(I've kept it simple and grotesque here for the sake of obviousness. Each of these is just a single character morph, except the last one, which I thought needed an expression. Do you see what I did here? ;) )

(Oops, accidentally posted the picture twice.)

I don't know what you see, but I see fun.  Especially that facial expression on the last one!  Almost a Tank Girl look.  Good job.

Satyress, located here:  http://poser-outerzone.com/freebie/freebie.html

...with Daz gen4 morphs, transferred with TDMT. Thought someone might recognize the morphs...the last one is Hiro, with one of his characteristic zany expressions. It IS fun! The Satyress is adorable, but she needed more morphs, and now she has plenty! (I will share the vmf file when I figure out how to fix the problems with it; I had to clean up some of the morphs after copying them, but the flaws are still in the original file.)

Tips and questions.

First, tips, that I learned the hard way, for anyone else attempting this.

1.    1.    DO NOT inject every morph package you have and then use the “copy all morphs” option. If you have morph packages injected (which you will, because you need the morphs to make the heads fit together), you will have to pick through the list manually, because it will make extra copies of the entire set of base expression morphs for every character in every morph package. This means, for example, if you have ethnic morphs loaded, you will get the full set for Norwegian, the full set for Ethiopian, etc., and none of these copies work as well as the original base copies (the ones that don’t have a bunch of letters before them) so to make it more manageable, you will have to spend an hour deleting them all.

2.   2.    Screen material areas, and use “exclusive” mode.  Screen out things like eyelashes, teeth, etc. In this character, the base of the horns is part of the head, and I wish I had screened them, because morphs that affect the top of the head cause the horns to warp. I had to clean it up in the copied morphs with the grouping tool and an Ockham script.

3.    3.   When copying morphs, make sure the figure and body part that you are copying to is selected as the current actor, even though the script lets you select anything in the scene, which might lead you to believe that what you’re copying to and from only have to be in the scene. If the wrong thing is selected, things start to get very weird. Figures start mixing up their parameter windows, forever, in that  scene, and you have to revert. Be careful. Save frequently.

Questions for anyone who actually knows how to use this: There are settings, but I don’t know what they do. “Distance cutoff,” “number of influences,”  “test normals,” “test using base geom,” “ray-casting.” I left them all on the default, because I was afraid to interfere with them.  

Is there anyone out there who uses this tool frequently, who is good at it?

There is a Restore Detail script that comes with it, but I couldn’t get that to work. It is supposed to clean up the messy morphs, but for me it either makes dials that don’t do anything, or it makes spikes shoot out of their faces. What am I doing wrong?

TDMT compares one geometry to another, to transfer morph targets between them.  Two different methods of comparison are used by the script, each with different options.  "Use Raycasting" will activate the raycasting comparison method, which can provide more precise results under some conditions.  The default option uses a sampling of "nearest" vertices on the source geometry to create morph targets for each vertex on the target geometry.  Raycasting results can be much better than nearest-vertex results, but nearest vertex works better when source and target geometries have dissimilar shapes.

"Distance Cutoff" is used with nearest vertex comparisons to rule out source vertices which are farther away than the submitted value.

"Number of Influences" determines how many source vertices will be used in a nearest vertex comparison.

"Test Normals" is used with raycasting to rule out source vertices which have normals pointing away from the normals of a target vertex.

"Test Using Base Geom" runs the normals test not on the current (potentially morph-deformed) shape of the source geometry, but on the un-morphed base shape.

The Restore Detail script may work best when run on a geometry which is not posed or transformed.  Zeroing the transforms of your actor (and its parent chain of actors) may give better results.

I was the author of Restore Detail and the co-author of TDMT.  Note that I have been away from Poser use for several SR updates now, due to Real Life complications, so it is conceivable that some recent change to Poser may be creating problems with Restore Detail.  What version of Poser are you using?

Thanks for the response, I love your script! I am using Poser 10. So, in restore detail, I zero the figure, and then run the script on the morph. Then, do I use the dial it makes along with the morph, or in place of it?

There is also the mystery of where my computer is keeping the vmf files. TDMT is pulling them from a certain folder, but when I look in that folder they don't exist (except for the sample ones that came with the script), and when I search for them by name on my computer it doesn't find them. (I have Windows 7.) Makes it hard to manage them, if I can't find them.

