bagginsbill,

Quote - The differences are not entirely due to shaders. Look at the edge of the iris. This is what I complained about many months ago - the hard edge. This edge was due to geometry, not to the texture. The new eye uses the same texture, but has a soft edge to the iris. And the iris-pupil transition is much more realistic.

I fully agree with the above. The right iris (left side of image) looks much better to me in all three examples.

just to add to what bagginsbill has said, the only thing P4 users will miss out on is the use of V4 and V3 textures on the new eyes.  he's saying he'll generate shaders to transform textures and apply them to a new UV map on the fly.  and it seems to me that if the  "sclera + iris + pupil" are one material zone, then talking about edges between the non-existent zones is completely moot.  the edges are whatever you texture them to be.  it's like worrying about the edge of a  mole on the torso.

Quote - and it seems to me that if the  "sclera + iris + pupil" are one material zone, then talking about edges between the non-existent zones is completely moot.  the edges are whatever you texture them to be.  it's like worrying about the edge of a  mole on the torso.

I wouldn't say completely moot. Those boundaries are in a sense prescribed by the way the geometry is built. There are no sharp edges between zones anymore, so there's a bit of wiggle room there. But if one made the iris or pupil significantly smaller or larger in the texture than they were supposed to be, I'm sure the result would look pretty odd.

Quote - You misunderstand. In the script, I evaluated the planar projection and then wrote explicit UV's that implement that projection into the OBJ file. These concepts of UV projection mode and so on are only meaningful while I'm in the scripting environment, dealing with the geometry in the object-oriented system. When I call "WriteOBJ" on the object, it becomes a totally standard OBJ file with no magic at all.

Well that I understand perfectly.
And you really don't need to do the dumbed down version for my benefit. ;-)

Quote -
Many of your responses and issues about compatibility with different apps seem to be all driven from this single mistaken premise - that the high-level features, like Level of Detail, like UV projection modes, etc. are lost if you target this app or that app. No they're not. Once I create the loadable prop, all those options are gone. I can create a new prop that expresses variations on those options, but Poser or Maya or whatever will never see those except as explicit frozen OBJ files with explicit topology, explicit vertex positions, and explicit UV values.

No I acknowledged several pages ago that "baked down" versions of what you're doign would be fine, and even accurately described your process before you did to this extent. In the post where I talked about how other programs deal with paramteric objects and history stacks.
My mistaken premises may have  been mistaken premises, but only because you've been real spotty with information as to how it all works until now. Hence I've asked questions to try to nail it down, not just because I was assuming it was one thing when it was something else.
What you've described above is exactly how some app like Maya would deal with it when exporting a static mesh as OBJ: all history/ability to adjust parameters gone and all projections baked into UVs. And your predefined UVs aren't much different than predefined UVs in mesh or NURBS objects.

So when are we gonna get to see this thing? I'm anxious to open it up in a modeling program and have a look.

i think it would look pretty odd more because of how big an iris usually is than anything else.  the geometry has a visible effect on the realism i'd like, but not so strong of one that i'd be likely to describe a moved edge as "pretty odd" myself.  not nearly as obvious as say,  missing the edge of a lip, and i've got at least one successful render where my totally inexpert morph moved the physical edge of the lip and the texture i used was pretty far in from the edge anyway.

but it's moot.  i was responding to the ongoing conversation about needing some sort of transparency and special geometry to achieve an iris with a realistically soft edge, when that's only needed when you've separated the iris from the scelera.  i guess it could even be different materials, as long as it was one contiguous UV map.

cobaltdream: No disagreement there. I wasn't quite sure which part of the conversation you were referring to.

Quote -
So when are we gonna get to see this thing? I'm anxious to open it up in a modeling program and have a look.

did i miss where you guys specified about what poly count you wanted?  because, iirc, that was the last thing he asked for, before these questions were asked. 

odf - cool, cool.  and it's a good point.  when she's set on standard UVs (and maybe one alternate?), someone should probably make a seam map like Snow Sultan used to do for the DAZ figures (and some of their clothes).

Quote - > Quote -

So when are we gonna get to see this thing? I'm anxious to open it up in a modeling program and have a look.

did i miss where you guys specified about what poly count you wanted?  because, iirc, that was the last thing he asked for, before these questions were asked. 

