Dude with ugly feet - now with dudette (CG nudity)

[Vertex_Pusher]

After switching OSs (to Ubuntu) and finding that open source apps run better under Linux, I tried out Wings3D seriously for the first time to get used to working with Ngons and picked up some interesting methods which turned out to apply to Blender modeling as well . So I returned to trying to box modeling the human figure for animation again and did a few figures before this one .

I am posting this here because it’s at the annoying tweak stage and I need to re-start from a cube again in my pursuit of the “ideal” method for modeling a human base model asset for my self . Bla bla bla …

I modeled toes into this one for the first time, so the feet are not particularly well thought out . The ears are not quite finished and there are no eye balls . But it is pretty much complete other then that, but because it has passed the 3K poly count I need to start over (I’m hoping for under 2K) . Maybe someone will find it useful .

The mesh is rigged very roughly for deformation testing .

And this is modeled with 1 BU = 1 meter in scale .

The file is saved in local/texture view (the “potato” view mode is great for looking at topology), so for newer users hit the “/” in the number pad or select Global View from the 3D header View menu .

The reason why the file is called PROJECT20 is because I try and save after each major alteration to the mesh from the base cube . So there are 20 blends before this that plot the modeling process .

Attachments

PROJECT20.blend (379 KB)

[ZombieJohn]

Hey, nice job! It’s always nice to see someone take a really solid crack at modeling the human body.

[Vertex_Pusher]

Hey, thanks ZombieJohn, it means a lot coming from one of the better modelers in the community . But it seems like you didn’t download the blend it self . I’d love a more in depth examination of the actual mesh data from the likes of you even though like I stated I won’t be reworking this .

[mouz]

The topology looks pretty good.

But i never get why people use so many e/n pols, except for lowpoly models :3

[Mats_Halldin]

I can’t help you with topology, but studying that file certainly taught me a lot of tricks and stuff. Thanks for sharing it.

[Vertex_Pusher]

But i never get why people use so many e/n pols, except for lowpoly models :3

That is mostly because I want to able to animate the mesh and the fact that I also use subsurf like all contemporary CG modelers . If I applied the subsurf modifier the poles would be obscured and look “neater” at the cost of needless resource usage . I want to have more then one character in a scene .

Taking on poles directly helps you see the structure better and faster .

@ Mats: glad you found it useful .

[Vertex_Pusher]

Another attempt . This time I managed to stop exactly at 2K quads, but I get the feeling that for a mesh with this LOD you need to go 2.5 ~ 3K quads to get a “complete” model … I only managed to get the rudimentary shapes for the hands, feet and head with little to no detail …

It’s rigged as usual with some minor adjustments from the initial bone heat weighting (I didn’t realize the crouch issue with the previous link, you’ll notice from the screen shots below I fixed it this time) to test for deformation .

There are some meshes on layer 2 one is the base mesh I started from and the other two are aborted previous attempts .

The deformation issues seem to be working out better . Hopefully after a few more tries they’ll get ironed out … Maybe for the next one I’ll try a chica … For that I might even finally use some references

Attachments

BASE10B21.blend (446 KB)

[Mats_Halldin]

2K quads […] I might even finally use some references

WOW, I envy your skills. Wish I will ever get even /started/ with organic modelling.

[Vertex_Pusher]

Well here is my first full body female mesh … It’s at the annoying tweaking stage again, so better to start from scratch again … I think I missed a critical step in the early stages …

I did learn a few things from this attempt … like how to simulate the movement of the scapula only through weight painting …

I hope this is useful for somebody …

@ Mats : thanks for the comment, the “references” comment was supposed to be a joke … “chica” is slang for a female (not sure if you knew that) … I actually didn’t use one in this case either, but that is because I am lazy, and because the way I work everything becomes pretty procedural (see the arm blend below, hit numbpad “/” to get out of local mode and look at the mesh with the step by step process in generating the mesh) . Figuring that out and tweaking a bit (with subsurf off) usually works for me …

I have an extensive art/sculpture background, so the transition into CG a few years ago was relatively easy . You might consider taking some drawing/sculpting class … Though as a general rule learning how to abstract what you see into simple shapes/ideas and building from that is something all the arts have in common .

I’m sure with some practice you’ll get to the results you want . Good luck !

Attachments

BASE_F7b.blend (304 KB)arms .blend (181 KB)

[Mats_Halldin]

Actually, “chica” is Spanish for “little girl” and ¡Hola chica! probably works all over the world.

Thanks for .blend files again – just opening them is a great inspiration. When I’ve tried to rig a human I always started off by weight painting everything bright red – your approach was fresh news to me. I’ll be having a look at the arm file – hopefully it will get me started with organic modelling.

Again, I admire your skills. Below is just an attempt to provide some critical feedback, hoping to give you something back:

• Back of knee – no “pit” (“fossa genu”), muscles tendons are supposed to form lateral borders
• Rotating the small shoulder bone (Bone 006_L.001) upwards causes shoulder (deltoid) to collapse
• Rib cage a bit large/wide for a female?
• Anterior superior iliac spine to low? With the hip joint flexed 90 degrees the “groin” (like in border between torso and leg ) looks “unnatural”. (To my knowledge, with spine bent laterally as much as possible, only body fat separates the lowest lateral part of the rib cage from the iliac crest. Maybe its just a problem with your spine rig.)
• Wrists/hands need more geometry, but I think you already pointed that out yourself.

