For my part, there's definitely interest in the low-res figures. Less in the super high res, although those could be great for portrait work.

Sub D in D|S probably means that the D|S crowd will probably be highly interested in a low-poly Antonia.

I think you're right that porting the morph transfer software to PoserPython probably isn't worth the effort.

About the Scala thing - would it be possible to compile your code to bytecode, and provide a link to the Scala download pages in the documentation?

Morph target transfer software - man, that's been a holy grail for ages! There are some solutions on the market, especially designed for transferring morphs from figure to clothing, with varying degrees of usefulness and quality of the resulting morphs. But if your solution transfers morphs cleanly, there'll be quite a market for that software alone!

Well, I for one would be interested in the low-res version! Like others have mentioned, largely as a "clone" when doing hair and cloth simulations, or in external programs where I can have live subdivisions to help smooth things out.

Quote - Sub D in D|S probably means that the D|S crowd will probably be highly interested in a low-poly Antonia.

True! I hadn't thought of that.

Quote - I think you're right that porting the morph transfer software to PoserPython probably isn't worth the effort.

Well, that would be a relief. :biggrin:

Quote - About the Scala thing - would it be possible to compile your code to bytecode, and provide a link to the Scala download pages in the documentation?

I could do that, but the bytecode is already coming close to a megabyte, whereas the source code is just a bit over 32KB. The Scala runtime library itself is less than 5MB, so essentially if I distributed the byte code, just adding the library to produce something that runs directly would be on the same order of magnitude. But I don't have to decide that right away. I could start out providing the source code for morph developers, and if enough people asked for a pre-compiled binary, I'd make one.

Quote - Morph target transfer software - man, that's been a holy grail for ages! There are some solutions on the market, especially designed for transferring morphs from figure to clothing, with varying degrees of usefulness and quality of the resulting morphs. But if your solution transfers morphs cleanly, there'll be quite a market for that software alone!

Well, let me be clear here. It wouldn't be a general purpose morph transfer program. It would simply transfer geometry back and forth between different SubD levels of the same object. SubD uses an algorithm known as Catmull-Clark subdivision. If you have a mesh that's been generated that way, you can simply reverse the process and reconstruct the original mesh. Now if you apply a morph to the subdivided mesh, you can still do the reverse subdivision in the same way, but of course you might lose some detail. I imagine that's pretty similar to what ZBrush does when you sculpt a high-res mesh and apply the result back to the original low-res version. You lose some detail, and that lost detail is then baked into a displacement map.
 

Quote - I would even suggest a ultra low resolution figure if that would be possible ( with no great expectation for bending and rigging just something like a mannequin)

Ultra-low-res, hmmm? That's a challenge. Well, my morph transfer process wouldn't work on that because it wouldn't be a reverse SubD anymore. But one could probably get the grouping about right, so that it bends similarly to the higher-res versions. UVs could get tricky, though, and might require some tedious editing by hand.

So I guess it depends on what exactly you'd expect from your ultra-low-res Antonia. Also, I don't think it would be any easier for me to do it than for anyone else who knows their way around 3d modelling tools.

*As I mentioned earlier, I'm basically developing her as a low(-ish) res figure

*What is the mesh res? *

Quote - *As I mentioned earlier, I'm basically developing her as a low(-ish) res figure

*What is the mesh res? *

Currently 9092 quads.

I have used Clothes Morph by Dimension3D to transfer hundreds of morphs from super hi-resolution figures to super low-resolution versions of the same figure in literally seconds.

Personally I am interested in the medium resolution and low resolution figures. I have rendered Posette well morphed at 4000x4000 with FireFly object smoothing and seen absolutely no resolution artifact. Resolution is better applied to clothing then figures given that even lower resolution figures look just fine nude with render smoothing. My opinion is that the optimal resolution is just only what is required for morphing no more...

Then she'll take a few subdivisions before you choke up Poser I should think. Looks promising

Quote - I have used Clothes Morph by Dimension3D to transfer hundreds of morphs from super hi-resolution figures to super low-resolution versions of the same figure in literally seconds.

