Here’s where I’m at with the base shapes for the morphable basemesh project I’m working on. I’m pretty proud with how these turned out, although I do think there’s still room for improvement, so feel free to share any critiques you have. I’m aiming for a slightly stylized semi-realism, so more “video game” than “digital double.”
Below you’ll see the morphs for race and gender. The faces look roughly the same aside from some minor tweaks. This is intentional as all of the bases are meant to be the “same” base starting point, just a different race or gendered version of it. Not shown in this post are the other 50 or more shapekeys to handle making the jaw/nose/eyes/mouth different shapes, as well as some body ones.
Basically, my goal is to take every complaint I’ve ever had with any other character creation tool I’ve ever experimented with, and build a basemesh that has none of the things I disliked about them. Not that this one is perfect by any means, and these things are somewhat a matter of taste. But that’s the value of making my own.
I may tweak it more, but I finally made a male body I’m happy with. (EDIT: August 2022 me was an optimist. Both base meshes have been reworked A LOT.)
The brows are placeholders, just some tranmapped polygons shrinkwrapped over the mesh. The eyelashes haven’t been joined to the mesh yet as I want to wait until I have all the eye morphs setup before I take on that hell. Haven’t 100% decided if I’m going to keep them this way. anyway.
@Okidoki Thanks, that’s what I was aiming for. Something fit, but not too crazy - especially for the base. I’m hoping to create a morph or combination of morphs for characters with more pronounced muscles.
Hello
Modeling is going well. There are a few anatomical problems that bother me. The belly button is slightly above. I sketched out my suggestions roughly. It’s up to you to do it or not. Stay cool.
I did say in my original post that critiques or suggestions are welcome. I appreciate you taking the time to do a sketchover regardless of whether I agree with all the suggestions or not.
You might be right about the belly button being a little high, and you are right about the the back of the arm-should meetup area.
For anatomical accuracy for an unclothed figure, you’re right about the breasts. However, I should’ve mentioned I’ve kept the breasts fairly rounded intentionally as the figure is meant to be used clothed most of the time, and it’s FAR more convenient to model and cloth sim clothing over a figure who is already shaped like she’s wearing a bra, unless of course the intention is that she’s braless. And in my experience morphs that add nipples on to smooth breasts work so much better than morphs that try to smooth them away, especially if there are ANY other morphs for the chest area that these morphs would need to play along with. That’s been one of my pet peeves with MakeHuman and MB-Lab.
Looking in a mirror, it’s not comfortable to stand in a position that arches my back that much and pushes my stomach out like that. I might tweak the shape of the legs though.
They’re subtle, but they make a difference. For the back of the head, there’s actually a fairly crazy amount of natural human variety there, whether in the volume at the back of the skull, roundness, or the angle where the skull and neck meet. So mine looked alright when I was eyeballing it but when I overlaid a reference in pure ref I liked and tweaked mine to match, I liked the new one with less volume better.
Navel height now matches the Aneta reference images from the Blenderella course. I didn’t use Aneta as my main referrence while modeling (different face and body type), but I have periodically used her to check my proportions.
Adjusting the shin/calves back also made it look more natural. Back arm-shoulder looks better with a little more x-ward volume as well.
While there, I tried a couple other tweaks to the under eye area and nose. The under nose area has been a source of frustration since, well, my first character. Undereye was looking too puffy as well.
Probably. With the caveat that I’m a slow methodical perfectionist working on this in my free time (and very much still learning), my current plan is to share it when it’s complete. It’s currently unrigged and untextured though, the shapekeys are incomplete and a mess.
I’ve mostly narrowed down the source. They don’t appear in the viewport, only in the main renders. It’s not the geometry (too bad, that could’ve been fixed), or hidden geometry that shows up in a render but not the viewport (also could’ve fixed). It’s a combination of the refraction shader, and camera location(?). Something camera related at least, as the location/viewing angle seem to be the main differences between the camera that rendered the above, and that rendered the below:
It shows up in two or three cameras in my scene which means even if it’s a rare issue, this is probably something I’ll see more of. The main one where it is obvious is kind of a problem since it’s exactly where it needs to be to get that perfect shot of all three angles at once, and minor adjustments to camera location aren’t helping much. It doesn’t show up on glossy/transparent non-refractive shaders, but the cornea looks stupid as a hollow glass shell in profile view.
