Page 1 of 4 123 ... LastLast
Results 1 to 20 of 78

Thread: Notes on 3D human modeling

  1. #1

    Notes on 3D human modeling

    I added some new experiment to the thread http://blenderartists.org/forum/show...2&postcount=53

    Original post

    During my recent adventure to learn 3D human body modeling, I seen a total absence (on the web) of tutorials trying to teach the relations between the principles of human anatomy and 3D polygonal modeling rules. I never found any tutorial, trying to show how to visualize the human body to approach correctly polygonal modeling.
    All that I found was really high poly or really low poly modeling tutorials, both leading to awesome and sometime amazing results, but none making me understand what exactly lies beneath.
    Maybe I did not search enough, but I started searching more than three years ago.
    However I do not mean that the tutorials, currently available on the web, are not usefull. Two of the best tutorials I found, for example, the first for High Poly and the second for Low Poly modeling, are Modeling Joan of Arc by Michel Roger and LowPoly Character Modelling by Ben Mathis. And, in general, www.3dtotal.com hosts good examples and totorials about many aspects of 3D modeling.
    What I mean is that, before reading a tutorial such those I linked above, a newbie, like me, should learn human anatomy bases, from the point of view of 3D modeling techniques.
    For example, a really good tutorial about face modeling, following this philosophy, giving a synthetical view of a base human face mesh, is hosted here in Elysiun, A Better Face Tutorial V 1.1(UPDATE).
    My effort is focused to obtain a base mesh for another complex region of the human body: the connection between the arm and the shoulder/breast.
    I am a comics style drawer, so I used my experience in human anatomy to solve the problem.
    This experiment was for my personal use, but I decided to post it for give help to other 3D modeling newbies, like me, and to see if any expert 3D modeler can give his opinion or some help to better understand and eventually correct my deductions.
    In traditional arts, it exist the convincement (absolutelly correct ) that an newbie artist should learn base concepts before doing more complex works. They start drawing schematic puppets to understand proportions of the body and so on.
    The same approach should be adopted to learn 3D human modeling. Even if you come from a painting experience, like me. That is becouse 3D modeling is completly different from drawing with a pencil or a paintbrush.

    So lets start.

    First of all, I show you schematically how really are muscles around the shoulders region. My drawing skills coming in help.



    As you can see, the great chest strap muscle (magenta) is connected to the upper arm bone, exactly under the deltoid muscle (green).
    When we tense our shoulder muscles and release the great chest strap muscle, we actually raise the arm. As you can see in the next image.



    This group of muscles is not the only one involved in this movement, but it is the group of the more visible muscles.

    The real problem is that we cannot simply connect the arm to the rest of the body as if it is a tube, like in this example.



    This is because the presence of the great chest strap muscle (with the great dorsal muscle) create a concave region between torso and arm, so called armpit.

    There is no online tutorial that helped me to understand the solution to this problem. Many of them simply resolve it, but do not explain WHY and HOW. And often, the more detailed they are and the technique, they use, is less understandable.
    In traditional arts, teachers teach to synthesize, in mind, the shapes we see in real world. And I think it should be the same for a 3D artist.

    So I started creating 3 meshes. For the neck and the breast I used a cylinder with 12 horizzontal subdivision, while the arm base mesh was a cylinder with 8 subdivisions. In Blender use Space Bar->Add->Mesh->Cylinder, and give the number of subdivisions as the parameter for the number of Vertices.
    The meshes created are: one for the breast (green region), one for the neck (yellow region) and one for the arm (blue region). I also deleted back faces of all the meshes and left side faces of the breast and neck, to work better and see the interested mesh regions without obstacles.

    I tryed 2 different methods for connecting them. The first mesh (upper in the next image) is the simplest one; the second one uses a single quadrangle (red polygon) as a bridge for connecting the three cylinders.

    This is a rappresentation of how to connect the simplest one.



    And this is the result.



    Than (for a reason of appeal??? ) I have mirrored the second mesh and started to work with it.
    I added one vertical subdivision to the central strip of polygons of neck (yellow region) and breast (green region), 4 horizontal subdivision to the breast (green region) and doing some manipulation (just move, scale or rotate), to obtain a draft model of the upper torso and arm.



    And, the same result obtained with the mesh without red quadrangle.



    I decided to go on with the red quadrangle mesh, and I deleted some faces: the red quadrangle, and 2 polygons strips (the more internal) of the shoulder until i reach the begining of the bicep.
    So, probably, the use of the red quadrangle was not necessary, but remember that I was experimenting.



    Than I selected 2 edges and create a face, that connects them.




    I subdivided the newly created face (use Knife or Loop Knife in the edge menů).



    The reason for this modifications is better seen in the pencil sketches of muscles I done. With this steps we are going to extends the breast muscle to reach the shoulder muscle, as shown in the image below.