The material screening is tricky, I'm still trying to figure out the best way to do it, I suspect that it involves the use of the editor? I haven't used the editor yet. Is the purpose of it to combine matches of different parts to make a full-body morph? And I just load them all into it and hit "merge"? Or is it more complicated than that? Can I also use it to merge different parts of the head? I'm currently working on a figure who has figure eyelashes. On the last match I did (Satyress, above) I regretted not screening the eyelashes because they warp on morphs that open and close the eyes, so this time I didn't match the eyelashes, and now they don't move with the eyes. I was thinking maybe if I matched them just to the "eyelashes" material area of the other figure, and then merged the files? Would that work?

So, raycasting is for if the original geometry is similar, and not just if I morphed one of them beyond recognition to fit the other one?

Just tried it. Of course it doesn't work, they're different figures.

There is also the mystery of where my computer is keeping the vmf files. TDMT is pulling them from a certain folder, but when I look in that folder they don't exist (except for the sample ones that came with the script), and when I search for them by name on my computer it doesn't find them. (I have Windows 7.) Makes it hard to manage them, if I can't find them.

Windows User Access Control is hiding them; they should be somewhere under /AppData/Local/VirtualStore (AppData is a hidden folder) which is in your user folder under C:/Users...

The Restore Detail morphs are additive.  RD does its calculations based on the current shape of the current actor, comparing it to the unmorphed base shape of that actor.  The resulting morph is intended to be used with whatever morphs are active when Restore Detail was run.

The vmf files should be in :Runtime:Python:poserScripts:TDMTfiles:matched:[files].  Where are you looking for them?  TDMT was written back before Poser offered so many Runtime placement options, and before Windows UAC was causing file placement problems.  The script may be doing something unexpected because the standard procedure has changed since the script was made.

You can use either geometry comparison method whenever you like, of course.  Based on my own experience with the script, though, raycasting can offer the best results when the shapes are similar.  The test case I always think of is comparing the ears on two different heads.  The shapes of ears are hard to match to one another, and as a result raycasting doesn't work very well.  In that case, nearest vertex sampling can work surprisingly well.

In the case of something like eyelashes, I think what worked best for me as screening out the eyelashes on the source but not the target, allowing target vertices to pick up influences from the source eyelids.  I believe I was using the nearest vertex method for that, though.  With raycasting, eyelash comparisons have some of the same problems as ears, in that it's hard to make the surfaces of the comparison geometries similar enough for a good comparison.  Maybe try comparing target lashes to source eyelids using nearest vertex, then run the face comparisons as desired and combine the resulting .vmf files.

Windows User Access Control is hiding them; they should be somewhere under /AppData/Local/VirtualStore (AppData is a hidden folder) which is in your user folder under C:/Users...

Yes, that's where they were. Thanks. Still playing around with it. Slightly off-topic but related: The Ockham script I was using to clean up the morphs after I copied them was MTexcGroup. When it works it is very useful. But, when it ignores the grouped section and excludes the entire head, why does it do that? I can't think of a single thing I'm doing different, it seems random. Is there a way to control it, so that it actually makes a morph excluding the grouped section, whenever I want it to? And I was just starting to think that my luck was turning around with Poser stuff...

The first time I used it, it worked. Then it immediately stopped working, in the same scene, on the exact same group. When I was working on the Satyress is was working consistently so I started thinking of it as an available resource that I could use if I needed it. Now all it ever does anymore is create something that ends in "EXChead," even though I have the grouping tool open with my group selected. So what I'm wondering is, what is the factor that determines whether it recognizes the group or not?

Because they tend to mess up in the same places, across all the morphs. With the Satyress it was the base of the horns, the bottom of the neck, and behind the ears, so I made a single exclusion group of those areas, and ran the script on all the morphs that needed it, and then they were nice-looking morphs. I liked it a lot. But alas, it died. :(

I think Poser changed how actor groups were represented, a version or two ago, after Ockham's script was released, a change that may have been working toward the alternate skinning methods Poser now uses.  IIRC, in the past Poser did not list the actual body part groups (such as "head") among the geometry groups listed by the Grouping Tool.  That change to Poser may be affecting how the script functions.

