Human Progress

[Tobian]

Chris, why aren't you using 'skin' as it's the most modern of the SSS nodes, and works fine with VPR. It's a little tricky to get right, but then all SSS is. and yes that Fresnel node does nothing, because Materials override anything beneath them. The 'gradient' is driving the mix. You shouldn't need to mix an SSS material and reflection if you use Skin as it has both things built in.

As a note: It is technically impossible to mix SSS + transparency, unless the SSS is provided by a node which also uses a pre-process. I was told this by Antti, and after a number of experiments, I have found this to be true... That said I don't understand why you NEED it for humans, since human skin is not transparent, and you can handle the cornea/thin film of the eye by simply making a secondary polygon layer and applying Dielectric to it.

Anyway Absolutely beautiful work, huge fan!

[Chris Jones]

Chris, why aren't you using 'skin' as it's the most modern of the SSS nodes, and works fine with VPR. It's a little tricky to get right, but then all SSS is.

Sigma 2 was giving me more sclera-like behaviour than the other shaders. There might have been some other contributing factors (already my memory is getting a little hazy on this)... I'm using Simple Skin for the actual body skin though.

and yes that Fresnel node does nothing, because Materials override anything beneath them.

Except for bump... (and possibly others?).

The 'gradient' is driving the mix. You shouldn't need to mix an SSS material and reflection if you use Skin as it has both things built in.

Sigma 2 has reflection as well; the gradient in the Material Mixer is for the blend from SSS to transparency.

As a note: It is technically impossible to mix SSS + transparency, unless the SSS is provided by a node which also uses a pre-process. I was told this by Antti, and after a number of experiments, I have found this to be true... That said I don't understand why you NEED it for humans, since human skin is not transparent, and you can handle the cornea/thin film of the eye by simply making a secondary polygon layer and applying Dielectric to it.

You'd think so, but I spent some time testing a variety of different geometry and shaders, including having a separate transparent layer over the whole eyeball, but there were always render errors (eg the iris would be black) or there would be a hard edge instead of a blend. The only combination I could get to work at all was the nodes I posted earlier + the geometry below:



By pre-process do you mean the nodes in the "Materials" section?

Anyway Absolutely beautiful work, huge fan!

Thanks.

P.S. I turned on "Show Normals" for the illustration above, but nothing shows. What's up with that?

[Tobian]

nope, ONLY bump/Normal are not overwritten. I forgot to mention that, but yes.

By pre-process, I mean *any* node which launches a pre-process, where when you render, you get a pre-process launched when you render, such as Sigma, Sigma 2, Fast skin etc. Simple skin and skin don't cause this pre-process. Note however Simple Skin is not energy conserving, which is to say it reflects more than 100% of the light which hits it. Skin is, however, it doesn't have the epidermis/sub-dermis split, though in a practical sense I don't know if this makes any difference.

I am not sure I understood that wireframe, so I made a horribly crappy model of an eye in 2 minutes, to explain how you should make it, so that it works correctly in terms of material stacking. The materials aren't necessarily correct, I just blocked something in. Note because the model is made this way the inside of the eye will actually reflect back, if light shines from beside the camera - aka you will get red eye (the back of my model isn't red, so it's kinda grey eye hehe)...

As a note on energy conserving materials, you should always make sure that the *sum* of all diffuse inputs =100% or less. I mean in Simple skin you have 30% diffuse, 90% epidermis and subdermis, which equals 210%. If you turn them down to equal the sum of 100% it looks a little dark, as compared to skin, which is properly balanced in this regard. Quite often the sclera looks too dark, without cheating, this is usually not because it actually is, but because the skin is just too bright (for the above reasons). Build your material and shaders correctly and then adapt the lighting to the shader, not the other way round, then your eyes should ballance to your skin.. In theory... It's been ages since I've messed on with skin shaders myself, and it does my head in because I don't fully understand the parameters for the SSS shaders in LW!

[Tobian]

Oh and as another thought... specular of objects underwater tends to be quite dead..(like when you see shiny fish like mackerel underwater, they look quite matt) so you can probably kill the spec/ reflection component of the sclera / iris, as in bright light it does look rather matte, behind the glossy reflections of the cornea / moisture of the eye... also save you some cpu cycles not calculating another reflection blur! :-)

[Chris Jones]

Thanks for throwing that together - unfortunately it's one of the configurations I tried that gives a hard sclera-cornea transition instead of a blend, and there's no "lip" to catch the light like it should. I even tried to soften it with refraction blur, which did not work well. It also drastically escalates render times if I remember correctly (on my version anyway, since there's transparency layered on top of SSS, with ray traced refraction, reflection and caustics, which is apparently asking a bit much I guess).