That's nice dear. :biggrin:

No seriously, that's cool. You apply a general-purpose program to a special case, and behold, it does its job. What I'm trying is much simpler. Probably faster, too, but that's a moot point when we're talking seconds. The important point here is that the method I'm using is precise. If you morph the lo-res figure, subdivide it and transfer the morph back to lo-res, you get exactly what you've started with. There's no guessing or approximation - well, up to arithmetic round-off errors, obviously.

In principle, this could even be made to work correctly if accidentally you scaled and translated the figure while making the morph and at the same time your modeler changed the order of points. But let's not get ahead of ourselves.

Quote - Personally I am interested in the medium resolution and low resolution figures. I have rendered Posette well morphed at 4000x4000 with FireFly object smoothing and seen absolutely no resolution artifact. Resolution is better applied to clothing then figures given that even lower resolution figures look just fine nude with render smoothing. My opinion is that the optimal resolution is just only what is required for morphing no more...

Hmm, you're giving me an idea here. If I could reverse the subdivisions process that Poser uses for smoothing polygons, I might be able to produce a morph for the lo-res figure that would optimize it for Poser display. That could actually be quite useful. Does anyone happen to have a reference handy on how exactly Poser's smoothing works?

I think I read somewhere that Poser uses Reyes subdivision for smoothing at render time.
Firefly was not developed by Curious Labs, E-Frontier or Smith Micro, it was a third party render engine called Vortex.
Maybe this could lead you to some info on the subdivision process?

Thanks svdl! I'll go and look that up.

No luck so far. I'll go ask the guys in the Poser Technical forum. Maybe stewer is around.

I think I'll fire up D|S next and see what it makes of this mesh.

Of course, for this to work properly, I'll have to a) find a way to get Wings to export smoothing groups and b) find out if D|S actually honors those when doing SubD. Oh, and c) unfortunately then get rid of smoothing groups for the Poser version, because they make ugly.

It never ends. ::sighs::
 

It looks really great. you dont even have to sub-D it daz studio. You can also make it softer with Zbrush( smooth option) or even in UVmapper pro you can smooth the model and then still every morph will work.

Quote - It looks really great. you dont even have to sub-D it daz studio. You can also make it softer with Zbrush( smooth option) or even in UVmapper pro you can smooth the model and then still every morph will work.

True that. I could just create a smoothing morph for the low-res version to make it render more nicely in Poser. That would be very easy. I  can even do it in Wings. As we say where I come from: sometimes one doesn't see the forest for the trees.

I think the SubD variant could still be interesting for D|S users because it would retain all the definition with a very light figure. If the SubD stuff is implemented well, one could do scenes with heaps of Antonias all rendering nicely and smoothly with the foreground ones still having all the bumps at the right places. And then one could change perspective without having to switch all the low-poly figures that are now in the foreground for high-poly ones.

sometimes one doesn't see the forest for the trees

heheheh you must live where i live:D i know that saying too;)

I dont know about D|S cuz i have never used it:) but a smooth morph for poser users is a great idea. I dont think poser will do Sub-D

First model i rendered with a crappy firefly render(smooth polys enabled) 2nd without smooth polys and third model is just the wireframe.

it does work a bit, especially on the breasts.

Quote - That's nice dear. :biggrin:

Man, you always seem so pissed off at me... what did I do to you? I signed off the thread for a while on account of it but then figured you were over with it. You posted a development and at one point promotional thread in the public forum soliciting input... people have been generous giving you that and you seem put out actually having to read it after asking for it...

Oh well, sorry to have bothered you... was my misunderstanding of the thread title I guess...

bye :)

Quote - > Quote - That's nice dear. :biggrin:

Man, you always seem so pissed off at me... what did I do to you? I signed off the thread for a while on account of it but then figured you were over with it. You posted a development and at one point promotional thread in the public forum soliciting input... people have been generous giving you that and you seem put out actually having to read it after asking for it...

Oh well, sorry to have bothered you... was my misunderstanding of the thread title I guess...

bye :)

Sorry if it appears that way. I've sent my personal apologies to momodot via site mail. But I'd like to say to everyone that even if I seem confrontational and dismissive of critique at points, it's not meant that way. I really, really appreciate the input and it's helped me along a lot.

