I'm having a strange problem with PMD and JCM (joint controlled morphs).

I made some JCMs to square off the shoulders (poser collars) of the new Antonia114 figure. First I implemented the JCM as a delta injection. Because Antonia114 does not have any empty morph channels as yet, I hacked the cr2 to add some.

Using the hacked cr2 and my JCM pz2 I injected the morphs and ERC/JCM slaving code into the hacked figure. Everything worked fine.

Next I thought that others might like to try the JCM, but as the standard Antonia114.cr2 does not have any empty channels, I would have to implement it as a PMD injection.

Now I don't have much experience using PMD, and have never used it in conjunction with ERC, but it should be simple, yes?

Here is the procedure I followed. I made a morph stripped version of the Antonia114.cr2. I copied my morph channels from a saved cr2 and pasted them into the morph stripped figure. Loaded the morph stripped cr2 with my morphs in Poser, then saved it with "Use external binary morph targets" turned on, to generate the PMD file. I then constructed a pose file to load the PMD and inject the ERC.

In Poser, I applied my pz2 to load the JCM. The result can seen at the left in the image above. The JCM morph looks totally different from the same morph applied via delta injection. I'm baffled, I don't have a clue what's gone wrong. Any suggestions?
 

Here is a screen shot comparison of the morph channels in the rCollar from two different cr2 files. The cr2 on the left used PBD injection, the one on the right used standard delta injection. As far as I can see there is no difference, except for the names of the morph channel. Apart from that the  PDM channel was added at the bottom of the channel stack, where as the delta injection was near the top, would that matter?  Why am getting different results?

Perhaps I answered my own question in the last post. If morph channel is at the bottom of the stack it is being applied after the joint rotation instead of before (or does it work the other way round?). Any way this might well account for the difference. I can't think of anything else it could be. But if my supposition is correct this makes PMD next to useless!!!

P.S. I'm using P6, if that makes any difference.

Les, you should take a look at my earlier preview Antonia-096 (status changed to public so you can download it again in case you deleted or lost some of the files). Used JCMs injected via PMD there with no problem at all.

You're stacking your JCMs on top of my existing JCMs which are controlled by the same channels. Could that be causing the problems?

Oh, you exported the PMDs from with Poser itself? All bets are off then. :laugh:

Another idea: the image looks almost as if the morphs had been subtracted rather than added. Is it possible that you exchanged left and right by mistake? I mean let the morph for the left shoulder be controlled by the right shoulder zrot and vice versa?

@ odf,

Quote - Oh, you exported the PMDs from with Poser itself? All bets are off then.

Is there another way to export PMD?!

You are correct that it is on top of your JMC. In the image from my first post, in the right figure of that image, it is on top of your JMC, but in the center image (delta injection) it is also on top of your JMC, and there is not problem there. The morphs for the PMD came from a figure that had all other morphs removed. So the PMD itself should not be corrupted by any extraneous morphs.

The proof of the pudding seems to be in the image from this post. The figure at the right is the standard Antonia114, center is the same figure after PMD injection, left, this is the interesting one. The center figure was saved to disk, then the channel from the lCollar was taken from the bottom of the channel stack, where the PMD injection had placed it, and moved to the top of the stack, no other changes were made. Moving the channel to the top of the stack cured the problem for the lCollar, and the morph now expresses correctly. So it seems to be the position in the stack that causes the problem.

In another experiment I deleted (not just REMed) all your morphs from the chest, collars, and shoulders. I also Removed the ERC slaving from my morphs, and set them directly with the dials. When the shoulders (upper arms) were rotated so the arms were at her sides the problem still manifested, even with no ERC. Again moving the morph channels to the top of the stack cured the problem.

Les, I hear you. Just checked the Antonia-96.cr2. The JCM are in by default, so the channels would be at the correct place already.

It is probably worth noting the following points. The problem only relates to injecting a new channel, and is because the channel is added at the end of the stack. The problem manifests in the collars (in this case), and seems to be only in the part of the collars that is bent (note bent, not rotated). The bend is via PJs in response to another actor, the shoulder. This raises a number of possible permutations, that I really should check out, but I'm not sure I have the time or patience to do that at the moment.

Right now I'm thoroughly sick of PMD injection, and don't ever want to touch it again. I'll probably moderate my opinion in a few days and realize it is very useful, but that's how I feel right now after wasting umpteen hours on this problem.