David,

I had exported the model, ran the install plug-in, reset the compensates and assigned the CPs. Then I decided I didn't like the placement of the shoulder and arm geometry bones... so I moved them all around. Now I wonder if maybe it is not that simple.

I have backups ever way from Tuesday and can fall back to any point. Do I need to go back to the action poses (Posable install) and correct it there and then carry forward again?

Thanks,
Rusty

The easiest way to fix it would be to go back to your saved Action and fix it there, then finish off the installation. If you are unable to do that, then you would have to unhide all the bones in the model in order to move things around and then reset all of your compensates. You can do it either way, but the one that has the least amount of possible problems is going back to the installation Action.

Hope that helps, Rusty.

Remember that David? Well instead of a new version of the rig, my circumstance now... would seem to fit. I'll reapply the action to the 'mesh only' model (and then make my shoulder changes and re-save the action) and carry forward (export, install plug-in, reset compensates) then... not wanting to redo the CP assignments (no weighting done yet) use Transfer_AW to do this!

However, I've read over the Transfer_AW documentation and right now I have no idea how this will be done... and perhaps it might be more trouble than its worth unless all the special groups were already setup. Its late and I will need to re-read the doc and also read the WeightMover documentation which it refers to when I'm more awake (I bought WeightMover... damn so many tools I've bought that I never got around to using!!!).

Anyway, you seemed to click right on to this idea of using Transfer_AW. If you could sketch out what your thoughts were I'd appreciate it -- just the broad strokes. My brain says something like one skeleton with 2 mesh models (one hires, one lowres) all in one model window -- WeightMover transfered the lowres model's CP assignments to the hires model... but... how will this serve for the application we envision?

Cheers,
Rusty

Since there isn't a lo-res proxy, the CP Weighting transfer would be 1:1...try setting the "lores CP's for each hires CP" to "1", then set the "smoothing" and number of passes as low as it will go ("0" if it will allow it). It makes sense in my head, but I haven't tried it. If I get some time I'll do an experiment.

Hope that helps, Rusty.

David,

Hey, I can put a Rubik's Cube together in under 60 seconds (great party trick) but right now I'm feeling a bit stupid here. I've read all the documentation on Transfer_AW and WeightMover and... I do not see how this can be done. You must see some creative avenue or approach that I don't.

I see a plug-in that uses 2 sets of mesh overlaid in one model with one rig and it transfers the assignments of one mesh to the rig to the other mesh. How would you use this to transfer old model (the mesh, rig and assignments) to a new new model (the mesh and a new rig)? Hey, I'm hoping to hear myself say "Wow! I didn't think of that!".

Cheers,
Rusty

Now you did it, Rusty ....I went and started experimenting with this. I haven't gotten it to work 100% yet, but, I'm pretty sure it can happen. I'll need another evening of messing with it to see if I can get past the problem I'm having at this point...I've run out of play time for today.

What I did was to take Squetchy Sam, strip out the Poses, bones and Smartskins, made some "hi-res" named groups and saved this as a "hi-res" version of the model. I then re-opened Squetchy Sam and made the same named groups, but with "lo-res" in there instead and saved that out as the "lo-res" version of the model. Then, I opened the "lo-res" model, imported the "hi-res" model and ran the "Transfer_AW" plugin with all the settings set to "1". After the plugin ran, the CP's looked strange (I think that's normal), so I saved the model with "combined" in the name, closed AM, re-opened AM and reloaded the "combined" model. It appeared to assign the correct bones for the most part, but I got some strange percentages on this initial attempt. I'll have to do some more experiments to figure out how to correct them. I'll pick it up again tomorrow and let you know if I find out what I've been doing wrong.

Hope that helps, Rusty.

David,

I can see what you did... I can picture every step. I just don't see how any of it applies.

Lets say you have Squetchy Sam-A all rigged and working. Now a new version of the rig comes out. I guess you have to take a copy of the model, strip out the poses and bones (all the CPs become unassigned) and do whatever to reduce it to just mesh. Then you import the new version of the rig, import the action to scale the bones, export the model from it, delete the install poses and reset the compensates and so on to create Squetchy Sam-B which now has the new version of the rig but all of the CPs are unassigned, right?