I typically use the Morph Trimmer script from the TDMT downloads page, for such functionality, but I didn't give it very robust group removal functions.  It works better for stripping select material zones from a morph.  If the Ockham script isn't functioning, that might be an alternative.

Thanks, it just frustrates me that things make so little sense sometimes. It sometimes works. That means it is able to work. Therefore there must be some factor that determines when it works and when it doesn't, but I have no idea what it is.

Also, similarly: I stupidly already deleted the eye morphs that I copied because I was going to try them again with the new settings.  Didn’t work. “Morph Creation  Failed.” So I tried to just copy them again with the old file, the one that successfully created them in the first place, and again I got “Morph Creation Failed.”  On every single one. But it made them before, and they were fine except that the eyelashes didn’t move with them, and I mistakenly assumed that the fact that I copied some eye morphs between these figures meant that I am able to copy eye morphs between these two figures.  The same figures, the same vmf file, the same morphs. Again, there logically has to be a determining factor. I'm frustrated; sorry for the rant. It was seeming so promising for a while.

Morph creation is failing with TDMT?  IIRC, that message should come up when the script doesn't find any morph deltas in the source morph or the vmf contained no comparison data for the relevant vertices. 

The code for the error follows:

if havedeltas == 0:

    #print "============ Morph Creation Failed =============="

    #print "Either that morph did not exist, or there was no"

    #print "valid vertex weightmap info for that area of the mesh."

    transfer_error("%s: Morph Creation Failed" %(morphName[0]))

    return 0

Well, it worked the first time. It's the same scene and I still have them both loaded with the same morphs. I added a few more that I thought of that I wanted, but I don't see how that would mess it up, I've done that before with no problems. Maybe it has somehow gotten confused by the times I had the wrong actor selected when I tried to copy stuff? Maybe I should just run the match over again? I was wondering: once the match has been made, I don't have to have them lined up anymore, right? They just have to be in the scene and have the morphs? I'm not at my computer right now. 

I re-ran the match and now it's copying morphs again (although it still fails on a higher percentage of them than my first my first two projects, but it was doing that all along, even though it was my best job yet of lining them up.) I still don't know what messed it up, wish I knew so I could prevent it from happening again.

As for the eyelashes: The figure I am working on now has figure eyelashes (they are a separate figure from the character, they load with him but if you delete him from the scene they remain hovering there like the Cheshire Cat's smile) and that's why I can't combine the files. How would you handle such a case? They move with his built-in eye morphs so should I just copy all the same morphs to the eyelashes and assume they will keep following the eyes if they have the morphs?

And the answer is: Yes, if you copy the same morphs to the eyelash figure as you copied to the head, and it is set to follow morphs when conforming, then the eyelashes will move with the eyes. :)

[quote]I was wondering: once the match has been made, I don't have to have them lined up anymore, right? They just have to be in the scene and have the morphs?[/quote]

This is correct.

With your eyelashes question, it sounds like the approach you're considering might be best.  Lashes as a separate figure is strange to me.  If I were using that figure, I think I would combine the lashes with the base figure, as new actors, and rig them as body handles to move the eyelids when the lashes are posed.  That could be a complicated conversion, though.

Yes, well, it worked.

Satyress was the second one I did. I think it’s good for a second attempt; I’m proud of the ears. Look how nicely they morph! They can morph even more than that and still look good. The first one I did, I am also happy with.  It’s this guy, the Open Source Centaur, who lives here: http://www.wa1gsf.net/subpage.html   He’s a simple lo-res kind of guy, but I’m fond of him. (I’m drawn to certain types.) I think the morphs look beautiful on him.

The first one is the V4 horse head morph available on the same page. The second is the M4 character Jarryd. The third is the Bushman ethnic morph. Then the Freak. Second row is Maldegar from Philosopher’s Egg, then with the expression “confusion,”  Orc from Creature Creator, “Flirting C right wink,” and something from the set “You Choose 2.” (I have also been refitting hairs and accessories to him.)

Now I’m working on Renithra; that’s the one that’s giving me so much trouble.