    In the next step I subdivided vertically the left most faces strip of the breast (green region).



    Also I deleted a face in the middle of the bicep and connected the right most faces of the bicep with the left most strip of the breast (green region).



    I selected the 2 faces, as in the images below, and subdivided them.




    Than I closed the hole in the mesh, as shown in the images below.






    Try to activate Subsurf button and you should obtain something like this. I think it is a good mesh, reflecting real muscle fibres structure of a human body.




    For the hole, remaining in the middle of the shoulder, you can choose to simply merge the 2 free points of the blue region with the 2 free points of the yellow and green regions.

    Anyway, I will try it, and I will go on later with this tutorial.

    Any suggestion, critique, or phrase like «I don't understand anything!» is welcome.

    Thanks.

    To be continued...



  2. #2
    Member cartoonM!ke's Avatar
    Join Date
    May 2005
    Location
    Walnut Creek CA, USA
    Posts
    5
    wonderful tutorial!

    I've been modeling the human form for two years now and you put into words & pictures what it took me that time to learn. I wish that this tutorial was around back then

    Also, following the direction of the muscles will come in handy when animating the figure.

    Thanks again and I can't wait till you get to the back muscles!

    m!ke



  3. #3
    Originally Posted by cartoonM!ke
    I've been modeling the human form for two years now and you put into words & pictures what it took me that time to learn. I wish that this tutorial was around back then
    I had the same problem and I spent the same time. This is the reason I write it.
    And I am still not completly sure that this is the right one solution. :0

    Originally Posted by cartoonM!ke
    Also, following the direction of the muscles will come in handy when animating the figure.
    Yes, I thought to animation when the idea, of this solution, born in my mind.

    Originally Posted by cartoonM!ke
    Thanks again and I can't wait till you get to the back muscles!

    m!ke
    Infact, the next step will be exactly back muscles and fine modeling of the piece I done with this tutorial. When I will complete that part, the rest of the body would be easier.

    Thanks a lot for your appreciation.



  4. #4
    Member womball's Avatar
    Join Date
    Oct 2005
    Location
    Wilmington Delaware
    Posts
    1,016
    An issue with flow in the pectoral muscle, the muscle itself does not flow in a curve way which you indicated, however it flows in a straight line directly under the deltoid. Due to this whenever the arm moves outward, the pectoral muscle is stretched. In yours if you animated the arm moving upwards, the pecs wouldn't stretch as expected. I may draw some pictures too, if you would like. The attachment to the arm isn't quite as wide.



  5. #5
    Originally Posted by womball
    An issue with flow in the pectoral muscle, the muscle itself does not flow in a curve way which you indicated, however it flows in a straight line directly under the deltoid. Due to this whenever the arm moves outward, the pectoral muscle is stretched. In yours if you animated the arm moving upwards, the pecs wouldn't stretch as expected. I may draw some pictures too, if you would like. The attachment to the arm isn't quite as wide.
    It would help us more if you post an example, well explained. 8)

    Thanks



  6. #6
    Donating Member NeOmega's Avatar
    Join Date
    Jul 2004
    Location
    Seattle, Wa
    Posts
    1,256
    Yeah, i gotta say, this is really cool.

    I have built low poly human models, and then tried to animate them, with unnatural results. This is really a neat perspective on how to properly model a human for animation.

    [blubber] I just want to say, I love everyone at Elysiun, [/blubber]



  7. #7
    Originally Posted by NeOmega
    Yeah, i gotta say, this is really cool.

    I have built low poly human models, and then tried to animate them, with unnatural results. This is really a neat perspective on how to properly model a human for animation.
    Thank you very much.

    Anyway, i think that the natural results, animating a human model, do not only depends on mesh structure. It also depends on fine rigging of the model. This to make understand that my tutorial is not the definitive way of doing a good animateable model. You also need a good rigging tutorial.

    If you look at these pages of a Elysiun user, http://kokcito.tk/tut/rig3.html and http://kokcito.tk/tut/rig1.html, you will understand that normal polygonal meshes never will simulate well the work that human skin do.

    One of the biggest problems is that real human bones, slide under our skin and muscles (not where they are connected with muscles, ofcourse), when we do some kind of movements. This appens mainly with Y axis bones rotations.
    Instead 3D applications bones actually move points of the skin surface, when rotating. This is the reason for some extreme pose is very difficult to simulate with bones animation.

    Thanks



  8. #8
    Member womball's Avatar
    Join Date
    Oct 2005
    Location
    Wilmington Delaware
    Posts
    1,016
    This shows the attachment points of the pectoral muscle. The bright yellow lines are attachment points. Note the attachment points stays consistent throughout the movement. Yeah this is pretty poorly drawn, the bones aren't quite in the right spot I think, couldn't find all of my notes. Also the muscle bunches up more at rest. It gets even thicker and shorter when its being used. These are deltoid movements really, just the pecs are greatly deformed by the movement of the femur (upper arm bone). Red=pecs, blue=biceps, green=deltoids, off white=bone. I can always revise the drawing if you want.




