Eyes looking straight forward

EarthQuake

Ok so with most of my models i have to rotate them about 10 degrees outward to make it look like they are looking forward, is this normal or am i just placing the eyes in the wrong spot?

Snowfly

Eyes point out 5-10 degrees outwards in real life. It's perfectly normal.

Michael Knubben

If i have a look at your last model, i have to say that the eyelids don't curve the right way. The eyelids should always look like they're draped over the eyeball, and with the girl you modelled, it looks as if the eyeballs are pushing against the lids somewhat, but they're not really draped.

JKMakowka

The iris is always looking strait forward if you mean that, but that shape of the eyelits around the eyeball is somewhat furtherback at the outside.

KDR_11k

Eyes point out 5-10 degrees outwards in real life. It's perfectly normal.

It's not normal. My eyeballs aren't aligned perfectly and this kills depth perception. Though I've heard that helps with art since you no longer see a difference between the 3d reference and the 2d canvas...

MoP

Surely it all depends on what you're looking at?

Snowfly

It works out anatomically. Try placing spheres in as eyeballs and see if you can structure the surrounding area properly without turning the eyeballs out. Leave them at 0 degrees and you compromise the shape and proportions of the skull itself.

MoP

Something I don't quite understand - essentially eyeballs are spheres... how will rotating them compromise the proportions of the skull? Spheres have 360 degrees of rotational symmetry...
The only thing is the cornea which bumps out slightly, and that hardly alters the shape of the eyelids at all. Surely you can model a high-poly face using a sphere as a stand-in for a "proper" eyeball?

ElysiumGX

how can two objects only inches apart focus on one point in space and have a 5-10 degree outward rotation...unless you're a blond chick? i wanna see proof of this.

KDR_11k

Oh, that's easy, they point into different directions and the brain mostly ignores one of the images. You can even tell your eyes which one to point at the object. Though it might be possible that they don't point at the object directly and the brain shifts the images around but that makes no sense biologically. Besides, having the eyes straight works for my models.

EarthQuake

just so there is no confusion here, heres some images to show you exactly what i mean
rotated 10 degrees outward:

no rotation:

KDR_11k

I think your eyeballs are too large and that makes you misshape the socket, resulting in the eyes being too far inwards.

Rick Stirling

Are you viewing your model in a perspective or ortho viewport? That makes a lot of difference.

MoP

You've modelled the eye sockets wrong!



Not saying I've got it perfectly right here... but I think I've got the shape and positioning of the eye socket and eyelids more realistic than you have so far.

EarthQuake

KDR: yeah you might be right there, ill try smaller eyes
Rick: I really dont understand how that could posibly make any difference what so ever, but its a perspective viewport with "perspective amount" turned all the way down in lightwave.
Mop: yeah that does look quite a bit better than mine

KDR_11k

The eyeball isn't much wider than the socket, try that size with the pupil centered and it should work.

MoP

Yeah, I usually find that eyeballs will be approximately the right size if there's about an eyeball's space between them. So duplicate one of your eyeball meshes, move it to the centre axis in the front view, if it just about touches each side of the existing eyeballs, then they the right size. At the moment the ones you're using are way too big.

Whargoul

Yep, you eyeballs are way too large, and the sockets are modeled off-center giving it that look. You eyeballs are centered in the sockets so you had to compensate by sliding the texture over.

Try putting a temp pupil texture on the eye sphere when you're modeling the socket, so that you can see where they line up.

CrazyButcher

ortho or not ortho makes a big difference,
you are used to see "perspective" and nearly all images would be perspective. so things might look wrong when you watch at them in ortho (more fat/wide), but are correct when in perspective.
I think a common mistake would be taking a photo as reference in the background of a ortho view and trace it 1:1, then again I am not experienced with faces, but maths tells me it would be bad to make no distinction.

Daz

One way to prove or disprove this theory that eyes point outward in the skull once and for all would be to find that visible human project where they took millions of slices through a deceased ex con. Can't seem to find it though.

MoP

Yep Daz, that's what I was looking for yesterday, but I didn't find anything on it. Shame.
CrazyButcher is also right about ortho and perspective making a difference... cameras for taking reference photos have different focal lengths and stuff that distorts "perspective"... which is why it's often difficult to make an accurate model by perfectly tracing photos in an ortho view.
I still don't understand why people use 3d orthogonal views, with no perspective, because to me that just looks really unnatural since we never actually see things like that in real life! And I have yet to hear a convincing argument why it's a good idea to do so... of course many people just "prefer" using orthogonal 3d views, but to me that makes no sense.

ElysiumGX

typing "visible human" into google often helps. try axial slice #125. looks straight to me.

Daz

Shucks, looks like were gonna have to cut somebodys head open then.

Yeah, you gotta have *some* perspective in a modeling viewport, but too much can be a problem too, obviously.

Daz

doh.

Actually I dunno. This is interesting, there does appear to be a very slight angling out. Clearly the eyeballs are somewhat off perfect spheres:

*EDIT* btw, this is probably NSFW Im not sure. Not for the faint of heart either

http://www.cs.umd.edu/hcil/visible-human/vhp/dl96/a_vm1109.gif

MoP

Interesting. Yeah, a couple of degrees outwards is what the HL2 people said made the models look more "lifelike" with their expressions... 2-3 degrees of outward rotation looks fine on my mesh too, I might try that in future. Still, it shouldn't make any difference when modelling, whether they're rotated or not.

ElysiumGX

Now the debate is, are those eyeballs slightly pointing outwards because the person is dead and no longer capable of sending efferent impulses through Cranial Nerve VI to the muscles that control the lateral movement of the eye?

eyes out

Snowfly

If that person is dead, and his eyes are skewed out, then that should make it a good default position to set the eyes in, don't you think?

I think I read somwhere that eyeballs are universally the same size, give or take a few mm. So babies and females with large, cute eyes with large pupils actually just have smaller heads.

That's related to my point about compromising the shape of the skull if the eyes were facing directly forward, cos you'd building the eye socket relative to the PUPIL while it's pointing in an unnatural direction.

EarthQuake

Yeah i understand about perspective in orthos i thought he was refering to the actual perspective in the program.

Rick Stirling

No, when I worked on Spooks we had a lot of problems with perspective and ortho. The other thing we found with using real people was that it is better to use a high res camera and take the images from several feet away and cropping, rather than being close to the face where the perspective got really skewed.