Chariots taking me for a ride!

[Bwian]

Been spending quite a bit of time trying to get to grips with how this can be done, and I have a solid plan that just does not seem to be working for me!

1) Took a selected set of horse animations, and created a dummy 'elephant' to try and get it so that a horse appeared as a horse...but was treated by the game wngine as an elephant. The only reason for doing this is so that I can have more than one rider on the chariot.

Sort of works, but I cannot make the rider actually appear! He is there, he shoots... but he refuses to display.

I thought it might be down to differences in the bone names ( saddle ... platform ) for the root... so I tried changing that in MS3D, and using the export extra bones tool to add an extra 'platform' bone. Exported teh animations OK...but the mesh converter will not convert the new model. Dies partway through.

Next effort will be to take a stock horse, add the crewman as an attachment, and try adding new bones for the chariot. If that still blows up the meshconverter, then I will try re-siting the tail bones to make the chariot wheel location. I would have to re-order the bones to attach the tail directly to the saddle 'root' bone .... more experimentaton needed!

[Bwian]

Update:

Tried with a stock horse, and re-used the tail bones to make a chariot pole.wheel setup. This worked fine, and would be great with a single rider on the chariot.

The stock animations, though, flap the tail around like a crazy thing. I need to have the chariot wheel a seperate bone so I can animate the wheels, or I would just attach the chariot to the root of the model.

Trying to change the animations now, but the animerge only seems to work one way. I can turn the modified MS3D into a mesh fine, using the mesh converter.

What I can't do is get the animations to work. I can merge the MS3D with the cas animation file, and load the result up into Milkshape. Then I tried to remove all the keyframing on the chariot part, so it remained still. This was fine...but I cannot turn this back into a CAS file. The animation utilities just don't do it.

I have not got any NEW bones... just some in new places.

Frustratingly, I have no idea why these things are not working

[Bwian]

Done a few tests to verify ... and animerge works fine with horses/elephants ... but the animation extract does NOT produce a cas file from the MS3D file. I did this with a stock unchanged horse.

This is a pain.... doubly so, since the Max plugin Caliban released does not work with Max as far as I can tell.... or at least, no-one seems to be reporting it as actually working!

[Bwian]

Next Test ... step by step as I do it :

The test model was 'mount_heavy_horse_lod0.mesh'. It was converted into an MS3D file using converter version 1.04, and the resulting MS3D file was opened in milkshape. The bones were 'zeroed' and the file was saved with the bones set at zero rotation.

Next, the skeleton was extracted using the latest release ( 1.1 ) of the utilities. The output in the skeleton file looks like this:

+0.0000000000  +0.0000000000  +0.0000000000    0  bone_H_Saddle
-0.0001452830  -0.0151966996  +0.4446060061    1  bone_Spine
-0.0006578590  -0.0245500989  +0.4198270440    2  bone_Spine1
-0.0008467390  +0.0139799975  +0.5470259190    3  bone_Neck
-0.0071678599  +0.5777239799  +0.4736510515    4  bone_Head
+0.2537220120  -0.2715969980  +0.5931140184    3  bone_RightArm
-0.0000306070  -0.3205270171  +0.0193389654    6  bone_RightForeArm
-0.0000488162  -0.4961239696  +0.0302742720    7  bone_RightHand
-0.0000340939  -0.3660830259  +0.0222389698    8  bone_RightFingerBase
-0.2555060089  -0.2715960145  +0.5923159122    3  bone_LeftArm
-0.0000304878  -0.3205200136  +0.0193104744   10  bone_LeftForeArm
-0.0000469685  -0.4961230159  +0.0303035975   11  bone_LeftHand
-0.0000362992  -0.3660819530  +0.0222394466   12  bone_LeftFingerBase
+0.0002688530  +0.0194744002  -0.2931849957    1  bone_Tail1
+0.0005695900  -0.3195210099  -0.3636940122   14  bone_Tail2
+0.2247990072  -0.2321140021  -0.1584759951    1  bone_RightUpLeg
-0.0000070184  -0.3785619736  +0.0039954484   16  bone_RightLeg
+0.0000009686  -0.4450979829  +0.0000083745   17  bone_RightFoot
-0.0000018179  -0.6024720669  -0.0000081211   18  bone_RightToeBase
-0.2339520007  -0.2229759991  -0.1583569944    1  bone_LeftUpLeg
-0.0000050515  -0.3876999915  +0.0031598955   20  bone_LeftLeg
+0.0000087619  -0.4450979829  +0.0000056773   21  bone_LeftFoot
+0.0000151992  -0.6024720669  -0.0000081062   22  bone_LeftToeBase