Just my five pennies, if you don’t agree, just ignore.

[Vertex_Pusher]

Thanks for the in depth critique Mats . And while I agree with all of them let me break down why they are so in the current file(s) …

Back of knee – no “pit”, the rig as it is currently set up is not suitable for a decent knee joint, it requires an offset (like in the first blend) for it to bend properly like when kneeling . I believe I have the geometry for it there, but I need to set up a proper knee set up .

The shoulder’s topology got a bit screwed up because I decided to radically change the procedure in the middle of modeling her . The new method that I am currently working on works a lot better in fact when it is instituted from the base mesh .

And yes the rib cage is probably too wide … for some reason tightening that up seems an issue with me .

The illac spine issue is also one of base topology issue, and the new version irons out some of those issues as well .

I didn’t add a lot of detail or a rig to the hands proper because at a full body level of resolution I don’t think you need it (or normal map the detail) and modeling hands or rigging them is some thing I am not trying to figure out here .

Thanks for the detailed critique once again, usually I get the “proper placement” of poles comments that insist that I will have problems with animation … I wonder why it is so hard to understand that you can get decent muscle simulation with manipulation of topology and poles only ?

[Mats_Halldin]

Sorry for a late answer. Yes, apparently most of those issues are due to a fast rig and your focus on low-poly count. I just wanted to give something back and I’d love to see your progress – keep posting.

you can get decent muscle simulation with manipulation of topology and poles only

Bad topology is bad organic modelling – with all the problems I’ve had I certainly know how bad things can get. Your efforts to sort things out on a basic level are very inspiring.

[Vertex_Pusher]

Thanks for the interest Mats .

Well time to move on again . Shoulders work mostly . I realized a conceptual mistake I had which I’ll try address in the next attempt .

Issues on the mesh : scale got slightly whacked due to odd compulsion to add more detail/geometry a little late in modeling process . topology on the legs is a bit of a mess . deformation at groin missing adductor longus tendons .

Issues on the rig (on layer 2) : arms and legs have been IK enabled along with pole targets . first time using pole targets for a chain length longer then 2 bones . the pole targets are the children of the IK target . this is an old concept . you can change the position of the knee or elbow by rotating the IK target . the leg offset which enables the mesh to “kneel” have to be rotated in the negative X direction for the IK to “kick in” when posing with transform cleared - will address a proper fix later .

Having an IK only arm setup makes shoulder deformation easier to solve - this can still be broken though if the pole targets are misplaced, but is easy enough to fix with the IK/pole target parent child relationship, but it is somewhat confusing for keying purposes .

Found an extra copy of Peck’s Atlas so will use to add “proper” muscle groups in next attempt .

Attachments

BASIS_R22ALT3RIGSHAPES.blend (469 KB)

[Vertex_Pusher]

Well I guess it was bound to happen … I start off intending to do a “simple” project and wind up back to where I left off 2 years ago … So well here is my second attempt at rigging a base “skeleton” …

If anyone cares to remember I petered out on my last attempt because I couldn’t figure out severe clipping issues in the shoulder area using constraints on the armature (in fact later I resorted to making actual physical models to figure out the shoulder joint, which gave me some vague insights that only recently began to make sense to me) … Doing the previous meshes/rough deform rigs here helped me unravel some of the issues . Repetition and practice not only makes “perfect”, it can also give you some interesting insights .

Just a few notes about the blend :

The mesh is a rough approximation of a human skeleton (not the detailed mesh from years ago - I’m sure I have it somewhere, but I did this from scratch), and is also non-manifold (I haven’t modeled this way for at least 2 years, since my last skeleton, this started off as an attempt to model a low polyish teeth and tongue, and just roughly indicates surface anatomical landmarks, especially on the torso) .

One reason I decided to model/rig this was the realization I had about the scapula . The fact you can get decent results simply with weighting on a continuous mesh, and now with separate “mechanical” setups in a skeleton was an interesting result for me . It’s far from being perfect, or even accurate, and even though there is a “scapula” armature bone (for major motions along with the clavicle), the constrained motions are all done through weighting .

The rig is a lot more “refined” then before, though I haven’t added bells and whistles like FK/IK switches here . Like I said in my last post I did give the legs a “proper fix” for the knee offset issue . The arms are only FK and the legs are IK, and the foot roll could be refined more, but this does include my “piano player” hand rig from 2 years ago (the thumb automatically controls the carpal roll of the hand) which can be used on piano/guitar player animations easily via the copy location constraint (it does not make a good rig for making fists easily though - for that an additional “IK”/action constraint switch would be applicable here) . And for the riggers out there you might want to look at the solution for the offset knees and how the “IK” is setup …

I did notice a weighting issue with the neck in some extreme poses, and I’m sure there are others, but I have noticed that posting/embarrassing myself and starting from scratch is a good thing from a learning perspective … and I have been staring/ thinking about this for almost a month and I need a fresh gray grid …

Attachments

skel_ref5a.blend (481 KB)

[onnevan]

Great models! I’m usin them for testing cycles, thanks!
and just to clarify some concepts, hehe…
“chica” is spanish for “girl”, not “little girl”, for "little girl we say “niña”