So you have two models with two rigs:
1. the old Squetchy Sam-A with the CP assignments and the old rig
2. the new Squetchy Sam-B with the new rig, no assignments

You need to move the assignments in the old rig in the old Squetchy Sam-A model to the new rig in the new Squetchy Sam-B model.

How do you proceed?

If you import one model into the other right now you wind up with two meshes but also two rigs in the same model -- one set of bones becomes renamed. You can probably transfer the assignments but they are to the bones in the old rig. If, before you combine them you strip the rig from Squetchy Sam-A you lose the assignments. If you strip out the rig from Squetchy Sam-B you lose the new rig.

What am I missing?

Cheers,
Rusty

Rigging, huh? What a nightmare!
I would love to use the squetch rig but I just can't get my head around all that re-compensating and not being able to make adjustments afterwards, due to the complexity. I think you have the stamina of a bull, Rusty.

David, I think you should find a programmer to help make the installation a little more user friendly and then start selling this rig. I want it. I want it bad! But I don't want to spend days just installing it. Maybe I should have another attempt with the updated re-compensating instructions.

Sorry. That didn't help anyone did it.

Sigh... it is true that I could have rigged an army of characters using TSM2 in the time I've devoted so far to the Squetch rig and, I have yet to completely rig a single character model. It is true that the TSM2 or other far simplier rigs will work just fine -- have worked just fine for years. Will the Squetch rig be faster to animate or animate better? The face yes! But I'm not even using the face yet. The body? Yes, I think a tiny bit. The hands if nothing else. Is the Squetch rig over-engineered? God yes... completely over the top LOL! Just the way I like it!

Why then?

This feels like the future. Anzovin has dropped support for the TSM2. Its working for TWO. And, I made a decision.

The Squetch is killing my current project right now... months and not a single model rigged yet. That's the horrible truth. But I'm like a dog with a bone... I've decided and I won't let go.

Rusty

I have and use an elaborate system that reduces the pain of the compensating. But it is not something anyone else could use... I'm using AM and AM features in ways they were not meant to be used and the system is falling apart at the seams. Knowing the internals I can still use it but no one else could without some study. Also you need a fast PC to use it or it would only slow the process down.

r

Okay, here's what I think about future installations...Transfer_AW is going to be a big help.

First, I'm working under the assumption that the compensates will be automatically reset eventually...I don't know how far along that is in development. So, that is one of the time consuming things that will (hopefully) disappear.

Second, to speed up the weighting, all that is needed is a generic weighted model to use as a lo-res proxy...I can modify Squetchy Sam for that purpose. Then, if that modified Sam is part of the installation rigs, the proxy will be approximately in the right position on the character you're rigging to do a decent job of transferring the weighting. After running the InstallRig plugin and the resetting of compensates (which, eventually will be automated, I hope), you would run the Transfer_AW plugin to rough-in the weighting. For the roughing-in, I would only do the broad strokes, leaving out the face weighting since that is going to be more specific.

For Rusty's present problem, there's going to have to be a little more tweaking. Here's what I think might work. Instead of deleting all of the bones from the originally weighted model, select the mesh, copy it and paste it with bones into the new model. That will bring all of the "geom" bones with it and rename them with a "2" on the end. Run the "Transfer_AW" plugin, delete the old mesh, delete all of the "geom2" bones, save the model and close AM. Then, open the model in a text editor and do a "search and replace" to change all "geom2" references to "geom" and re-save the file. It makes sense in my head, but I haven't tried it.

Mark's suggestion might be the way to go for when there is a rig update though:



@Paul:



But then it would cost money, wouldn't it? Since it's a community designed rig, I don't think it would be right to charge anything for it. If someone makes a program that eases installation and sells that, then that would be fine, but the rig itself can't really be sold...in my mind, at least.

At this point, it is still harder to install than TSM, absolutely true. Eventually, it might get that easy, but it takes time...especially for free. It started off extremely difficult to install, is now moderately difficult (in my mind) and will eventually (with any luck) become easy enough for a new user...at least that's the goal.