Looks fine so far, and this is correct with regard ot my understanding of how the thing should be built up

Next, being completely happy that my base MS3d file is a good one, I run the animmerge, and choose the solid MS3D file and .... for preference ...the 'run' animation for the horse. All animations were given the 20 frames per second option in the dialoge box.

The resulting MS3D file is called 'mount_heavy_horse_lod0_animby_fs_fast_horse_run.ms3d' and opens up fine in milkshape and the animation plays just fine.

Next ... I run animation extract. No actual output is created, but there are error messages in the DOS window.

(hope that has come out clear enough to be helpful)

This was run using a 100% stock model on a completely stock animation, with the latest animation toolset.

When I try to run a similar process for the elephant ... animerge fails and I cannot create an MS3D animation file to look at or convert back. Error in line 525 ... with a 'list index out of range'.

Camels convert to and from CAS file perfectly, though the gallop file I created went from 5Kb to 8Kb in the process. I have not yet tested it in game, but I am just pleased that content has been created.

[uanime5]

I can't help you with the technical details but I have an idea regarding the number of people in the chariot.

You could alter the chariot mesh so that it looks like there is a person controlling the chariot inside the chariot (the controller would be part of the chariot, not a seperate person, but would look like a seperate unit). So long as the controller moved slightly as the chariot moved (to make the controller look less like a fixed object) you'd only need one rider (independent of the chariot) in the chariot.

[KnightErrant]

Hi Bwian,

Bummer on the internet connection being down. This is good,
since no mods you've obviously isolated a conversion problem.
Can't run it tonight, but I'm e-mailing myself at work and can try
it at lunch tomorrow. Glad you're testing mounts, this is the weakest
part of the utilities because they have the most non-standard .cas
formats and this (hopefully) is what you've found.

Regards,

KE

[KnightErrant]

@Bwian,

Well, my early onset dementia must have been kicking
in when I worked on the mounts part. Could you edit
animationutilities.py and on line 1094 could you comment
out the line like the following:

#        float_vec3.fromfile( fidcas, 7 )

using the pound sign "#". What the heck that is doing there
I have no idea. Anyway, it ran and produced a .cas file.
If your editor doesn't do line numbers, do a search for

def getcasfooter

to get to the function, then search for

fidcas

That should get you to the offending line.

Mea Culpa,

KE

[GrumpyOldMan]

Hi All

Just floating through in a cameo appearance . Just a thought for anybody that's planning to make any new skeletons for chariots, or anything else for that matter. With the shader technology there is a maximum number of bones that a skeleton can have before it has to be split up into separate pieces for animation, without anything else in the shader you can have a maximum of 64 but with everything else going on in the shader a more realistic number would be 40. Since I haven't seen any skeletons in MTW2 go over this amount I'd suggest you use this as a working maximum.

So for example a chariot as one entity with four bones for the actual chariot body, driver with 20 bones and two horses with ~ 25 bones each would not make a viable skeleton

Something to bear in mind if you're designing something from the ground up, if CA don't have any skeletons above 40 then they probably won't have any built in mechanisms to split up skeletons over two calls to the shader.

Cheers

GrumpyOldMan

[Bwian]

Trickery will reduce that

the skeleton will only have a couple of extra bones to make the pole/axle setup. The horses will use the same skeleton, and the animation will just mean they are in step. Custs donw a LOT on both the extra bones and the need to revise animations for the horse. I don't see it being a problem, but it is good to know that there is a limit to work around, and something to make sure I don;t exceed!