I did argue against things I didn't agree with, though, and I will keep doing that. I see it as a discussion, not a debate. If we argue our points, that gives us a chance to clarify our positions and understand each other better. Unfortunately, I can be a tiny bit stubborn, too. But if people make good arguments, I usually give in.

I'll try to be more careful with my words in the future.

It happens to everyone at one time or another where you could misinterpret someone response....(Smiling as he was typing) Maybe we all have to type in first our expression and moods at the time of typing...LOL...

Excellent work with the texture.It's looking very good:)

To clarify some more:

That morph transfer process I've mentioned is actually part of a larger program that I use to transfer information between different versions of the mesh during development. I use .obj files as the lowest common denominator to get the data from one program to another, but even that doesn't always guarantee that I won't lose grouping information, vertex order, etc. So that transfer program has helped me enormously to stay sane during the development process.

One aspect of this is that the final grouping and the UV mapping were done in the high-poly version, whereas for pretty much everything else I tend to work with the low-poly one. So the need arose to somehow get that information transferred. That's why I don't care about other morph transfer programs. They just wouldn't help me with what I need to do. But I haven't really made that clear before, so I can't blame anyone for misunderstanding.

With that out of the way: easy morph transfer between different resolution versions of the figure would indeed be a neat side effect of what I have done. So my question was whether it might be worth it to translate some of my code into Poser Python so that a user could - for example - easily apply morphs to the figure that were  made for a different resolution than the one the user has currently loaded. Again, that wouldn't quite be the same as using an external program for transferring the morphs. For one thing, if one only needed the transferred morph once, there would be no need to save it to an extra file.

This is all a bit far-fetched at the moment, but the reason I'm asking now is this: if I end up translating my code into Python anyway, I might as well do it now while it's only a thousand lines long. But so far it seems like, although there is considerable interest in the lo-res figure, people would be quite happy to use external programs for transferring morphs. Which is fine by me, because it obviously saves me a lot of work.

I get the impression that this morph transfer program would be useful to someone who creates custom morphs for Antonia, and wants to transfer them to different resolution versions.
Didn't really understand that before.
Still, someone who creates custom morphs will be used to using external applications to do his/her work, so adding tthe Java/Scala application to his/her toolkit wouldn't be too much of a problem, as long as it's got some decent documentation.

By the way, I am seriously impressed by your Antonia work. I'd like to make her some nice fantasy outfits and dynamic dresses, when she's available.

Quote - I get the impression that this morph transfer program would be useful to someone who creates custom morphs for Antonia, and wants to transfer them to different resolution versions.
Didn't really understand that before.
Still, someone who creates custom morphs will be used to using external applications to do his/her work, so adding tthe Java/Scala application to his/her toolkit wouldn't be too much of a problem, as long as it's got some decent documentation.

Yes, that would probably be the best solution. I'll write up some documentation and make the program available to morph developers if they want it. Soon I should have a file locker here at Renderosity where I can put stuff like that which is not likely to generate millions of downloads.

Quote - By the way, I am seriously impressed by your Antonia work. I'd like to make her some nice fantasy outfits and dynamic dresses, when she's available.

Thanks! I would be flattered.

odf are the Uv maps ready? If so maybe you can put a higher version up so some people can begin texturing for antonia, so she has some different maps when she's out.

Quote - odf are the Uv maps ready? If so maybe you can put a higher version up so some people can begin texturing for antonia, so she has some different maps when she's out.

Yep, I'm on it. I'm just waiting for my file locker to become active.

well it doesnt have to be that hires just put up a 512x 512 or 1024 x 1024 on maybe imageshack.us :) so people can download them

Hmm, you mean just the templates? I thought you'd want at least the mesh with that to see the textures in action.

Lessee ... hmm, the 2000x2000 didn't upload here, apparently. I can try 1000x1000 tomorrow. It's time for bed  now.

So, phantom3D has sent me a new version of the rig he's making. It looks amazing, and he says he's making JCMs for it too. So I guess it's time to look at the mesh again and make the seriously final changes before beta testing.

Here's a question for you modelers out there, or more specifically the morphers: how does your favorite program deal with hard edges or, respectively, smoothing groups? I've used them for Antonia at a few special spots because they're economic and more predictable than multiple loops close together. But I've already come across some technical problems. Nothing I can't handle or the normal user would have to notice, but it might make some things a bit more tedious for developers.