LOL! Yes, that happens when trying to be all things to all people...just trying to make sure that the things the rig can't do is as few as possible. I realize it is off-putting to some people...nothing is stopping anyone from stripping out the things they feel they don't need, using another rig or making their own rig. I have isolated and stripped out sections when it has been requested and maintain a separate standalone version of the FACE controllers along with a few add-ons that could be adapted to any rig if necessary. I'm like you, Rusty...I want all of those toys in a rig I'm using.

I think that covers everything...if I missed something, let me know.

Hi,

I'm using the system I put together for easing the 'compensates' right now and as I'm using it I'm realizing that with a few changes in the order of the instructions this would be 100 times slicker. It is working better in V14 and, I think changes could be made to make it work better in V13. Additionally, if tags could be placed on the names of the constraints it would work even ten times better.

The draw back to my kludgey system is the number of times the batch file has to open and close AM; LOTS!. The above changes might reduce this but basically for each relationship that must be edited to adjust the constraints within it, I close/save/reopen the project file. I tried editing multiple relationships and it did not work but this was more because of the way I constructed the project file.

Right now, as is, it is actually dangerous because each filter exposes all of the constraints that have the same name and mistakes are easy to make. I think ALL of these constraints are adjusted anyway but I strictly follow the order of the instructions as published so these same-name constraints are offered up to different times, all of them showing making mistakes easy, when a change in the order things are done would likely offer all of them once to all be changed!

I should ask: is there someone working on some program to automate this right now (or planning to). This would of course be far far better, if it is possible.

I think I have enough room to upload the thing to my site now if anyone is interested in looking at the monster -- and it is a MONSTER! Basically it uses a huge amount of resources to save a little time. To give you some idea here is the batch file that drives it.

Click to view attachment

Again, it does this; AM opens with a bite size group of instructions displayed and filters that are matched to the instructions which 'get you closer to' the constraint you need to change (sometimes you get just the constraint you need and if tags were added to the constraint names, that's what you'd always get).

Cheers,
Rusty

Yes. You are right.
I have watched you developing the Squetch rig since the early stages of TWO and it has turned into a solid machine that is a joy to use. It is too easy to temporarly overlook the sterling work that Steven Cleary and Mark Strohben put into it and of course Mark Skodacek. Was there someone else? Martin and Noel and... well, yes, a whole community.

Free's good.

I can wait.

No I can't. No I can't! I must have it NOW!!

David, you have the stamina of two bulls...
and a chicken and a goat and one of those old London buses that go on well over a million miles.

Absolutely, I wouldn't rule anything out. It makes a lot of sense, bordering on genius really. To add to that, if there were a lo-res proxy included with the geometry bones installation, the Transfer_AW plugin could make roughing them in a lot faster the first time through.

Sorry I neglected to respond to that idea, Mark...in gathering my thoughts, things got jumbled a little in my head and it shook loose. I'll have to get something to strap my head together...there's a lot of junk rattling around in there.

The list is pretty long, lots of people were and still are involved. I made a list in the original "Squetchy Thom" thread that thanked everyone that I could think of up to that point.



Bullheaded I've been called...and a lot of other things not safe for television. Thanks for tipping the scale the other way a little, Paul.

(Deleted)
I just asked a stupid question before realising how stupid it was.

Falloff would work too...it may be the best way to go. I don't think the spline layout has to be the same for the proxy method, at least not from what I can tell. I haven't done enough experiments yet to determine how well it would work. The documentation I read was promising though. I know that Transfer_AW can designate groups as non-bending, for something that is supposed to be made of metal for instance, or a large group of scales or whatever...if I remember correctly, there's a way to do that with falloff as well though.

Sure, it's possible...I haven't tried it, but I'm sure it's possible with identical meshes.

Okay Mark and David, ignore my last post. See if I care. :-)

But I do need an answer for one question that was in there:



'It' being the reset compensate step. It will determine how much effort I put into my invention for myself.

As far as all the ideas flying back and forth, all the ones I can get my head around (most) sound great. Starting with a weighted template you can 'stamp' on your model (if you want) is a great idea! The text editor find/replace is unfortunate -- I don't mind, do it all the time but, it's borderline for some users. It knocks me down that Windows doesn't come with something like grep or awk that will process/find/replace text! Yeah I'm an old UNIX dude.