KE ... I will test this tonight and let you know how it goes. If it works, I will have a rough chariot runnig around in the game before I go to bed tonight!

[Bwian]

OK ...base test went OK. With a standard horse structure, I animmerged to cerate an animated MS3D file, and then turned that back into a working animation file.

Next test is to add some extra bones, and see how the thing handles that ...

Imported the heavy horse lod 0 file, added 3 extra bones to make a chariot pole, base and axle. Rigged and added some very crude bits to make the mesh, and zeroed the bones.

Exported a skeleton, and manually re-ordered the bones in the skeleton file.

Exported the new skeleton with the 'export extra bones' button, and that ran fine. Recognised that I had added 3 bones in teh dos window..so thats good.

Seems to be working in game ...but my grasp of animation in MS3D is not good, and I am not sure exactly what I am getting in terms of end results. Will tweak some more, but I think the tools are working as designed, and the issues now are with me not the code!

[KnightErrant]

Ok, that's excellent!
I'll just post a note in the release
thread about commenting out the bad
line; not worth doing a whole new release
for that.

[Bwian]

KE... any word on the new skeletons for mounts? I am poised to test it if we can get to the bottom of the wierdness I spotted with the new mount skeletons!

[Abokasee]

Camels convert to and from CAS file perfectly, though the gallop file I created went from 5Kb to 8Kb in the process. I have not yet tested it in game, but I am just pleased that content has been created.

another victory for camels

[SigniferOne]

KE... any word on the new skeletons for mounts? I am poised to test it if we can get to the bottom of the wierdness I spotted with the new mount skeletons!

Tie it up to some elephants :P

[Bwian]

'Tie it up' is easy enough .... things go wrong when you start messing about with bone positions to a large degree. The animations on the horse started to go very badly wrong.

I can give it a go with some camels as a base, and see if it works better.

Elephants ... same. Need to test.

I went for the horse base, since that way I do not have to re-do all the animations for horse movements as well as adding those for the chariot!

[KnightErrant]

@Bwian,

This is going to be one of those weasely posts that people having
actual problems hate. Alletun has sent me his orc files, descr_skeleton.txt
battle_models etc. I haven't put them in game or anything but it seems
that this is not a bug in the mechanics of animations but something
we don't understand. (Not trying to weasel out on this, of course )
Remember the dwarves and trying to get their feet on the ground?
This seems to be like that. I've checked and Alletun does have the
basepose files transformed in descr_skeletons so it isn't that easy.
From what he described, the problem gets worse the more you depart
from a vanilla skeleton, even if it is a skeleton using the standard
bone names like the dwarf.

Scaling should not be a problem because that is only applied to bone_pelvis
(and the deltas but they are small). Already tried the tools->zero joints
but even though that changed things a little for Alletun it doesn't change the
basic problem. I know you always do the zero joints thing from the dwarves
so no easy answers to this one. I wish I could say I'll be on it but external
things have my time spoken for for awhile. (Plumbing problems, I'm told the
store that has what I need is where "the density of muggers and prostitutes
will exceed your comfort zone". My comfort zone cuts off at 1 so Saturday
should be interesting.)

If I can get the orcs in game at least I'll be looking at the same problem.
Maybe imagination will strike (or fire and brimstone, depending on the
GPU settings.)

KE

[Bwian]

I noticed the same thing as Alletun when I was doing my Orcs first time around. I say first time because I haven't gone back to them yet ...

What I noticed was far less extme than Alletun's and only really seemd to affect theweapon. This was moved away from the hand by some distance. I cured it easily enough by making the vertices of the weapon linked to the hand, not the weapon bone.

Alletun pm'd me about the issues he was having, and I explained what I knew to him, and told him what I did to sort it. Doesn't sound like it worked!