I did try creating a hybrid shader that takes care of the worst of it from the front:
I think my third option is to toss the inner mesh for iris and sclera + outer mesh for cornea eyeball* build completely and use parallax to fake a refractive cornea. I’ve seen it done on youtube so theoretically it might work. That might be going too far though.
(* I know the traditional and “correct” approach these days is an outer mesh that’s sclera + cornea, with inner mesh for the iris, but since I favor Eevee that approach doesn’t work if you want AO to work on the sclera and have a refractive cornea. Checking the Refraction box on a material in Eevee disables AO, and then you get burning white scleras unless you texture it darker. Fun adventure figuring that out).
Ups did i miss something ? Never saw a wireframe of the eye… ??
Anyway: i guess there are mutliple layers of geometry… could it be that this is the cause of this effect because the face edges are at almost the same position? I mean for example slighty rotating one layer may just lead to a differetn ray calculation… another idea: it seems to be that there is no subdiv modifier added and since blender does have this nice adapitve subdivision in the experimential feature set… maybe it’s (in general) an artefact because of he slight transparent material so only this has to be more smooth??
Thanks for the suggestions. I don’t think you missed anything, I didn’t post a wireframe of the eye before. I’m not really sure the best way to show both layers but if it helps:
I made the cornea shell comfortably bigger than the eyeball to prevent artifacts related to proximity. There’s about 1mm of size difference And I’ve checked for errors like disconnected geometry, double vertices, or modifiers being enabled or disabled in viewport but set differently in render, as well as face orientation and other normals issues.
I assumed it was intersecting geometry until I started looking for the culprit and discovered the artifact exists even if the scene has no objects other than one eyeball and cornea, no modifiers except subdivision, and disappears if I change shader to one without refraction, or a different camera. If I use 0 to view through the camera in the Viewport in rendered mode the artifact doesn’t exist just rendered (left image is rendered):
Mesh smoothness is a sensible suggestion as well, but there are two levels of subdvision in the rendered ones. The last image in my previous post was just straight out of the viewport, it didn’t have the same subdivision smoothness as the rendered images. It’s pretty smooth with the two levels of subdivision:
I also ruled out the geometry or weird shader issues by applying a regular glass BSDF over an unaltered sphere the same size and location as the cornea model and got the same exact artifact. As far as I can tell, it is genuinely just an issue with screen space refraction from a specific viewing angles or distances or something, and specifically during render.
Which, admittedly makes less sense than your much more sensible suggestions or the other things I thought it could be, but those seem to be the only things that effect it.
Recently learned about the magic button to render out of the viewport without using workbench. Can’t promise it’s the best topology, but it follows the traditional principles to the best of my knowledge, so it should animate well - in theory at least.
I may change my mind, but hair will probably be cards for the eyelashes, textures for the eyebrows - probably not the ones you see in my earlier renders, those are very much placeholders. Nothing ruins a good looking character like badly done particle hair. I haven’t been super impressed with the results I’ve been able to get with particles.
I’ve been considering giving eyelashes another go when the new hair system is added though.
As for actual hair on the head, that will definitely remain hair cards for the foreseeable future. Haven’t gotten great results yet, but I’ve done some with enough promise I’m a lot closer to the quality I’m looking for with hair cards than particle hair. Easier to edit too.
nice and good topo
i’v never been in character guru
prefer mechanical things LOL
i’m trying to make a new character model
did a first approx with quad for the head and it looks ok but topo is a real mess
but i think i need to bring it back to the standard topo
much easier to anim but also looks good
are you ready to share the basic topo mesh for the Male Female
even if it is rough i mean it is better then redoing it from scratch
and spent one week to get nice results
keep it up - nice work here
there is a tut but don’t remember the link
but did you use this one here
Thanks. Yeah, characters are hard. Trust me my first head didn’t look nearly this good, I’m sure with time you’ll get there. It may be easier if you start by trying to follow a tutorial rather than starting out by trying to box model the whole head from scratch.
I’m not quite ready to share my meshes yet. They still have a lot of baggage from development and need some cleanup. Hoping to get there soon.
I’m familiar with that image, but it’s not one I referenced while modeling as it’s not really one of my preferred references. There’s just some odd topology decisions, particularly around the mouth/upper lip and chin areas, and the model isn’t shaped well.
For my referrences, I mostly relied on the Blenderella model tutorials, combined with Dikko’s tutorials and CG Cookie’s videos on modeling humans and topology, as well as a folder of topology referrences from various sources, that has a couple dozen images in it.