Cheers,
Rusty

I suppose that is possible. As long as I don't do anything to change the CP numbers and geometry bones don't change names. And as long as there is a single identifiable XML section in the model file. Good call.

r

But this could be very restricting unless you build all your models from an uber model. Build in all the extra geometry that you might need for different physiques and then just dicipline yourself to not add any extra geometry until it is rigged. I guess for a cast of human models who have almost identical clothing it would work and it would get you most of the way there even if you were to make changes.

I can't give you a definitive answer on the automating of the resetting of compensates, Rusty. I know that there were two different efforts toward that goal, but their status is unknown to me. I'm not a programmer (and don't want to be), or I would make the attempt myself. Both the efforts I heard about were being done in the spare time that those people had, so, there is no promise when or if it will ever get done. You know the saying (I assume), "You can have it fast, cheap or good...pick two". I don't have a problem waiting for "cheap and good".

Yep. But what we're talking about is making a decision to upgrade a working rigged model to a new release of the rig. There would have to be a compelling reason to do this. Furthermore, I always archive the post-rig model and as for post rig changes I keep good revision notes so I'd know what I was losing to fall back to the archived copy.

Cheers,
Rusty

You are tireless, Mark! Thanks for all of your (constant) hard work!

Okay, I figured I should post the results of my experiments with Steffen Gross' Transfer_AW plugin in case anyone else wants to try some things.

What I was trying to do was to transfer the weighting from a rigged model to an unrigged version of the same model. I kept getting some strange weighting that I couldn't explain. Things like a point with the weighting between three bones being, for example, 200, -100, -100...which was strange. The fix was silly and obvious (aren't they always), it was that both the "lo-res" and "hi-res" geometry were the same exact size. The assumption made by the plugin is that they aren't exactly the same size because the "lo-res" version would normally be made using fewer splines...and I'm not doing that here.

Here's what I did in steps:

1. First, I opened Squetchy Sam, which is already rigged, made a group that contained the entire character (excluding the FACE interface) and named it "lo-res". Then, I saved that version of the model with "lo-res" in the name.

2. I kept that version of Sam open, changed the "lo-res" group name to "hi-res", deleted all of the Poses, Pose sliders, Smartskins (the Smartskins should probably be left in place, I deleted them for this test though), bones and geometry for the FACE controls and saved this as a new model named "hi-res".

3. I then closed that Project to clear everything out, opened the "lo-res" version of the model, selected the "lo-res" group and scaled the mesh using the "Properties" menu to 101% on the 'X', 'Y' and 'Z' axes.

4. At this point, I imported the "hi-res" unrigged version of Sam, right mouse clicked on the name of the model in the PWS, selected "Plug-ins/Wizards/Transfer_AW", set the "Lores cp's for each hires cp" to "1", the "Smooth with hires cp's" to "1" and the "Smooth factor" to "0" and hit "OK".

5. To finish up, select the "lo-res" group, delete the associated mesh and extra groups and save the model out under a new name.


This did a pretty good job of getting things into the general ballpark except for the really tight places, like between the fingers near the base and some stuff in the face. There are things that would make it better, like using "exclusion" groups...which I didn't do for this test. To make an exclusion group, you wait until you have the "hi-res" model imported, make a group that includes both the "lo-res" and "hi-res" versions of the model (like selecting the right foot for both at the same time) and name the group with "exclusion" in it's name...like "right_leg exclusion".

This may or may not get used for situations like this, but it helped me understand how the plugin works...which could help in a number of other situations.

That's it for my play time today.

***** Suggestion *****

If the Squetch rig is stable right now (and I think it is) I would humbly suggest that you call the rig as it is right now V1.0 and freeze it. Then, all the changes you are thinking of, put in V2.0 alpha. Then, later on, drop support for v1.0 so you're not supporting 2 versions.

As an aside...
1. I don't mind you hijacking my thread but its a shame all this great discussion isn't under a more appropriate thread.
2. I have not had time to read all the recent posts and won't until later today (just wanted to get this in now).
3. FYI, it is very unlikely that I will move to a new version of the rig that contains massive changes until after my current project (I guess one reason for my suggestion). I'm too far behind and need stability and an end to the learning process... not, LOL, a journey!