Source code, please! :laugh:

Quote -
but it's moot.  i was responding to the ongoing conversation about needing some sort of transparency and special geometry to achieve an iris with a realistically soft edge, when that's only needed when you've separated the iris from the scelera.  i guess it could even be different materials, as long as it was one contiguous UV map.

If you're referring to SaintFox asking about how to get transparency between the sclera and the iris a page or two back and where the two met, that's because they were separated.
Separate UV islands, but not separate objects.

Quote -
did i miss where you guys specified about what poly count you wanted?  because, iirc, that was the last thing he asked for, before these questions were asked. 

It's probably my fault then. I asked for 1,532 polygons exactly and said I wouldn't vote on it until I had exactly that number, no more, no less.
But I don't want to be the hold-up here, so I'll settle for anywhere between 1,531 and 1,533.

Just for reference: Antonia's current eyes have 128 polygons in the lo-res and 512 polygons in the hi-res version, if I recall correctly.

I was unhappy with the way Poser deals with the lines converging to a point in the middle of the cornea. It tended to make the specular reflection shaped like a tear drop.

I took a look at Antonia's existing eye mesh and, odf, I see that you did not "spin" the cornea, but rather made a mesh of quads sampling the semi-sphere of the cornea, and then merged that into the sclera which was spun.

So I was thinking about how to do this in the general case. I came up with an interesting geometry. I bet it has a name, but I don't know it, so I'm calling it a "staggered spin". The central point is eliminated, and replaced with a single quad, basically spin x4 of a single point. Then from there, the next "ring" has 8 points, connected to the central quad via quads. Then the next ring has 16 points, and the next 32. From there on, there are 32 points in each ring.

It's an improvement, but still not producing the highlight shape I expected.

Here's the wireframe.

Quote - Just for reference: Antonia's current eyes have 128 polygons in the lo-res and 512 polygons in the hi-res version, if I recall correctly.

Did you change that with AP118?
Because I'm counting 176 and 704 with AP 116.

Quote - > Quote - Just for reference: Antonia's current eyes have 128 polygons in the lo-res and 512 polygons in the hi-res version, if I recall correctly.

Did you change that with AP118?
Because I'm counting 176 and 704 with AP 116.

No, I didn't change it. Maybe I just counted the sclera and cornea and forgot about the iris and pupil. Anyway, the order of magnitude was correct, no? :laugh:

Quote -
If you're referring to SaintFox asking about how to get transparency between the sclera and the iris a page or two back and where the two met, that's because they were separated.
Separate UV islands, but not separate objects.

no, i don't think they were, according to bagginsbill.  they might have been on your eye.  but the posts went roughly, iirc:

• bagginsbill - hey, i have a cool new eye using the principles i was talking about.  the scelera, iris and pupil are all one material (and implied same UV).
• SaintFox - well, i'd rather use the other eye where i can have a soft edge to the iris because i have the geometry (i say geometry, because transparency needs some geometric consideration to work) and mapping to use transparency and soften the edges.
• bagginsbill - well, you can use shader magic to soften the edge of an iris.

my point is just that it's not a question of shader magic or anything else.  when it's contiguous, you can pretty much make the edges as soft was you want in the map itself.

edited to add- and maybe my perception is wrong, but that's how i read it.

That's definitely unique, BB.
I think Olaf did it best, with his quads in the middle and that nice smooth transition out from where the iris meets the sclera.
Spheres can be a real PITA to work with.
Could have something to do with non planar polys not quite aligning the normal to the direction of the light source. shrug

Quote - And here's a render. The highlight is shaped like a jelly fish. I'm stumped. Is the only thing that works the grid-like sampling you did, odf?

I honestly don't know the answer to that. The reason it looks the way it does is because I modeled the eye with very few polygons (be it 128 or 176) and then did a Catmull-Clark subdivision to obtain the hi-res version. In the original mesh I did something pretty similar to what you've described in order to avoid 'poles' and reduce the mesh density in the center.

I think in order to understand this phenomenon fully one would have to know Poser's algorithm for interpolation surface normals.

Quote -
edited to add- and maybe my perception is wrong, but that's how i read it.

cobalt dream, she was talking about the physical eye that Olaf modeled, but that I UVd differently. (And just for the sake of saying it again, I mentioned several times I am unhappy with the way I did it, but she likes it, so I didn't change it except to fix the warping in the sclera.)