When I try to use the editor, I get an error message.  I don’t know whether I’m using the editor wrong, or if I’ve made some other mistake.  What I did: First I went to Open, and opened one of the files I wanted to merge. Opening a second one seemed to be switching it out instead of adding it, so I went to Merge and then chose the one I wanted to merge it with. And then a window pops up that says:

“Exception in Tkinter callback

Traceback (most recent call last):

  File "C:Program Files (x86)Smith MicroPoser 10RuntimePythonliblib-tkTkinter.py", line 1410, in call

    return self.func(*args)

  File "C:UsersLisa_2Documentspython scriptsTDMT_Match_VMF_Editor2b.py", line 118, in mergefile

    gz = check_gzip(fname)

  File "C:UsersLisa_2Documentspython scriptsTDMT_Match_VMF_Editor2b.py", line 145, in check_gzip

    f = open(dataPath,"rb")

IOError: [Errno 2] No such file or directory: u'{' “

Is that the correct procedure for the editor? The files I was trying to merge both work. I would like for them to work together instead of separately. 

Version 2b of the editor is affected by a bug in Tkinter for Python 2.7, which reads file names incorrectly when askopenfilenames (plural) is used.  I thought I had uploaded version 2c, which corrects that issue, to my site, but it seems not.  I really need to find a good html editor, so I can do site updates again.

At any rate, I have uploaded 2c to my site, although the TDMT page is not updated to reflect this.

http://www.morphography.uk.vu/~cagepage/TDMT_Match/TDMT_Match_VMF_Editor2c.py

Thanks!

Cage, I’m sorry to keep bothering you with every little question, but I really appreciate your being available to answer them. My next 2 questions are:

1.       In TDMT, if the majority of the morphs are failing on a certain figure, and I know I lined the figures up as well as it’s possible to do, what setting needs to be changed? The distance cutoff , or the number of influences, or what?

2.       I finally opened your morph trimmer, and it is more complicated than I expected. There are 2 columns: “Morph to trim” and “subtraction morph.” What is a subtraction morph?

Thanks.

Cage, the link in your signature goes to the loop script instead of tdmt.  Can I have the link please?

Yes, what Kalypso said.

http://www.morphography.uk.vu/~cagepage/TDMT_Match/TDMT.html

Wow, my sigline is messed up?  Sorry about that.  Renderosity fought me when I updated my site signature last year... and now it seems that they've found a way to break even more site features than before!  I'll have to see if I can summon the courage to attempt a sigline correction.  Frightened, I am.  Thanks for posting the correct link, AsteroidLady.

Instructions for the Morph Trimmer script were posted about halfway down this page:  http://www.renderosity.com/mod/forumpro/?thread_id=2795602&page=7&page_number=10

With the TDMT failure, are you having problems with the comparison step, or the transfer step?  We should be able to figure out what's gone wrong, but I need more information.  Are you comparing using the raycasting?  What are your current script settings?

Thanks AsteroidLady and thank you Cage for all your cool and innovative scripts!

It was only the one figure it was failing with. Renithra, the little elf boy. I set it aside and started another project which I didn't have the same problems with (I had different ones, but I'll get back to that.) I tried it both ways, with and without raycasting, and I tried changing the distance cutoff as well. Some of the morphs are copying, and the ones that do look fairly good, but the vast majority are not. I used Michael 4 for the comparison, maybe not the most obvious choice but he's the figure I have the most morphs for. I am unfortunately away from my computer all weekend, it's a desktop one so I can't take it out of town with me. I left the number of connections at the default. I aligned them very closely, better than the first two, I was being perfectionistic about it and took my time.  On the material screening, I started out conservative and gradually added more parts. That's about all I can tell you from memory, I hope you can extract some meaning from it.

I think I would have to see how your geometries are aligned for the comparison, to offer much speculation.  My best guess otherwise would be that the cutoff distance you've used is still too low.  The raycasting box is un-checked, right?

Yes, it's unchecked. What would you recommend changing the distance cutoff to? There wasn't much change at .02. 

I would try raising the cutoff distance and number of influences, at least to see if you can coax the process into giving you better results.