Later,
Rusty

No time to review but that would be right... if that is so.

r

Yikes! Why would ever want to do that???? Its the same rig, the same model...????

Now I really have to go!!

r

Sorry Rusty. I removed my posts, as to not to hijack your thread any further.

Using a number like "1.0" turned out to be more confusing than using the date when I was initially working on the rig. What I've been doing is only releasing what I think is a stable version and naming it using the date is was released. If there are problems with that release, I try to correct them as quickly as possible and use the date the fix is released in the name to differentiate it from the previous problem version. With so many updates, you would get numbers like "1.04.56" or something else pretty crazy. I don't want to do a feature freeze because I want the new capabilities available as soon as possible...it also makes it easier for me to manage with all of the updates released, in case someone adds or modifies something and wants to get it put in (if we're all working from the same base, it makes it much easier).

As for the rig being stable...there are at least two updates being worked on at the moment. The current version is stable (as far as I know), but there are more upgrades possible (some you asked about, Rusty).



Sorry if that's what happened, Rusty...this stuff is loosely related to your original question, it's an attempt to answer/solve it.



A smart move, that's what you should do.



It was the first step in the experiment. If there are changes to the rig, this method could possibly allow the plugin to adjust the weighting to compensate for changes in the geometry bones.

Sorry if this became a hijack of your thread, Rusty. Discussions and experiments like these are sometimes necessary to devise a better mouse trap...I'll go back to what I was working on before this started.

Sorry for any hijacking on my part too, guys.

So, to answer your original question, you are better off going back to the installation action. Trying to go into bones mode and adjusting the bones manually can be very time consuming and errors could be made.

BTW, the hijacking on my part was coming up with a way to adjust the geometry bones only and to run an installation action with the control rig installer, to help you and others out who have assigned cps already. Since you don't seem to be interested in this process and you consider this hijacking, I will no longer pursue it. I don't have the time to waste, since I don't need this process.

You know I was kidding about the hijacking yes? I was doing it myself no? And why would I care?

But the discussion deserving a better thread title was true. :-)

Rusty

Mark,

Sounds very slick! That said, the Squetch rig contains techniques I've not dealt before with so some of what you say, for myself, is kind of out of context and I may be guessing wrong about what is meant. That said, 'sounds' awesome.

r

Mark is right. Sorry about the side trip on my part.

Answered long ago and already done! But thanks!



I understand.

Sincerely,
Rusty

Since you didn't respond to any of my posts, other than a reply to Paul I made, you didn't seem interested at all. You've only responded to my replys to the hijacking of the thread. The non response to my posts is more the reason for me deleting my posts, than the hijacking comment. That's what I find insulting, more than anything else. So I'll just take my toys and go play somewhere else.

As I attemted to convey in my post this morning (item number 2), there was little time to read and reply to things over the weekend and I wouldn't be able to review anything more until later this evening. Our family was visiting from LA this weekend through today and today we took my mother to Sea World. It never ocurred to me that this would cause problems on the AM forum and I'm sorry this resulted in any kind of insult to you!

I don't know what to say. I hope your okay.

Rusty

But found time for everyone else's posts, for the most part.

Don't sweat it Rusty, I'm just not interested in helping anymore. You seem to have things under control.

Mark,

Well I do sweat it Mark. I feel bad. To be perfectly honest, I had very limited time on-line and I read and replied to all that I could. I did read your posts but I did not reply because they went over my head -- in the time I had I could not absorb and understand what you were saying. I planned to come back to them when I had time to think about what you were saying. I don't know the internals of the rig like you and David. Like I was standing behind my chair reading and typing as fast as I could while, in most cases, my wife was already calling for me. If you knew the in-laws I had to deal with the last three days you might understand the mush that existed where my brain normally resides. Perhaps I should have replied saying this. When I did get back to it, your posts were missing but at first I didn't know this. I saw David was replying to stuff that wasn't there. I actually replied to one or two of your posts through David's replies... look and see. After a point I became very confused and I jumped forward trying to see what was going on and landed on your post saying you'd removed your posts. If you knew the weekend I'd just gone through you'd understand why that post of yours didn't land well with me. Like I did my best on all fronts the last 3 days.

I don't know about you but I'm blaming all of this unfortunate mess on my in-laws.

r