So I'm trying to imagine workflows now. I'm not quite sure how the software that people normally use for morphs deals with hard edges, and how much they would actually influence the process. Basically, if your program ignores them, your smooth previews will look wrong, especially if you're working with the low-poly version to make morphs. Also, if you then subdivide to produce a corresponding morph for the high-poly version, it might do weird things if and where it involves the hard edge.

Those hard edges are not all over the place, though, just where hard boundaries are needed: nails to skin, teeth to gums, pupils to irises ... you get the picture.

Any input on this? How's your workflow, which programs do you use, how do they deal with smoothing groups / hard edges (if you know)?

I guess with hard edges there will be limitations for what type of morph you can do to it.But because the hard Edges are in those small areas it shouldn't be a big problem...I use ZBrush 3 to make my Morphs.

As it is, the outlines of the areolas are formed by hard edges, too. I could maybe just get rid of those and leave the more fiddly ones in.

Finally finished the JCM's.  The last two for each leg took me 2 weeks! Didn't realize how hard it is to get two different JCM's for two different bends to work together seemlessly.  Anyway you can get 80% on the bend and 80% on the side to side together at the same time with good results. and between the two of them you can get most all human movement, with out creases and dents you see in most models.  All the other JCM's are done too just have to load them and maybe a little touch up and I'm done, yea!  Here's a pic:

And a little movie on You Tube:

http://www.youtube.com/watch?v=T0gTdocOqnA

Yay! Very cool.

Very nice Phantom3d. I've been following this thread for awhile now and I'm very impressed by odfs model and the rigging you've applied to it. Being a person that's interested in animating I'm just curious, have you tried her in the Walk Designer and have you tried her with BVH files? If you need any here's a very good resource: http://www.mocapdata.com/ Looking forward to trying her out when she's ready. Great work guys.

I have not tried her in walk designer, I will do that.  BVH files will have to be adjusted for the extra bones of the model but can be done. I used a BVH file made for the freak once, on my model Brad, which uses much the same rig, and it worked, but had to be adjusted.

Phantom
Im my best James Earl Jones "Vader" voice...Impressive!!!

ODF
I know wings3D and blender will use hard/soft edges but for poser you have to bevel things a bit to keep hard edges ( because of the the way poser smooths things ) AFAIK. Im probably misunderstanding what you're talking about here. Are you worried about harder edges not transfering over from low to high-res? Possibly making morphs alot different from low to high-res? If that's what your talking about I don't think it will matter too much.

I will probably only do morphs for the higher res one. I want to make some general shaping morphs for specific regions like hips-x, nose-y, and some FBM's such as BodyFat etc. Just general morphs that combine and reverse well, but I would also like to make a few very defined ( not necessarily muscular etc. )  complete character morphs that sort of stand alone and don't reverse or mix as well with other defined characters but still mix with the general shaping morphs and I know the low-res one will not support the defined morphs much, so Im thinking I may just do morphs for the higher-res one. Did that make sense? Lol.

Cheers,
JB

So when can we get our hands on her? Soon? Pleeeeze?

JB123: Yep, that's what I meant. But I've decided not to worry too much about the hard edges issue.

I'll leave them in for the low-poly version for now, so that D|S users get the correct boundaries when rendering with SubD. In Poser 6, I see some artifacts when I use polygon smoothing with hard edges. Since in Poser, though, the low-poly version only makes sense for background figures, I imagine those artifacts will not be too much of a problem in most practical cases. If I turn out to be wrong, it's dead easy to get rid of hard edges with a little script.

For the default, high-poly Poser version, I will of course remove the hard edges.

I'm hoping that morph-makers will consider supporting both versions. As momodot pointed out, there's software out there that can transfer morphs between different resolutions of a figure very reliably based on the actual geometry. So even if someone uses a modeling package that does wacky things with the hard edges, it should still be easy to get the morphs over to the other version and do a quick check to see if it still works nicely.

By the way, I've put the current mesh together with the texture templates in my file locker for inspection. Please note that this is not the final version yet, so if you play with it, keep in mind that some things might still change drastically before the official release. Also, please do not share either the file or the link yet and note that this is only up temporarily until I've established a proper development environment. So I might pull the file anytime.

