Looking good!  Will this model be released?

Do the lips remain a separated part of the UV map or are they now part of "Face"?

adp001 posted at 10:06 PM Mon, 29 August 2022 - #4443506

Do the lips remain a separated part of the UV map or are they now part of "Face"?

But that reminds me that I was wondering about reducing the number of material zones for the eyes a bit. Maybe pupilLeft, irisLeft and scleraLeft could just become eyeLeft, and similarly for the right eye?

Speaking purely for myself, I would prefer to have separate mats for the iris and sclera at least, so I can put SSS into the iris but not the sclera. Come to think of it, separate pupil mat as well; I generally set it to be something light absorbant or at least with no specularity at all.

I suppose if the entire eye is one mat you could work with masks to achieve the same thing, but that would make things complicated, so might as well deal with three separate mats.

Hmm, I hope that with the new eye geometry there will be less of a need for that kind of trickery. But I don't have a problem leaving the mat zones as they are, either.

Let's hope one day a blackbird comes to take a stab at that can of worms and makes a LaFemme, V4 or G8 mapped version.

adp001 posted at 10:06 PM Mon, 29 August 2022 - #4443506

Do the lips remain a separated part of the UV map or are they now part of "Face"?

There also appears to be asymmetry in the face UVs around the eyes where the rim of the lids join the inner socket. At first I thought it was simply the face map not perfectly center X on the UV tile, my thinking here because painting an eyeliner on one eye, a very tight to the lid rim liner, does not copy flip horizontal exactly right to the other side. Actually, relative to that kind of flip in that area for that kind of makeup, it's way off. You can do this with most centered Poser face UVs. I then tried doing it in Mudbox with the brushes set to symmetry X and had a similar problem doing precision work accurately with symmetry. I think actually the problem might be twofold here. I think there might be a bit of stretching or distortion in the lid rim UVs and a bit of a uncentered on the tile problem in the placement. No offense to whoever UV mapped Antonia but there's some problems... Which really I'm surprised slipped through the QA process, or maybe it's just no one looked as closely at the minutiae. This would explain why the eyeliner makeups created by saintfox aren't very close and tight but more raccoon/Smokey eyed variety, or that's my theory there.

And I think you should probably leave the eye material regions alone. Most Poser figures have those zones condensed single though. It does feel a bit odd having to copy paste materials between them when used to working with figures that have a single mat for sclera, Iris, Pupil.

If you' ever get curious, have a look at Antonia-A.obj. That's got the UV mapping I made. I seem to distinctly remember that the version of UVMapper I used didn't support symmetries, so don't expect too much. May have been a free evaluation license or something. Everything else I tried was unusable. From what I've picked up so far, the situation is not so different today. Good UV tools seem few and far between, and probably somewhat expensive.

Anyway, that one has the lips attached to the face, and the eyelids were mapped in a closed position, so I'm guessing makeup should work better. But it's got some unusual seam placements, so it wasn't very popular.

The UV-mapping question is especially can-of-wormy for me because there is some very interesting recent research in that area, complete with published source code. So if I got started on anything UV, I'd feel much more inclined to try and wrap my head around that stuff and turn it into something I could use in practice than dishing out money for some software that, if I'm lucky, is maybe a decade behind.

Also: I'm now in the JCM-fixing stage. While I'm at it, I'm also fixing some older issues that weren't caused by transferring them to the modified mesh geometry. Still, I don't expect it to take too long. Staying in Poser and gradually learning how to work around the quirks of the morph tool keeps the turnaround times low, although it's not the most fun way of shaping a morph.

Interesting. Could you give us a pointer to that research?  

FVerbaas posted at 4:41 AM Wed, 31 August 2022 - #4443587

Interesting. Could you give us a pointer to that research?  

ETA for the non-mathematicians: conformal means angle-preserving.

Also: I'm now in the JCM-fixing stage. While I'm at it, I'm also fixing some older issues that weren't caused by transferring them to the modified mesh geometry. Still, I don't expect it to take too long. Staying in Poser and gradually learning how to work around the quirks of the morph tool keeps the turnaround times low, although it's not the most fun way of shaping a morph.

odf posted at 3:05 AM Wed, 31 August 2022 - #4443586

Also: I'm now in the JCM-fixing stage. While I'm at it, I'm also fixing some older issues that weren't caused by transferring them to the modified mesh geometry. Still, I don't expect it to take too long. Staying in Poser and gradually learning how to work around the quirks of the morph tool keeps the turnaround times low, although it's not the most fun way of shaping a morph.

So if an exporter takes about 30 seconds either way ( that's a generous estimate) and you have 30 JCMs to make, you'd say 15 minutes all told is a significant turnaround increase to workload? Lol.

I see what you're saying. Generally it's pretty predictable what the blend will be like between off and full. I'm used to it, but yes I do end up making a few difference morphs and baking down my final result. JCMs aren't a task I find unpleasant though. I'm not the fastest morph maker but that's relative.

Could it be this is the paper you referred to?