The error does seem to be related to how far away you have actually moved the bone from it's original start position. I would assume that the animation converter has to take each translation and apply it to the new animation to fit the bone as it now is. Logically, knowing how the skeleton is built up, you start with the root, nd work outwards down the tree....with each bone being dependant on the prevous one for it's actual location in virtual space.

So...the further you move the bone from it's start point, the more you would multiply any rounding off type errors. If you moved a bone a long way AND there were multiple bones dependant on it, the errors wouldcarry over into the start numbers for the next set of calculations.

Mf Orc and Chariot issues both seemed to be worst at the ends of the 'tree' with the Orc weapon and the horse front hooves. These pieces both move the most and are furthest from the root. Alletun seems to have a similar issue with the arms being wonky for him. All the rest looked fine in the pics he showed me.

Same was true of my chariot. The basic mesh and the body/head/neck areas were fine. Then as you progressed outwards, things went wrong.

Now...when we were testing this stuff, we worked on Dwarf skeletons. These had the dimensions REDUCED in terms of height, so this might actually serve to hide some issues with the calculations.

Confession here though.... several of my Dwarves also suffered from 'floating weapons' ... the Thunderer's especially.... and these guys have the gun attached to the hand bone, not the weapon one.

That aside...I have plenty of experimental experience playing with the tools, and have gone through Alletun's stuff as well, so I don't think there is anything glaring in what he has done to create the odd effect. I think we are looking at something that has been hidden under Dwarfishness, and is starting to rear it's head now we are doing other things.

Phew...

Oh yes.....and make sure you keep the doors locked and the windows up when you go shopping! Bring a friend and keep the motor running while you buy the stuff... and have the right money ready! Otherwise...while he rummages for change, your car loses it's wheels!

And ...yes ... I have lived in that kind of neighbourhood before

[Alletun]

Now...when we were testing this stuff, we worked on Dwarf skeletons. These had the dimensions REDUCED in terms of height, so this might actually serve to hide some issues with the calculations.

now thats pretty interesting. im doing a goblin with a smaller skeleton right now so im gonna experiment with it.

[KnightErrant]

Actually, not so scary. No muggers or pros at that time of
thie morning, just some homeless who actually got me to the right
place for some cigarettes. Sort of a Frick and Frack thing though,
"sorry have to buy the whole faucet for a replacement", "maa'm the
faucet is cemented into a brick wall on the side of my house", "not
my problem, would you like to talk to a manager", "unless he's a mason
I don't see how that would help", and it went downhill from there.

Yes, I think as well, this is something that was hidden under the dwarfishness.
The basepose problem came out clearly about the feet but with the larger
than normal skeletons something else is happening and I don't know what it
is. If worst comes to worst we can post to the powers that be (Caliban) and
see if that elicits some aid.

Alletun's experiment with a smaller goblin might help with quantifying the
problem size with the skeleton size.

KE

[Bwian]

Ahh....the joys of customer service! IF you had stayed longer, I am sure the manager would have tried to sell you a new wall as well!

Ahh...you want the X27914 wall mounted faucet. Yes...comes complete with it's own wall...only $970.00 plus tax. No...it's not available seperately. Oh and don't forget the fixing kit...and delivery charge. And the washer in the tap is extra.

It's all one big conspiracy to stop us doing it ourselves, and force us to pay tradesmen. Me...I blame the Masons

[GrumpyOldMan]

Hi Guys

Just passing through and I thought I'd add my two washers worth (in keeping with the plumbing theme). The strangeness in the bones seems to be a problem with the matrices we're applying. I've been totally immersing myself in dx9 and shaders at the moment but to get back to animation - each bone requires three matrices for successful animation. First a global matrix ie it's position in world coordinates, this is used (as an inverse) to find the local position of the vertex in relation to the bone. Next we have a local matrix, which is an offset matrix from the parent bone. Last is the transformation matrix for that frame ( and of course the final transformation matrix for the parent bone).

