The Perspective Problem

Have you ever think your recent 3d face model from picture references is not good enough to be realistic??.....
How in heaven do I dare to even think in it?. Some models are perfect but the really truth is that you want to model in 3d a picture that is already deformed by camera lenses fool -no, you fool-

If I were you, I get back to my model and then with some additional references and enter to edit it with more care but damn perspective and lenses deforming this a lot you can't trust any photo now.....you think about it............................

Amoo Project

Article: references for build more references.

Amoo is a huge project that evolve from a basic to an advanced high quality rig for blender. It has been created with worth features and is unique in her kind due to it`s complexity and research.

The objective to make Amoo is to bring to virtual life a specific person with regard of all it`s features. So Amoo is the result of good research of the realistic.

So far this article brings you my point of view to discover some features I think other artists do not count or forget about nature of their 3d models.

Since Amoo was born, knowledge about how to model is changing over these last years so to be in touch Amoo has been evolving from different modeling methods.

Amoo Model History

Amoo Character in Cephis Project

This is a topic for non orthodox rigs documentation.



several stuffs to do

clear more and more topology

Low res mesh for testing animation in semifinal ............

face expresions and deforming muscles with drive keys

oh my head i been thinking lot..................

Amoo history (Model)

Article: Classic Pipeline to Create a Character

(See also Amoo for information about production).

In the beggining Amoo`s model wasn`t easy to make, it started (like someone would say) "fooling around" in 3D about August 2004 inside 3ds Max platform which hadn`t disctintive features by itself of actual 3d programs, and the sculpt module wasn`t developed yet. It wasn`t existed something to build somewhat decent but only my will to made her, and it took about three months and one more to give form to a mesh that started out with a cube. It would take pretty time to me to compile all images of this process that I`ve taken from the precept of genetic algorithms (just using my criteria of genetic algorithm), which I will explain it later in an interesting topic called `genetic algorithm pipeline`.

My concept model

Amoo version alpha

This pic shows Amoo`s model in her very firsts steps. After about 100 essays with my starting cube (about a year after my first session). Main modeling process featured in 3ds Max was to move down vert by vert. With the time this process was started with a speed increment of editions thanks to proportional edit feature.

Then, lot of time after improving my modeling performance I decided to get a more flexible 3d suite cause 3ds Max still had troubles in the process, time process, lot of things and mainly for troubles in data format so I wanted to get a tool that help me up instead screw me out. I found blender and I could export it out and started to edit it.

Mastering 3ds Max... but with an exotic weird topology....how in heaven could I even think I was master in Max???

Amoo alpha model exported to blender and afterwards it was refinated

In process of learn again 3d modeling I was still modeling in 3ds max.

Then when I was ready to go in blender my Amoo model begun to have improved features

So after about 3 or 4 years from the start, Amoo beta version appeared. Face topology from this beta version (actually with 936 vertices -cheap resolution-)

I started then with blender particle system.

Render and blur with photoshop

Refining the model in blender. Improving T pose for better model deformation. Applying vertex paint and improving surfaces. But still this model was a beta version

Rigging the model using my rig criteria...some of these rig features are good.

Applying hair mesh (not a render). Model still in beta version.

Accumulative surfaces by vertex and edge increments (for me, before sculptris voxel method and multires sculpt, this is one of the best modeling methods of a specific surface via subsurf mesh, )

Amoo v6 with improvements in proportions. Rig and model were still incompatibles

After lot of improvements this model ironically would decrease its quality with the time, I thought was bad topology that for god sake I couldn't to improve.....what could I do?. Retopology tools........that`s what I did, I got some topology references for this purpose and then I got a smooth model.......but.........

Amoo v9 topology, result of copying topology from different artists. So naturally, this version was a failed version, note the lack of vertices

Improving topology, some mad issues about this damn brand topology.....shooooot!!!!

OK, this is a try of beauty mask.

Amoo v9 with some bad issues in uv mapping

Render with Blender 2.49, there are weird issues in hair mesh for this blender version

Good News (Dec 17, 2010):

Amoo v10.03

After 1 year later a totally rebuilted mesh of Amoo. These are the improvements

• Several patches of good edge loops over and over new retopo surfaces
• Decent medium resolution mesh
• Topology ready for animation with fixed features

I`ll post some more details............wait..................