It comes with code indeed:

primorge posted at 7:15 AM Wed, 31 August 2022 - #4443717

I see what you're saying. Generally it's pretty predictable what the blend will be like between off and full. I'm used to it, but yes I do end up making a few difference morphs and baking down my final result. JCMs aren't a task I find unpleasant though. I'm not the fastest morph maker but that's relative.

And sure, I can see how one would be able to predict the blend quite well with some experience. Mine is pretty outdated. :-)

Could it be this is the paper you referred to?

It comes with code indeed:

FVerbaas posted at 7:28 AM Wed, 31 August 2022 - #4443718

Could it be this is the paper you referred to?

It comes with code indeed:

Looks like it.

SamTherapy posted at 4:35 PM Mon, 29 August 2022 - #4443491

Looking good!  Will this model be released?

Yes, but I can't promise she will load with a smaller nose.

odf posted at 4:57 PM Mon, 29 August 2022 - #4443492

SamTherapy posted at 4:35 PM Mon, 29 August 2022 - #4443491

Looking good!  Will this model be released?

Yes, but I can't promise she will load with a smaller nose.

Fair enough, provided it can be dailled down from pickaxe proportions. :D

Oh well, I just fell into a JCM rabbit hole. The thigh movements had been bothering me for a while, and making morphs in Blender is so much fun. Going with primorge's process of making difference morphs on top of the original JCM and baking them at the end.

See ya all when I see ya!

Quick snapshot of my hip/thigh JCM noodling so far: Toni 1.2 versus Toni 1.3 WIP in one of my favorite poses (rendered at subdivision level 1). Ask me about the health benefits of squatting deeply and frequently.

An improvement. I go for less crease where the thigh meets pubis, usually more relevant to a more acute zrot, but that's just a personal preference.

I can tell you one of the disadvantages of squatting deeply and frequently. Stoop labor.

JCMs for those type of rotation combinations in the thigh are tricky because something that might look right at one degree combination won't look quite right at a different combination. The only solution I've found for this are either multi axis JCMs which can only be done by value operations, or additional JCMs that kick in over top at degrees far into the rotation at a specific point. That is, starting not from zero but already into the rotation. It's a balancing act. Probably waaay too OCD for regular use cases.

If you' ever get curious, have a look at Antonia-A.obj. That's got the UV mapping I made. I seem to distinctly remember that the version of UVMapper I used didn't support symmetries, so don't expect too much. May have been a free evaluation license or something. Everything else I tried was unusable. From what I've picked up so far, the situation is not so different today. Good UV tools seem few and far between, and probably somewhat expensive.

Try Blender :)

Activate add-on: "Magic UV". It can rotate, flip, mirror and more.

https://docs.blender.org/manual/en/3.2/addons/uv/magic_uv.html

Another one: "Dream UV"

https://github.com/leukbaars/DreamUV

An improvement. I go for less crease where the thigh meets pubis, usually more relevant to a more acute zrot, but that's just a personal preference.

Agreed! I've started to work on decreasing the creases, so to speak. Takes a bit of practice.

I can tell you one of the disadvantages of squatting deeply and frequently. Stoop labor.

JCMs for those type of rotation combinations in the thigh are tricky because something that might look right at one degree combination won't look quite right at a different combination. The only solution I've found for this are either multi axis JCMs which can only be done by value operations, or additional JCMs that kick in over top at degrees far into the rotation at a specific point. That is, starting not from zero but already into the rotation. It's a balancing act. Probably waaay too OCD for regular use cases.

Quite true! I actually "invented" delayed-onset JCMs for Antonia back in the day (invented in quotes because it's unlikely I was the first person to ever do that in Poser). But they're much easier to implement now via keyed dependencies. This kind of work is generally much more pleasant now with the tools I have at my disposal than it used to be. Less energy spent on coercing the software and more on the actual problem is generally a win in my book. Still tricky, though!

odf posted at 5:15 PM Tue, 30 August 2022 - #4443558

If you' ever get curious, have a look at Antonia-A.obj. That's got the UV mapping I made. I seem to distinctly remember that the version of UVMapper I used didn't support symmetries, so don't expect too much. May have been a free evaluation license or something. Everything else I tried was unusable. From what I've picked up so far, the situation is not so different today. Good UV tools seem few and far between, and probably somewhat expensive.

Try Blender :)

Activate add-on: "Magic UV". It can rotate, flip, mirror and more.

https://docs.blender.org/manual/en/3.2/addons/uv/magic_uv.html

Another one: "Dream UV"

https://github.com/leukbaars/DreamUV

Here's some progress on the de-creasing. Still needs additional de-creasing for the higher x-rotation angles, but that's mostly hidden in this render.

odf posted at 8:50 PM Sat, 3 September 2022 - #4443892

Here's some progress on the de-creasing. Still needs additional de-creasing for the higher x-rotation angles, but that's mostly hidden in this render.

I do like creases sometimes, as long as they look plausible. The main problem, I think, is that deep creases are hard to control and tend to look ugly in at least some positions. Over-smoothing of bends on the other hand can lead to boring, unrealistic shapes, so it's always a bit of a balancing act. As of course you well know.