I don't use either of the figures you're using here.  I assume M4 is pretty hi-res.  How does the mesh density of Renithra compare to that of M4?  Whether you're comparing "uphill" (low-res source to hi-res target) or "downhill" (hi-res source to low-res target) can make a difference in the results.  You may want to try running the comparison with raycasting activated, at least to see what different results the changed settings can offer.

Says here that the vertex count for M4's head is 15078. Renithra's is 17618.

Poser is behaving very strangely for me right now. I ran a comparison between M4 and Tyler, mostly because I wanted Tyler's ear morphs for a character I've been working on for a while that I can't seem to get the ears right. It doesn't want to do that one at all. I was able to copy the ear morphs that I wanted, but they only work on the copy of M4 I was using at the time. I tried exporting them to use on the character that I really wanted them for, which should have worked, but when I tried to load them it said the vertex count was wrong, even though it was from one M4 head to another. I make morphs that way all the time, there is no reason it shouldn't have worked unless the one M4 has somehow had his vertex count changed. I tried it repeatedly, double-checking that I was doing it right. Finally I decided I would save the M4 that had the ear morphs and load them into the scene, temporarily zero my character and use TDMT to transfer the ear morphs from one M4 to the other.  Instead of loading a second M4, and having 2 of them in the scene, it merges them into this one weird multicolored thing. I tried it 3 times, and the colors are different each time. (That's just how it looks in the preview window, though; it renders as a single figure.)  As for Tyler, I can't copy any morphs to him. Can you think of any reason why this particular pairing would be so difficult? Or, have I just broken Poser?

For a 1:1 comparison of identical meshes, I use the nearest vertex comparison with number of influences set to 1.  This has worked for me, but note that I pretty much always use Antonia for these things and I have limited experience with the Daz figures.  I'm not sure what cr2 oddities might be built into the Daz figures to maybe run interference somehow.

If you want to port deltas between two copies of M4 and Poser isn't cooperating, you might try directly pasting the morph deltas from one M4 cr2 into another, if you have a cr2 editor.  Recent versions of Poser also have the "Copy morphs from" option, which generally seems to work pretty well.  (For that matter, IIRC shvrdavid posted a tutorial in the RDNA Poser 10 forum, explaining how to use Poser's new morph transfer tools to do the same sort of transfers you've been doing with TDMT.  I haven't tried his method, but it looks like a good one.)

For an uphill comparison like your M4-Renithra, I think we found in tests that the raycasting method tends to work better than the nearest vertex.  It's been a while since I've delved deeply into these things, though.  I keep hoping to be able to get back to Posering one day soon, but stupid Real Life issues haven't settled down yet.  :(

Forgot to mention: Before that, I brought Tyler into my scene and tried to transfer the morphs directly to my character, but they wouldn't copy to him, even though they copied to that other M4.

With Renithra, increasing the number of influences to 10 solved the problem. Now I can copy as many of them as I want to.

I gave up on Tyler, though, it just doesn't want to do that for some reason.

Next question: Can this be used to create full or partial body morphs? Like, that affect more than one body part at the same time. I was assuming it could, but now I don't see how. It only lets me select one body part when copying the morphs, even if I've merged the files.

The specific case that drives me to ask, it's the one I said I would get back to: I decided to take a break from matching figures with completely different shapes and match two human women of the same approximate size and shape, which I figured would seem really easy by comparison. And it was, except one of the figures was Dawn, whose jaw, it turns out, is not actually part of her head. So I transferred morphs to both the jaw and the head, but it's going to be really tricky to morph them separately and make it look good. Her own morphs seem to affect both at the same time so there must be a way to get her jaw to do whatever her head does, but I don't know what it is.

Suggestions?

I have learned the following things:

1. The morph generated by the restore detail script has to be set at a negative value. You set the original morph at 1, and the restore detail morph at -1, and spawn a new morph from the two of them combined.

2. The Ockham script that I mentioned, MTexcGroup, is sensitive to the number of figures in the scene. It only works if there is more than one figure in the scene. It doesn't work if there is only one. I don't know why, but this has consistently shown to be true.

3. Conversely, if you are exporting an obj to use as a morph target, make sure there is only one figure in the scene. That is what causes the error that I mentioned above where it thinks the geometry is wrong even though it is from the same figure. It gets confused somehow.