So , to correctly move a vertex to it's new location :-

new.vector = pose.vector * M.inverse.global.bone * M.offset.bone * M.current.frame.bone * M.final.parent

I haven't gone through anything yet but I'm just wondering if the strangeness is being caused by applying a M.global.bone instead of M.offset.bone. This would cause strange behaviour and placement.

Cheers

GrumpyOldMan

[Alletun]

Hi. I've been testing with a smaller skeleton ... and they still have weirdo arms Think im gonna backtrack my entire process to see if it's something fundamental im doing wrong.

[Bwian]

Thanks for the info GOM...sort of makews sense, but to a humble polygon pusher, it's not really something I understand in anything other than outline!

Oddly enough...I am much happier with plumbing! If I lived closer, I would nip round Ke's place and fix his tap while he fixed thje animations... but I can't stump up the cash for a transatlantic flight for a plumbing job! Besides...the missus would flay me alive if I did..she would much rather I finished off the tiling I started 6 months ago!

Alletun's stuff looks odd compared to what I was gettiong. I will double check my pipeline and see if I can duplicate his results. I got much better behaved models with only occasional glitches.

[Alletun]

I decided to finish my orcs today and the arms looked good i have no damn idea what i did different from when my arms looked funny. see here for a look at the arms:
https://forums.totalwar.org/vb/showt...26#post1640826

Im gonna see if i can find out what i did differently all the other time.

[KnightErrant]

Sorry to be MIA so long,

That's encouraging, Alletun's orcs look much better. GOM is probably on
the right track; if the conversion stuff isn't just plain broken then it probably
is a bad sequence of rotations that isn't apparent to me. Sorry, but I'm
basically a code hacker, still haven't gotten the bigger picture of the whole
3D graphics world in my mind.

Apologies re the plumbing theme, just got a new sprinkler system and my
whole world is now consumed with water problems and fixing the dug up
yard. Fall is coming, the blessed time when the d*** grass stops growing!

[Bwian]

Alletun ... which animation set did you apply your skeleton to? I am wondering if there are animation sets that 'work' and ones that just plain don't! I used the Maceman ones for my Orcs, and had the weapon bones go strange.

I was wondering which ones you used and whether you used the same one when it failed as opposed to when it worked. Might help to pin down an issue.

[Alletun]

Bwian i also used the mace animation. Now that i think of it, for the goblin and for some time with the orc i used a two handed axe animation, which you saw didn't work.

also when the animations suddenly worked i noticed that the weapons (and the shields) floated further out, as far as i remember you also had that problem? and as you did i also fixed it by assigning the weapons to the hand bones. Before, the arms stretched themselves freakishly to still hold onto these floating weapons. perhaps that is what caused the weirdo arms?

[Bwian]

This should help KE with the bug-hunting.

We have basically reached the same result independantly, and have both got most of the animation working fine, but with errors in the weapon bone. We both found the weapon bone had drifted away from where it was supposed to be ..and I think your stretched arms could well have been where the arm was going with the weapon.

We have both got workable models by linking the weapon to the hand bone, not the weapon bone.

Something defintiely goes wonky when you get to the end of the chain of bones!

[Alletun]

Dont know if it's supposed to be like this, but the exportskeleton is case sensitive. Couldn't understand why the cas file called "HR_lance_basepose.CAS" didn't want to convert .. had me scratching my head there for awhile, until i took .CAS down to lowercase .cas. Luckely only a small fraction of all cas files how uppercase letters.

just thought i'd mention it so others dont run into the same problem

[KnightErrant]

Sorry Alletun,

The case problem is probably mine; that is something I can look into.
However, the interesting thing is the bone assignments. Amazing how
something I was so into just a short while ago goes out of the mind.
IIRC I don't think I've ever messed with the weapon/shield bones; they follow
along for the ride. I remember Caliban commenting on how animations get
odd if you do stuff to the weapon bones and also GOM does things differently
with the extra bones in his mesh converter so I never tried to do things with
them. You may have hit on the problem here, assign things to hands but not to
weapon bones?

KE