  9. #9
    Donating Member NeOmega's Avatar
    Join Date
    Jul 2004
    Location
    Seattle, Wa
    Posts
    1,256
    Oh my, I am afraid that tutorial uses too many bones for what I need to do. Even so, the better the mesh structure, the better you can do with the bones available.



  10. #10
    Originally Posted by womball
    This shows the attachment points of the pectoral muscle. The bright yellow lines are attachment points. Note the attachment points stays consistent throughout the movement. Yeah this is pretty poorly drawn, the bones aren't quite in the right spot I think, couldn't find all of my notes. Also the muscle bunches up more at rest. It gets even thicker and shorter when its being used. These are deltoid movements really, just the pecs are greatly deformed by the movement of the femur (upper arm bone). Red=pecs, blue=biceps, green=deltoids, off white=bone. I can always revise the drawing if you want.
    What I need is not a drawing but a mesh example. I know how is done muscles and how they attach to the bones. And if you look at my drawings at the beginning of the tutorial, I done exactly the same you do. Except that I did not draw the bones.

    The almost horizontal direction of lower part of pectorals, when arm is down, you draw (and me too), is something academical, in my opinion. In my experience, I have seen really few persons that have pectorals with that shape, most of them doing body building.
    But for a normal man or a woman, things changes a lot.
    Just for refreshing my memory, when you posted your comment, I went around the web looking some artistic nude photos, and I saw many different shapes or pectorals.
    And looking my self at the mirror, for example, I do not have a pectoral that goes horizontally from nipples to deltoids.

    Anyway, if this is the only problem, i think that the points of the shape I done in the tutorial can be easly moved to obtain the shape you suggest. This becouse I have done a strip of faces following the lower part of pectorals.

    Moreover, the shape I created with the tutorial is a really base shape to be completed.

    I have also tryed to rigg the mesh to test it. And it seams to do exactly what I expect it to do. Of course it need more complex rigging if I want it to stretch naturally.

    Thank you



  11. #11
    Member osxrules's Avatar
    Join Date
    Jun 2004
    Location
    Scotland
    Posts
    3,979
    Originally Posted by ender.saka
    One of the biggest problems is that real human bones, slide under our skin and muscles (not where they are connected with muscles, ofcourse), when we do some kind of movements. This appens mainly with Y axis bones rotations.
    Instead 3D applications bones actually move points of the skin surface, when rotating. This is the reason for some extreme pose is very difficult to simulate with bones animation.
    The new b-bones in 2.4 do quite well at sorting that problem. Basically, you segment a bone into multiple parts and as you rotate around y, the bone rotates less the nearer to the base of the bone you get.
    Pinky: Isn't life wonderful, Brain? Just think, we started out as lab mice forced to spend the whole day working our way through frustrating mazes that went absolutely nowhere. Now we get to do what humans do!



  12. #12
    Member womball's Avatar
    Join Date
    Oct 2005
    Location
    Wilmington Delaware
    Posts
    1,016
    I look forward to the completion of this. Although its quite a task to do the whole body. Wasn't sure if the muscle attachment was right, looked off. There are threads on other forums concerning topology which may be useful for this. I'm rather interested in this, since I am going to attempt to do a humanoid figure next. I think using the new features of the bones for muscle sliding would be great. If that was figured out, this could be sticky material, I think.



  13. #13
    I seen a total absence (on the web) of tutorials trying to teach the relations between the principles of human anatomy and 3D polygonal modeling rules. I never found any tutorial, trying to show how to visualize the human body to approach correctly polygonal modeling.
    Here is a link to tutorials on Human Modeling by Peter Ratner, which shows the relations between anatomy and modeling (page 13-27), maybe it is useful:
    http://cal.jmu.edu/ratner/tutorials/



  14. #14
    Member osxrules's Avatar
    Join Date
    Jun 2004
    Location
    Scotland
    Posts
    3,979
    Originally Posted by SoftWork
    I seen a total absence (on the web) of tutorials trying to teach the relations between the principles of human anatomy and 3D polygonal modeling rules. I never found any tutorial, trying to show how to visualize the human body to approach correctly polygonal modeling.
    Here is a link to tutorials on Human Modeling by Peter Ratner, which shows the relations between anatomy and modeling (page 13-27), maybe it is useful:
    http://cal.jmu.edu/ratner/tutorials/
    I don't recommend using that book. I think it's a bit fraudulent the way it is presented. The image on the front cover was actually made by Stahlberg and IMO it makes it look as though using the author's techniques will get you the results on the cover. That maybe says more about the potential customers but I don't like it at all. I've looked through the book and after seeing a vast number of references concerning the topology of the human body, his is really quite bad. Just look at the back of the knee.