PS: That's a good-looking squat Nova's doing there, especially from the back.

odf posted at 7:38 PM Mon, 5 September 2022 - #4443998

I do like creases sometimes, as long as they look plausible. The main problem, I think, is that deep creases are hard to control and tend to look ugly in at least some positions. Over-smoothing of bends on the other hand can lead to boring, unrealistic shapes, so it's always a bit of a balancing act. As of course you well know.

PS: That's a good-looking squat Nova's doing there, especially from the back.

primorge posted at 11:56 PM Thu, 8 September 2022 - #4444162

Looking forward to seeing what you've been doing. 

Still not planning on weight-mapping any further joints at this point, though. Beware the feature creep!

Here's a quick snapshot of where I got to last night, and I'm really emphasizing the word snapshot here, lol. The pubis and inner thigh area as well as the buttocks definitely need more work, but it's amazing how easy and fast it is to produce clean shapes with a bit of pushing and smoothing. I should be able to get a nice basis for JCMs at the very least.

That's a tricky one you have there, the butt crack height.

I'd say the results would look better if a JCM kicks in when both thighs are bent to that extreme that lowers or pulls the start of the crease downward from the tension of the flesh. Now I'm saying in theory... the reason it would be a nice theory is because such a JCM is on the center of the figure so you can't simply create the morph, split it, and attach it to both rotations under that condition only. It wouldn't be right if it kicked in when only one thigh bends, the asymmetry would be wrong. It would have to be a strict value operation that applies the morph under the condition of both rotations happening, and only then. Does this make sense?

I've never been able to do this. Multi axis rotations on a single actor JCMs are doable, and I've managed to solve the math between negative and positive simultaneous rotations (I was over thinking the instructions that Nerd gave me on that, as I usually do). Multiple actor rotations single JCM? Center line?  Stumped on that one. I have the strong feeling it's possible though, it just escapes my ability to figure out and apply a formula.

Long story short, it would look better if her glute crease wasn't so long. Obviously this can be done easily by a manual dial fix, but that wouldn't be fancy ;)

Probably the easiest fix there would be raising the glutes higher or tighter, de-emphasizing the length of the crease. Probably the most sensible, easily done with weights or morphs. And more asymmetry "friendly" or correct to the tension of the individual thigh bends.

Yeah, to me the long buttocks seem to be the primary problem here. Both the bend and the side-side have a tendency to pull them down, but I think that should be fixable. Will see about the butt crack after that.

Here's a small update, only influence weight-maps on thigh bend and side-side at this point, no bulges yet and no JCMs. The long glutes problem is still quite apparent in the bottom-left image.

(Word of warning from Antonia: "Don't stick regular tape on your crotch when you take your anatomical reference shots for the gram. This is special skin-friendly censoring tape I'm wearing here, applied by trained professionals.")

Bulges get a tad tricky because they can go in unexpected directions, so it's a bit of back-and-force sometimes between bulge WMs and influence WMs to get the correct shape. That's why I'm trying to get as far as I can with just influence WMs first.

Are you applying body mechanics to the squat?  I know that you are trying extreme posing to get the bending and joints just right but I don't know of anyone that can realistically squat straight back like that.

Some people can

The leg spread looks perfectly fine, and so does the kneeling and deep knee bend, but the center of gravity for the squat does not look right.  I would think that if you are going to base your modifications on a particular pose, the pose should be as close to an actual position that an individual can maintain.  There should be an upper body tilt forward for a squat; Antonia looks like she is leaning against a wall.

I have to agree, hborre: upper body movement should always accompany those type of movements unless there is a support behind like you mention. On the picture above FVerbaas posted there is a definite backwards tilt and rollback of shoulders which is noticeable due to the hand and arm position.

Here's a side view of the squat, if that helps. Her forward lean looks sufficient to me, but I could be wrong. Ideally, she'd want her center of gravity above mid-foot, but without a loaded barbell on her, mid-heel should be fine.

That looks totally plausible to me. The same thing crossed my mind when trying to get Nova in a similar pose. The one I made is not plausible, probably, but it was just to show the similar bend issues. If amputees can eat breakfast with their feet I'm pretty certain someone can maintain balance in such a pose.

That looks pretty plausible on the side view.  Maybe a little too much bend on the knees, otherwise, it looks good.

I have a feeling that my old knee JCMs may be lengthening and thinning the knees when bent, so that might be contributing to the impression that they're beyond their natural range. It's also possible that the rotation center for the knee needs adjusting. I remember checking this pose in Antonia WM, and the legs looked considerable shorter.

Makes me hurt just looking at it

Well, I'm a youth of only 58 years, so obviously I'm a bit more bendy. But I've looked at some references, including myself in the mirror, and I think I'll change my test squat to a more relaxed position with an overall stronger forward lean, some lumbar flexion and less extension in the upper spine. It's important to make sure that the thigh bend interacts well with the hip bend, anyway.