So, when the Iris and the sclera are separate UV islands, it can be a bit tedious trying to make a nice soft transition since you're texturing over different areas at once and trying to line them up.
Well that's what I got out of it at least.
Not sure about the need for transparency now that you mention it though.

well, that wasn't what i read.  my understanding was that she was talking about that other eye, but as a comparison to bagginsbill's eye.  and saying she wanted to stay with the other because she could make the iris have a soft edge.  hence him talking about how to soften the edge with shaders on his eye.

Oh my bad, I thought you meant something else entirely. Too much eye discussion going on these days, too many eyes in the mix. ;-)

I think she was talking about how to reduce the sharp edge of the sclera geometry, since it stopped and had an edge and the iris is a physically separate object. That was never modeled with Poser or Antonia in mind though and was just meant as a refraction test which I uploaded in case anyone wanted to mess with it.
I think she was just going along with the experimental fun of it though and I don't remember her saying she wanted that one to be part of "official" Antonia.

Stop talking about soft edges - I have jellyfish!

I'm wondering if I was not as clever as I thought. I came up with an algorithm for doing elliptical vector interpolation that I thought could work for any pair of vectors. I'm certain it is correct for a right angle (quarter sphere) but mayhap in this case the turn from edge to center of the cornea is not perfectly circular. I'll get out my math-explorer tool and check it.

I'll be back.

Quote - Stop talking about soft edges - I have jellyfish!

:lol:

I'm not a big developer. But if you now and then test what you have for me, it's been OK.

So it turns out that my interpolation of vectors over an elliptic path for any transit other than 90 degrees was incorrect. I have fixed my code. My cornea and eyeball are now spherical.

I'm still getting somewhat distorted specular highlights, but then again, so does the old eye. See attached image - existing Antonia eye on the left, my experimental eye on the right. Both use the refractive shader for the cornea. You can see that the old eye highlight is a jellyfish in this particular orientation.

I haven't done enough studying of reference photos, but casually looking at a few tells me I need to figure out what is the exact shape of the transition between sclera and cornea. Should the two be joined by a cone, or should it curve smoothly between the two? I have it curving smoothly, but some images lead me to believe it is curved too much. Images of dissected eyes are suspect, as I think they are distorted by the preservation process. Many of them are not even close to spherical.

This is the first direct comparison I've made at scale. I see now that my iris is smaller than the original. But this can be compensated for by making the eye larger and moving the center of rotation back a bit. That's what I did in earlier comparisons that I showed in Antonia's head.

Does anybody have a definitive answer as to what the angle subtended by the iris typically is?

Also, I made some very subtle changes to the iris shape that improved the response to lighting at shallow angles of incidence.

At the moment the poly count is 480 for the eye and 395 for the cover, total = 875.

Eye pointing straight forward.
Camera orbited -20 degrees.
Infinite light rotated 70 degrees.

I think the eye is optimal. When I come to my eyes, look in the mirror then this has in the last shot in the last picture basic similarity. Except for the large Ires and Pupil is better in the original. There, it is true to the others. (Proportions)

Well I will continue to work on this as I have time, but I know many of you want to play with it. I have not integrated it with Antonia - it's just a pair of props.

Go here:

http://sites.google.com/site/bagginsbill/free-stuff/bbeye

THIS IS A WORK IN PROGRESS. Expect more changes. Discussion and suggestions, crits, etc. are welcome. Please help me. This is hard.

After I've reconfigured your parts a little with the help of EditPad Lite. I had reduced the version number to 7.2. And I have the Tranz set in the cover material on the 1, it was possible for me to see them. Otherwise, everything was black. I could but only with P7 render Render Status Poser4. At FireFly Renderer everything black.

P.S. They also accept other eye textures.

Heh - you didn't read what I wrote on the BBEyes page. You must enable raytracing. The cover material uses reflection and refraction. If you do not enable raytracing, then it renders black, because the entire cover is nothing but reflections and refractions. It has no color of its own.

It is OK to use transparency, but you basically turned off the cover.

And you should not use textures that have baked-in reflections like that one above. The whole point of this eye is realism, not fakery.