More Updates

Amoo version 12

Amoo with SSS shader

Index

Topology Workshop: The Pole Poly Solver

Ooooray!!!!!!

Here it is!

The pole poly solver: Note poly samples with six external edges in each. (vertices have been enhanced) 

I will post an extremely interesting modeling topic about the quad flow that builds any organic model...............................but first i start with building this page.............................................................................................

It is established that quad polys build a very well topology but just if it has a good flow of quad polygons running around over a well determined surface.........how is it possible? and how to perform it well?...I just heard about that the key are the poles...........

So before me there's a topic about this explaining here (a very comprehensive organic modeling external site).

You must see this for understand what I'm going to explain..........

So, It is a must to avoid triangles in a mesh for a best performance. I really don`t care really what is the E pole and the N pole but I know that the pole poly solver can be the solution for most of topology problems for a topology form only by quads, for creating better edge loops, for increasing and decreasing resolution and for create useful edge loops flows. ..............

This topic is made in parallel with recomended face topology topic cause they have lot of similarities but specialities, so to give a better explanations I should made them together but apart to a better not redundant comprehension.

Ok this is my main character and it`s completely made from vertice increments. That means that it doesn`t have to do with edge loops rules or face topology, it has only refinated surfaces and few edge loops to give a smooth appeareance but with an advantage of be imperfect and it like me a lot..........

I hate the perfect topology I think is a must to be few imperfect with everything, perfection doesn`t exist but equilibrium does. So when I finish Amoo, I will give an imperfect touch.

Recomended Face Topology

Over these days I`ll post some explanations about my huge request of good face topology, I'm still in some tests about all stuff of this................

You could need to see the topology workshop article to dig more about topology.

Mission: I'm going to make it possible, right?

The main mission is to have a decent topology with the maximun edgeloops possible of any organic model to perform its best animation

To transform this topology:

Into this one (click to enlarge): 



Topology Tools

Ok, this part is a must to animate a face at its best, so the only way to be in touch is to practice until understand the flow of edges (Holy phew!, this job is terrible and you could spent lot of time deciphering topology science, crap!. It must have to exist a some better pipeline to reach this up in a not so boring way indeed......). But in the meantime I just know that just with practice the outcome will just be more easy to reach.

Yeah, First I've seen some face topology tutorials even a new face topology tutorial by Angela Guenette but I didn`t like at all, some details were putting in practice........then here`s my essay of topology number 60:

So Once my original model has been finished I have to start with topology tools to paint a flow of edges (a hard work to do) and then I will have:

Some boring tests over the face.....some references were taken from tutorials, the last from durian team explaining face topology but I don`t sure about all this stuff so I start to search my personal good topology

Checking for normals to check some sinks or topology mistakes

Muscle enhanced in green is my seek of principal feature that is seen in all faces around world...

But it is wrong so I start again....-Several hours to achieve-

Finally this muscle is splitted in two. These muscles are important and I think they need to be constructed well.

New model finished with 1889 vertices. Also I think I've defined the most important muscles without references needed, this time with some references about muscles and intuition (actually I spent about a year to get a good topology cause I've been copying to other models but it doesn`t work so each model is different of each other)

But wait........some there were some troubles at sight, finally I think were solved

Oh yeah!, this model now has edgeloops around mouth, orbitals......

Here we can see some edgeloops resulting from solving some troubling areas from triangles.

A very hard to achieve edgeloop, just the edgeloops I needed

Here: the result of my tries

Modeling Techniques: Benchmark

Incredibly topology for my model results in awesome similarities with sintel model by Angela Guenette. Sintel model has a beautiful topology with an exact number of faces for detail, moreover high detailed topology for sintel was made for production from Durian Project

Here, sintel (fantasy character) with perfect edgeloops, but perhaps lack of character details and pretty smooth and perfect.

Here perhaps for my likes, lack of an edgeloop for lower jaw edge....this model I will edited with permission to give birth Moora (character).

You have to see Amoo's Topology History topic .............. (under construction).

............