parts.

The plug-in will also automatically pelt map (or Ptex) the model for easy texturing. It comes with a skin shader, too.

The good part about this is that it allows the artist to create non-standard figures. I was thinking for another project even animating them procedurally.

These are papers I will use to help me. This list is not filled yet.:

Sketching Interfaces:

Amit Shesh and Baoquan Chen
SMARTPAPER: An Interactive and User Friendly
Sketching System

Automatic Meshing:

Rigging:

Jianhui Zhao1, Ling Li, and Kwoh Chee Keong
3D Posture Reconstruction and Human Animation from 2D Feature Points

Andrei Sharf, Thomas Lewiner, Ariel Shamir, and Leif Kobbelt
On-the-fly Curve-skeleton Computation for 3D Shapes

Curve Manipulation:

Pelt Mapping:

Texturing and Shading:


This is related to stuff I probably won't implement but would be cool additions.

Clothing:

Philippe Decaudin, Dan Julius, Jamie Wither, Laurence Boissieux, Alla Sheffer, Marie-Paule Cani1,2,3
Virtual Garments: A Fully Geometric Approach for Clothing Design

Ambition will get the better of me.

Planning Notes II

I realized my blogs are very much not "weekly." I just kind of post whenever I've finished something.

Here are some more moderately intelligible notes. They describe in detail the class structure of the plug-in - like what node possesses what and which methods call which inside each other.


I have to work on the soft selection and mesh manipulation algorithm, but after that, I'll be ready to put it all to computer (given that I have, already put it all to paper).

Here are the tasks I am giving myself, about one or two a week:

1. Do the plug-in writing exercise from my Maya plug-in writing class.
2. Create a functional CurvePair.
3. Make a TetraCurve where muscle curves affect fat curves.
   
4. Make the TetraCurve manipulate the mesh.
5. Model the figure. Going to start off with just the human male for now, since it's from the Goldfinger book.
6. Set TetraCurves up on the figure.
7. Create GUI.
8. Allow for importing and exporting of skeletons and meshes.

These are things I would like to add if I have the time.

1. Save and load from text files.
2. Gender toggles.
3. Viewer toggles (such as "hide curves").

Well, then. To work.

Planning Notes

I've read a lot more than I've needed to, I think, but it's given me a better grasp on currently available technology for human modeling: volumetric methods, template methods. Hyewon Seo, et al. even did "An Automatic Modeling of Human Bodies from Sizing Parameters" for a clothing story.

Most use CESAR data, so I've decided to take my project down a slightly different path. The thing is, CESAR scan data produces ridiculously high resolution models. More importantly, the resulting figures cannot be easily stylized (pushed beyond the available data set). My editable curves methods aims to allow artists to create stylized models, to fit provided skeletons, and accomodate lower resolutions.

These are my planning notes. They are relatively intelligible planning notes. The next step is to get them off the paper and into the proposal.

Sorry, 3DS Max users, but I've settled on writing on Maya plug-in. The GUI (and as much as possible) will be done with Maya's Python Commands. However, the data structures (such as bodyPart and triCurve) will be written in C++.

I have also broken down the figure into seventeen overlapping regions and demarcated important landmarks - specifically using terms from figure drawing classes.