Also, in this render, I turned off the specular effect completely and just used self-lit environment props, particularly a one-sided square with a picture of a window on it. This is much better than the little round dot you get from the Glossy node.

I also did the blurred-iris trick in this shader.

I'll post an update to the zip file with these new shaders later today.

Click for full size.

(Note: Need more detailed skin textures for this sort of image.)

Note 2: If you scale the eye I posted up to 110% the iris looks the same size as it used to. I don't know if I should adjust the size of the iris relative to the eyeball, so that the eyeball radius stays the same as it was, or is it fine to just use bigger eyes. Thoughts, people?

Rather than hide the covers, which means you may forget to show them again before rendering, change the Element Display Style on the covers to wireframe.

Another tip - load a primitive and have both eyes "point at" the primitive. Now you can move where they eyes point easily by moving this primitive around. Make it hidden during rendering.

Go into the cover material. Find the Glossy node and set Ks=0 on that node - this turns it off.

Go find a photo of a bright window on a dark background. I found this one by Googling for "bright window". Mount it on a one-sided square. Scale the square big. Connect the image to the Diffuse_Color and also the Alternate_Diffuse in the square's material. Set the Diffuse_Value = 0. If you want the photo to be brighter, you can use additional nodes to alter its appearance. I anti-gamma corrected mine and then ran it through an HSV node with Value=2.5 making it really bright in the bright parts, and really dark in the dark parts.

Add a spotlight. Position the photo square centered on the spotlight. Then parent the square to the light. Now you can move the spotlight around and the photo will move with it all the time. The reflections will match the lighting and be much more realistic.

I used bump maps, no vss yet and SF lights for this imagE.

Now back to M4 for a bit...sigh.

Sorry for the edges in image, zoom and they're gone.

Well thanks for uploading that BB. :-)
I got to see the part I was interested in so I'm happy now. I think it could have used a few more edges to soften the cornea, but hey, maybe too many was the problem in the first place. Can't argue with it if it works as-is though. Nice job.

One very strange thing..
I opened the eye OBJ in Lightwave and the center point of the pupil is actually 32 vertices stacked on top of each other, resulting in 32 non planar polygons that are actually quads, that anyone would guess are triangles just by looking. Strangely enough, they act like welded points, but aren't. That is, you can grab one of those polys and move it and all the rest go along for the ride. if those points weren't welded, it would come apart at the center. I have to say I've never seen that kind of behavior before.
Not that that doesn't work, but it could create render artifacts in some programs. I'll do some render tests and see what happens. Putting it into SubD typically solves any such issues, but you never know til you actually try it.
The UVs are distorted really badly towards the outer edges,which is what will happen when you planar map a half sphere straight down the Z axis.

Yes... I know, it's not designed for texturing in the normal sense, but people are still going to want to be able to, or at least try to.

I have no idea what's up with the missing polys on the UV map at the top and right side, but they're there on the model. Probably a UVLayout map rendering bug.

I have no idea what's up with the missing polys on the UV map at the top and right side, but they're there on the model. Probably a UVLayout map rendering bug.

Oooh - good catch, Mike on the stacked vertices. In the script I only weld exact matches and I probably forgot to force those points to 0,0,0, resulting in them not being welded. Any tiny deviation from 0 results in those points moving during the "revolve" operation, resulting in duplicated points.

I thought about added code to weld up to a certain tolerance like the big boys do. I hadn't bothered because most of the time I was able to force things to be exactly the same. I seem to have failed to do so in this case.

I know the stretching happens on the sides, but you can't rotate eyes far enough to even see them, as evidenced by the fact that I only built half an eye. I figured since people put me on a polygon budget, I traded the back polygons so I could make the whole cover and still have under 1000.

BB: Very impressive work! I'll test your eyes with Antonia this weekend. My gut feeling is that your irises are probably a bit on the small side when compared to real eyes, but I might be wrong. My only real concern is that if we make the eyes bigger, then we need to change the center of rotation for them. That's fine for Antonia since we're planning to simply replace the existing eyes with your version. But if you're thinking about making eye-replacement props for other figures, you'd have to make sure your eyeballs have the same size as theirs, so the relative iris-size needs to be whatever is appropriate at that size.

MikeJ: I'm almost willing to bet that those missing points on the rendered UV template have u and v at 1 or ever so slightly above 1, respectively.