The wireframe I posted above is an object within an object, and the normals of the inside object are pointing inwards, and the front polys of both objects (where the cornea is) are sharing the exact same space. I shrunk my vitreous chamber down because it was adversely affecting the SSS. The iris is matte for the reasons you outlined, but not the sclera since I only have one layer of geo there and the exterior needs to be reflective. Interestingly Sigma 2 tends to give the impression of an underlying surface anyway, along with giving the bump map a nice volumetric appearance.

As I mentioned earlier, there's more to this than meets the eye.

[prometheus]

Sigma 2 was giving me more sclera-like behaviour than the other shaders. There might have been some other contributing factors (already my memory is getting a little hazy on this)... I'm using Simple Skin for the actual body skin though.


Except for bump... (and possibly others?).


Sigma 2 has reflection as well; the gradient in the Material Mixer is for the blend from SSS to transparency.



You'd think so, but I spent some time testing a variety of different geometry and shaders, including having a separate transparent layer over the whole eyeball, but there were always render errors (eg the iris would be black) or there would be a hard edge instead of a blend. The only combination I could get to work at all was the nodes I posted earlier + the geometry below:



By pre-process do you mean the nodes in the "Materials" section?



Thanks.

P.S. I turned on "Show Normals" for the illustration above, but nothing shows. What's up with that?

show normals doesn´t show on subpatched objects..toggle it to off and it will show, if selected that is, can´t recall if it ever worked on subpatch in earlier versions of lightwave, probably not since it would have to re-evaluate a freezed object from the subpatch object to be able to present all the new faces the geometry is made of and
thus adding a lot more of normal guides.

[Chris Jones]

show normals doesn´t show on subpatched objects..toggle it to off and it will show, if selected that is, can´t recall if it ever worked on subpatch, in earlier versions of lightwave.

Ah, right you are. I can't remember the last time I needed to look at normals (or used polygons for that matter!). Put that down as a feature request - show normals on subpatch.

[prometheus]

Ah, right you are. I can't remember the last time I needed to look at normals (or used polygons for that matter!). Put that down as a feature request - show normals on subpatch.

yeah..maybe, I added some further explanation in my post earlier above, why that might be tricky. (probably not since it would have to re-evaluate a freezed object from the subpatch object to be able to present all the new faces the geometry is made of and
thus adding a lot more of normal guides.)

[erikals]

maybe using a clipmap as transparency >

[Chris Jones]

yeah..maybe, I added some further explanation in my post earlier above, why that might be tricky. (probably not since it would have to re-evaluate a freezed object from the subpatch object to be able to present all the new faces the geometry is made of and
thus adding a lot more of normal guides.)

..

Couldn't it just show the normal of the centre-most poly in each patch though? Seems that it's already displaying subpatches as polys, though I know not of these things...

maybe using a clipmap as transparency >

Not a bad... I shall ponder this!

Thanx

[Tobian]

I'd still say do my config, but put a lens ever so slightly inside the aqueous layer, to handle the slight lensing at the edges and make the ior's different. also some of the softness is just a texturing issue (stares at eyes in mirror)...

And I never said faster, i'm just going for accurate. not that my eye model is very good, but i spent like 15 minutes at 4am :-)

[Chris Jones]

Yep, tried embedding an aqueous humor in there. I think that might have been the one that made the iris go black! There's always a catch...

[RebelHill]

there were always render errors (eg the iris would be black)

I notice in your image that you appear to have co-incident polys. The "lens" is occupying the same space as the cornea... This will give you render errors such as you speak of a lot.

[Chris Jones]

Actually no, that's the one that's not giving the errors (at least not severe ones). It's the most successful of all my tests.

[lardbros]

Just out of interest... was it rendering black under all circumstances? So, for instance... VPR and F9? And how high were the ray recursions etc?
Sometimes it's surprising how high the ray recursion has to be set to get perfect refractions on objects... especially crystals or diamonds etc, so was wondering if this was the same?