I have another question for Cage, if he is still around (or for anybody who happens to know the answer):  What is the VMF editor actually for? What happens when you combine the VMF files? As far as I can tell, you can't make a PBM with it, because TDMT will still only transfer a morph from one part to one part. So, what is the advantage in combining the files? (I trust that there is one, or you wouldn't have written the script, but I don't understand what it is.)

Thanks.

OT: hea cage any plans on a V4 Yvonne Craig (Barbara Gordon/Batgirl) character? I'm a fan of the actress.

@AsteroidLady:

The vmf editor is used to combine existing vmf files.  If you run two partial comparisons, using material screening, the resulting vmf files can be merged using the editor.

Some old TDMT tutorials I wrote years ago have been put back online.  They can be accessed via the Cage Page menu page:  http://www.morphography.uk.vu/dlcage.html

@Boni:

The only time I've ever used V4 for anything was when I was part of the Poser Place V4WM team.  I don't really have any plans to create characters for the figure.  I still tend to use Antonia pretty much exclusively, even though that figure has become unpopular with the user base.  My Antonia downloads page is still up (sadly the characters lack full utility while the Antonia Free Site remains down), and anyone who wants to try to use one of the available tools to port my Antonia morphs to V4 or another figure is welcome to do so.  :D

So that the material groups are appropriately matched. I see. Thanks.

Cage, thanks for the link. In your experience, what are the best settings for copying eyelashes? I am still having the problem with them warping. After reading the tutorial, I had the idea that I could do them separately, on different settings, and then merge the files.

IIRC, I had some luck with eyelashes (and eyebrows) using a closest vertex comparison and comparing the target eyelash material with the source head/skin material.  That allowed the eyelashes to pick up the motion of the nearest eyelid vertices, avoiding the fiddley need to match one figure's eyelashes directly to another figure's.  (After which you would want to merge the data files, yes.)

That can be one of the benefits of splitting up comparisons by material, that ability to use close-vertex comparison where it it useful and then use raycasting where it is useful.  Ears are a pain to compare between figures, using raycasting.  But they're a breeze, with closest-vertex.

Thanks. I assume the distance cutoff and number of influences should both be very low, on the eyelashes?
I was wondering what is the distance that is referred to in the distance cutoff. Is it an actual unit of distance in Poser space, or number of vertices away, or what? It would help me to make a better guess about what to set it at, if I knew what the distance in question actually is.

I think I usually have had the distance cutoff at 2 or 3. A setting of 1 will give you a 1:1 comparison, which is (or was, before Poser transferred morphs natively) useful for moving morphs between multiple instances of a figure or object within a scene. The more samples you use, the less precise the transfer is, on a vertex-to-vertex basis. I believe that distance in Poser Python is always in Poser Native Units.

The best setting I have found so far is 2, with a distance cutoff of .003. I've only been using Poser for about a year, so I'm still a relative newbie, and I still haven't read the manual either. I don't know what size a unit of Poser space is, I'll look into it. I was just thinking that if I put it on the wireframe view, I could see how far apart the vertices actually are, and make a better guess.
After matching the face with raycasting and the head without (because of the ears), on something close to the default settings, and then merging the files, when I apply a morph that affects the size or shape of the head, Apollo gets either a bulge or a dent on his head. This is after running it through restore detail.
What have I done wrong? It's like the vertices where the face and head meet have been left out.
Does what you said about raycasting not working well for ears still apply if it is the second run and I have gotten them perfectly matched by copying the head shapes of both figures and making them meet halfway? I'm thinking I need to do the head and face together, unless there's a remedy for that problem.

If the surfaces are matched fairly well, then raycasting is probably the better comparison option. I never had the tools to match areas like ears in a better than approximate manner, so I found the closest vertex approach suitable for such things.

That ridge along the edge is at the join between two merged comparisons, right? That's where you would want to use the "inclusive" edge setting for one side, and "exclusive" for the other. Then that edge belongs to only one of the areas, rather than both (or neither).

Yes, that fixed it, putting one on inclusive and the other on exclusive. Thanks for the tip. This one might actually be good enough to share. The only thing is the eyes and jaw don't morph with the head (I don't know how to make them do that) and sometimes the neck looks funny if the head morphs a lot.