Here's a WIP.

I've started a new skin shader. Nothing special just yet. I didn't work on the eyes or brows at all.

I tried using light-based ambient occlusion here. It went nuts on me. I don't like it. The lacrimals went berserk because they are deeply inset. I'm going back to material-based AO.

Now that I've rendered her up close, I'm not crazy about the ears. Does anybody else think they're kind of thick and puffy?

odf,

You might consider lengthening the lash geometry, and perhaps curling it a little more.

I wanted to put parameters in the lash shader to let you change the contour (longer on the outside, etc.). But they're so short, I really can't work with them much.

If they were longer, I can make them shorter in the shader just by hiding the ends a bit. But I can't extend them longer in the shader - I need longer geometry.

The inner area seems to be on a different scale than the rest.

I'm using a pseudo random number generator (Fractal_Sum) and I can dial in a different seed to produce a new pattern.

Every pattern I generate is clumping near the lacrimal. Close to the temple it's perfect.

Here's a different pattern using a different seed value.

You might want to shrink the lacrimal a bit - it's huge.

Hi Bagginsbill,

Those definitely look a lot nicer than the procedural lashes I tried.

You made some good points:

1. The corners of the lashes got pushed inward when I subdivided the mesh. It's just a consequence of how Catmull-Clark works. I had that fixed at one point, but lost it again, unfortunately. I guess what I'll do is make the lashes considerably larger, add some more polys and push them together really closely at the corners. That way I'll still get the subdivision artifacts, but they will be tiny and far out.

2) I checked my own lacrimals in the mirror, and apparently they are really that large. I hope I'm not a freak. But where they meet the eyeball they match its color so well that they're virtually invisible. Someone correct me if I'm wrong about this. Anyway, I guess I can just hide that part behind the eyeball surface and add a morph later to push it out if someone wants it.

By the way, I've noticed that V4 even has a tear geometry for adding that extra reflection between the eyeball and lower lash. I can't imagine many people are actually using that, so I decided not to include it in the figure. If someone wants it, it should be easy enough to make a prop later on and transfer all the morphs that would affect the lower lash to it. Again, someone correct me if I'm terribly off.

3) Never noticed that about the ears. :biggrin: The thing is, ears vary hugely, and most real ears I've looked at were pretty far from my idea of an ideal ear shape. So I'm not quite sure I can manage to shape a pretty ear that's not vastly unrealistic. I'll check some more references, though, and see what I can do. If all fails, I guess people can morph them any way they want once the figure is out.

It's good to finally have more people look at the actual geometry.

I did some more work on the lashes. I generate displacement so each hair is 3D instead of being flat.

I don't think I've seen anybody do that before.

So now it makes sense to apply some diffuse and specular to the lash hairs.

They're starting to look good, I think.

I wouldn't worry too much about the spacing. I think I can hide it or compensate for it with a little math.

Just make the lashes bigger.

I'm not even sure you need to curl them anymore, because I'm making them curl with displacement. Heheheh.

Yeah, I don't know much about the proper size and curling of lashes. I imagine it should be easy enough to morph them into submission, anyway.

Your latest lashes have a nice structure, but look a little smeared now. One thing you might want to check further on is how they look at a distance. My impression was that Poser (P6, anyway) doesn't always antialias detailed procedural transparency that well.

The skin is starting to look skin-like.

I tried your procedural iris shader from node cult, by the way, and I thought it looked really good on her. When I'm back home at my own computer, I can post the proper parameters for Antonia, or maybe just attach the material files I came up with. I think I made two greens, two browns and one blue.

I agree with BB that the lacrimals may be a little wide. I think the flat part could be scaled inward about half way ( just the flat part ) or maybe even less. It really is just a bit wide on the eye. I agree about the ears being a little thick ( particularly the upper part and where it rolls over ) and they kind of have a detatched look where they meet the head when looking at them from behind, but she doesn't have those perfect looking CG ears and I do like that alot.

Im really very impressed with the topology, proportion and shape of Antonia. A+ outstanding.

Cheers,
JB

I'll take a look at the lacrimals and ears.

That's pretty cool to see her first incarnation. She has come a long way.