    I'd say the link the Wim posted before is a much better one:

    http://secondreality.ch/

    Go to tutorials and subdiv modelling section.
    Pinky: Isn't life wonderful, Brain? Just think, we started out as lab mice forced to spend the whole day working our way through frustrating mazes that went absolutely nowhere. Now we get to do what humans do!



  15. #15
    I've looked through the book and after seeing a vast number of references concerning the topology of the human body, his is really quite bad. Just look at the back of the knee.
    Stahlberg tries to model a human as realistic as possible. The way he has modeled the knee apparently has a reason (realistic detail?). I don't think his human models are intended to be animated, it is more about the outside look. For animation there are better techniques (the seconreality tutorial is certainly a good one). But how Stahlberg models clearly relates to the anatomy which was the point I was bringing up. It still is one of the better tutorials I've found on the web, next to the SecondReality one and Joan of Arc.



  16. #16
    Originally Posted by osxrules
    Originally Posted by SoftWork
    Here is a link to tutorials on Human Modeling by Peter Ratner, which shows the relations between anatomy and modeling (page 13-27), maybe it is useful:
    http://cal.jmu.edu/ratner/tutorials/
    I don't recommend using that book. I think it's a bit fraudulent the way it is presented. The image on the front cover was actually made by Stahlberg and IMO it makes it look as though using the author's techniques will get you the results on the cover. That maybe says more about the potential customers but I don't like it at all. I've looked through the book and after seeing a vast number of references concerning the topology of the human body, his is really quite bad. Just look at the back of the knee.
    Yes I am agree. I have already read that book, and it is quite difficult to understand anythig he do. And, not only, but he works a lot with "optimal" subsurf not showing exactly polygon mesh.

    Originally Posted by osxrules
    I'd say the link the Wim posted before is a much better one:

    http://secondreality.ch/

    Go to tutorials and subdiv modelling section.
    I am quickly looking at the modeling steps of the girl.
    However he obtains basically the same base mesh I obtained at the end of my tutorial, if you take in consideration that I model a man breast and he models a woman breast. But the flow of the faces strips nearest to the shoulder/breast connection are basically the same (look at the latest two images in this page http://secondreality.ch/tutorials/mo...dy/body02.html).

    Anyway I think that that tutorial is probably the best I have ever seen. But still it does not explain why he do what he do (from the point of view of human anatomy) as like as all the others. And if I had not done the studies I done, I would not be able to understanding for what reason faces are positioned that way.

    However, it is a great tutorial.

    Thank you.



  17. #17
    Originally Posted by SoftWork
    It still is one of the better tutorials I've found on the web, next to the SecondReality one and Joan of Arc.
    Yes about Joan of Arc (that I have linked at the begining of my topic) and SecondReality I am completly agree with you.



  18. #18
    Member osxrules's Avatar
    Join Date
    Jun 2004
    Location
    Scotland
    Posts
    3,979
    Originally Posted by SoftWork
    I've looked through the book and after seeing a vast number of references concerning the topology of the human body, his is really quite bad. Just look at the back of the knee.
    Stahlberg tries to model a human as realistic as possible. The way he has modeled the knee apparently has a reason (realistic detail?). I don't think his human models are intended to be animated, it is more about the outside look. For animation there are better techniques (the seconreality tutorial is certainly a good one). But how Stahlberg models clearly relates to the anatomy which was the point I was bringing up. It still is one of the better tutorials I've found on the web, next to the SecondReality one and Joan of Arc.
    That's not Stahlberg's book though. I've seen a few of Stahlberg's models and the models of people who are in close contact with him and their models are not like that. They are more like this:

    http://forums.cgsociety.org/showthread.php?t=108412

    Also Stahlberg does animate some of his models. He did the Quake movie with Sarge dropping his cigar.
    Pinky: Isn't life wonderful, Brain? Just think, we started out as lab mice forced to spend the whole day working our way through frustrating mazes that went absolutely nowhere. Now we get to do what humans do!



  19. #19
    My fault, had been reading a body modeling thread by Stahlberg at CGTalk and mixed it all up.



  20. #20
    I have get a look also at the models of Stahlberg, in the links you posted.

    I think he done great works.

    But I saw the model wireframe in the CGTalk topic, linked by Softwork, and that is not a base mesh or a learning tutorial. That is the final result of a work.

    What I would like to analize is mesh structure (starting from a base mesh) related to human anatomy and 3D animation.

    Too high detailed meshes can make confusion to someone that is beginner.

    Thanks.



Page 1 of 4 123 ... LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •