Animations

Well just to let you guys know: I started looking into the animation files a bit.
I already wrote a python script that can extract them from the .idx/.dat file - it seems they aren't packed or anything since Verci's old cas importer was able to import them (well they didn't work properly, but it didn't crash, either).
So what I'm planning to do right now is to figure out how the .cas files are described by leeching a bit of know-how from the cas importer :P

As for the idx file (only pack.idx):
It has a header consisting of the string "ANIM.PACKV:|", then a long 9 and then a long with the number of entries.
Each entry then consists of:

Bytes Variable type Explanation

0-3 long strlen+10 of the tag (below)

4-7 long byte offset in pack.dat

8-11 long byte length in pack.dat

12-15 float Unclear so far, often 1.0 (first four additional bytes of the strlen+10)

16-25 - These are the other 6 additional bytes in the strlen+10*

strlen String tag (qualified path for the animation)

*They seem to be 0x3F**00140014 - 0014 is 5120 as a short which sounds probable. No idea about the first two bytes.

Re: Animations

Just briefly, yes the old Verci script works halfway. It can extract, if you replace the title string above with the RTW version, and the animations are half-way ok, but still not 100% good. Caliban wrote a post somewhere here that the reason is that in M2 bones are weighted, whereas in RTW they weren't.

Caliban, is it a secure fact that once the animations are extracted and work properly, they don't have to be put back into the packed files again, and work using that anim_db switch you once posted?

Re: Animations

The bones section at the top of the .dat file has
the idle pose of the standard skeleton in the first three
floats after the null terminated ASCII string. There's 19
floats in each data section between the bone strings,
3 are the idle pose offsets, 3 probably 0.0 rotations, and
who knows for the remaining 13 floats.

Edit: Sorry, not the first three. Skip the first float after the
bone string, then the next three floats are the idle pose offsets.

Re: Animations

KE: Are you talking about the skeletons.dat? I was looking at the pack.dat :laugh4:
I got the impression we could still extract the skeletons with xidx?

Edit: By the way, don't forget ICQ XD

Re: Animations

Oops, my mistake, wrong .dat.:laugh4:

Edit: What does ICQ XD mean?

Re: Animations

Well I don't really know how the skeletons work at any rate.
Do they just link the bones in the .cas/.mesh files to the anims?

Re: Animations

Don't know, the only .cas files I've worked through are the
_idle ones in animations/engine and they're short.
They just have the idle pose for rigging the skeletons.
I was digging for where the human animations were.

Re: Animations

Hi KE et al

Quote:

Originally Posted by KnightErrant

Don't know, the only .cas files I've worked through are the
_idle ones in animations/engine and they're short.
They just have the idle pose for rigging the skeletons.
I was digging for where the human animations were.

I've just about finished the CAS figure to Mikshape converter, and I've already got code to convert m2tw animations to Milkshape, including the quaternion to euler changes. It shouldn't take too long (famous last words?) to do the re-convert.

The main issue is that the way the figure cas' are set up it's easier to convert to Milkshape ascii.

Cheers

GrumpyOldMan

Re: Animations

Awww, no fair, does this mean the adventure doesn't continue
with animations?

@alpaca
Sorry for the hijack.

Re: Animations

Hi KE

Quote:

Originally Posted by KnightErrant

Awww, no fair, does this mean the adventure doesn't continue
with animations?

Yes, KE, there will be more adventures, we've only just begun to convert. I finally ironed out the conversion of RTW cas files into a milkshape file with a m2tw skeleton (I'll put up a pic in the mesh thread soon). You'll be happy to know that they work ok with vanilla m2tw animations. More adventures planned
- get the milkshape animations back into cas (pass me the 4 pound hammer :laugh4: )
- once we've done that, we can look at straight converting of RTW animations to m2tw
- there's always the bogey of adding new skeletons (once we get the skeleton stuff straight in our heads, it's doable because Caliban said they rebuilt the skeletons and animations on the fly so they're not hard coded)

Lots of stuff to do yet.

Quote:

Originally Posted by KnightErrant

@alpaca
Sorry for the hijack.

Me too

Cheers

GrumpyOldMan

Re: Animations

Haha well if you already got it figured out :laugh4:
So you already extracted the pack.dat? Why didn't you say?

Anyways, I'll have a look at the skeletons then.

Edit: By the way, does anybody have any resources on bone weigthing?

Re: Animations

Well I have to start learning somewhere, so I wrote an extractor
to pull the cas's out of pack.dat using the offsets in the index file.
Pulled out 30 to look at. The first 5 bytes are the header where
the first short is the number of anim frames. The other 3 bytes are
14 00 14 hex or 20 0 20. The only question is what comes first,
quaternions or pose data. I saw some 0.0 0.0 0.0 1.0 values so that
answered two things: quats come first and CA uses the {q1, q2, q3, q4}
convention where q4 is the special one. I transformed them to Euler
angle using the 321 convention and got this for the first anim frame
for CR_spear_charge.cas:

Code:

Number of animation frames: 11 Remaining header: 20 0 20 Quaternion data and maybe Euler angles (using 321)? q1 q2 q3 q4 sum of squares yaw (deg) pitch (deg) roll (deg) ============ anim quaternion frame 0 ============ +0.2538955510 +0.1681092978 -0.0226481762 +0.9522412419 sum squs = +1.0000000095 +2.5657103993 +19.3696683429 +30.2967888994 -0.3953365684 -0.0070954142 +0.1987117827 +0.8967565298 sum squs = +0.9999999935 +21.4590405866 +8.3019773363 -46.0049588155 +0.1825654358 +0.0634186864 +0.0113932779 +0.9810800552 sum squs = +0.9999999497 +2.6276101260 +6.9081227284 +21.2413852449 +0.0005502258 +0.0181907974 -0.2293124646 +0.9731827378 sum squs = +1.0000000555 -26.5251008404 +2.0435025817 -0.4169129706 +0.0379701219 +0.0379873700 -0.0190575924 +0.9983747005 sum squs = +1.0000000050 -2.0214710134 +4.4332999853 +4.2777858328 +0.1421776563 +0.1788186878 +0.0416587740 +0.9726633430 sum squs = +1.0000000412 +8.0495681889 +19.6342635762 +18.0275697089 -0.2416000813 -0.3342688680 +0.0259645525 +0.9106149077 sum squs = +0.9999999436 +15.0756420060 -36.6006764153 -34.7298720077 +0.0000000000 +0.0000000000 +0.0000000000 +1.0000000000 sum squs = +1.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +1.0000000000 sum squs = +1.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 -0.0013705824 -0.0047476389 -0.0055176513 +0.9999725819 sum squs = +1.0000000275 -0.6315546125 -0.5448992324 -0.1540581944 -0.0002179843 -0.0194163416 -0.3695584238 +0.9290046096 sum squs = +1.0000000350 -43.3998593387 -2.0766736500 +0.7995938917 -0.0381897017 -0.6430811286 -0.0234443955 +0.7644858360 sum squs = +1.0000000244 +4.4177386253 -80.0780864235 -9.4322343545 +0.0172815081 +0.1676424295 -0.1667936444 +0.9714819789 sum squs = +0.9999999898 -19.7157479464 +19.3591213839 -1.3571896291 -0.0011378765 +0.0047027008 +0.0202233475 +0.9997837543 sum squs = +0.9999999494 +2.3170456268 +0.5414182810 -0.1194701201 -0.2264718562 -0.1660457850 +0.4406757355 +0.8526102304 sum squs = +1.0000000133 +56.0534307300 -4.7922679610 -32.3030476656 -0.0418538116 +0.8028841019 +0.0247014631 +0.5941508412 sum squs = +1.0000000070 +7.4251721953 +72.9670051832 +1.9701113044 +0.0022718122 +0.0688987225 +0.0323999561 +0.9970948100 sum squs = +1.0000000124 +3.7581457004 +7.8887411361 +0.5203097663 -0.3605583012 -0.4344316721 -0.1733115017 +0.8069881797 sum squs = +0.9999999651 +3.4140766971 -55.7042659859 -49.9541032693 +0.2243260294 +0.1274575591 +0.0513146892 +0.9647793770 sum squs = +1.0000000404 +9.2202763455 +12.8802680946 +27.2221190368 -0.0548719205 -0.0131644774 +0.1631041616 +0.9849938154 sum squs = +1.0000000151 +18.8304236382 -0.4603295875 -6.4533948925 ============ anim quaternion frame 1 ============

The next block should be the position or pose data so its done as
triples:

Code:

Pose data: ================== anim pose frame 0 ============ +0.0000000000 +0.0000000000 +0.0000000000 +0.0952388272 +0.0007522855 -0.0000000016 +0.0225613043 -0.4644488990 +0.0143959904 +0.0241626762 -0.3995066285 -0.0316338763 -0.0000000072 +0.2124620080 +0.0000000010 -0.0002945314 +0.2115578204 +0.0000000317 -0.0000617032 +0.2349731475 +0.0000000667 +0.0003562687 +0.0108105456 -0.0034470414 +0.0016836965 +0.1178482845 -0.0744606182 +0.0132546313 +0.1300114095 -0.0273839347 +0.1653589308 -0.0517836995 +0.0034832763 +0.3022064567 +0.0111386590 -0.0137692019 +0.2838370204 -0.0030556649 +0.0263375491 -0.0102220755 +0.1300113946 -0.0273839198 -0.1678023189 -0.0517838039 +0.0034833569 -0.3021733761 +0.0111954454 -0.0144314468 -0.2838012874 -0.0032004488 +0.0267044865 -0.0952387974 +0.0007523632 +0.0000000220 -0.0216093566 -0.4641438425 +0.0230784863 -0.0250774939 -0.3986378312 -0.0406115353 ================== anim pose frame 1 ============

This looks right because it matches the standard skeleton:

Code:

0.0 , 0.0 , 0.0 -0.095239 , 0.007520 , 0.0 -0.022561 , -0.464449 , 0.014396 -0.024163 , -0.399507 , -0.031634 0.0 , 0.212462 , 0.0 0.000295 , 0.211558 , 0.0 0.000062 , 0.234973 , 0.0 -0.000356 , 0.010810 , -0.003447 -0.001684 , 0.117848 , -0.074461 -0.013255 , 0.130011 , -0.027384 -0.165359 , -0.051784 , 0.003483 -0.302206 , 0.011139 , -0.013769 -0.283837 , -0.003056 , 0.026338 0.010222 , 0.130011 , -0.027384 0.167802 , -0.051784 , 0.003483 0.302173 , 0.011195 , -0.014431 0.283801 , -0.003200 , 0.026704 0.095239 , 0.000752 , 0.0 0.021609 , -0.464144 , 0.023078 0.025077 , -0.398638 , -0.040612

(Don't worry about the signs on the x values, The standard skeleton
was made to work with Milkshape so there's always to x -> -x transform
for that.)

So far, so good. The problem is I still have 76 floats left over that look like
this:

Code:

What the heck are these data: We have 76 leftover floats +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0145724751 +0.0491957590 +0.0002147975 +0.0301976539 +0.1019454375 +0.0004450644 +0.0437174141 +0.1475874335 +0.0006442972 +0.0519736372 +0.1754599810 +0.0007659923 +0.0510471202 +0.1819434315 +0.0005802182 +0.0442738384 +0.1494659334 +0.0006524611 +0.0348668732 +0.0312059894 +0.0020629363 +0.0234865602 -0.0744927898 +0.0031001847 +0.0119119827 -0.0558995418 +0.0018968056 +0.0000000000 +0.0000000000 +0.0000000001 +0.5000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000001 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000 +0.0000000000

For a big 4.5 second animation like CR_spear_die_galloping.cas it's
worse, I have 556 floats left over, mostly zeroes or repeating values
but no idea what they are. The pose data on the last anim frame
still looks ok, I don't think I've slipped data by miscalculating the
frame offsets.

Another GrumpyOldMan question is this:

If I have my rotation matrix parametrized by the quaternions as

Code:

_ _ | | | q1^2 - q2^2 - q3^2 + q4^2 2 * ( q1*q2 + q3*q4 ) 2 * ( q1*q3 - q2*q4 ) | | | | 2 * ( q1*q2 - q3*q4 ) -q1^2 + q2^2 - q3^2 + q4^2 2 * ( q2*q3 + q1*q4 ) | R = | | | 2 * ( q1*a3 + a2*q4 ) 2 * ( q2*q3 - q1*q4 ) -q1^2 - q2^2 + q3^2 + q4^2 | | | |_ _|

and similarly for Euler's

Code:

_ _ | | | cos(psi) cos(theta) sin(psi) cos(theta) -sin(theta) | | | R = | - - cos(theta) sin(phi) | | | | - - cos(theta) cos(phi) | |_ _|

is that Mete's convention or is it one of the other 11 sets?

Re: Animations

Hi KE

I didn't quote your reply, so I hope I get everything clear for you.

The next lot of floats after the skeleton are global position vectors for the bones, so number of frames by number of bones by x, y, z. (I get a lot more than 76 floats here :dizzy2: are you sure you just haven't printed off the first three float values of each frame's worth??). Although the local positon value hasn't changed, you have to remember this is preprocesed to plug into the engine. The only significant one to worry about is the y value for the first bone this give the movement perpendicular to the ground plane.

Next there is number of frames by x and z 'incremental' movement for the first bone (pelvis). This shows you how much the figure moves parallel to the ground plane each frame.

Below is the function I'm using for Quat to Euler conversion :-

Code:

QuatToEuler(a4,a1,a2,a3);data read in from original file - a1,a2,a3,a4 Function QuatToEuler(w#,x#,y#,z#) Local sint#, cost#, sinv#, cosv#, sinf#, cosf# Local cost_temp# sint = (2 * w * y) - (2 * x *z) cost_temp = 1.0 - (sint * sint) If Abs(cost_temp) > QuatToEulerAccuracy cost = Sqr(cost_temp) Else cost = 0 EndIf If Abs(cost) > QuatToEulerAccuracy sinv = ((2 * y * z) + (2 * w * x)) / cost cosv = (1 - (2 * x * x) - (2 * y * y)) / cost sinf = ((2 * x * y) + (2 * w * z)) / cost cosf = (1 - (2 * y * y) - (2 * z * z)) / cost Else sinv = (2 * w * x) - (2 * y * z) cosv = 1 - (2 * x * x) - (2 * z * z) sinf = 0 cosf = 1 EndIf ; Generate the output rotation roll# = -ATan2(sinv, cosv) ; inverted due to change in handedness of coordinate system pitch# = ATan2(sint, cost) yaw# = ATan2(sinf, cosf) End Function

Don't worry about the QuatToEulerAccuracy, I thought it was a bit fancy-shmancy, so I just deleted the constant but haven't updated the function yet. Read it as '0'.

I wrote this code a while ago and it exports the animation but facing in the -z direction. Shouldn't take much to fix but the Menu/Tools/Mirror All fixes the orientation.

Now the results come out as calculated roll equalling pitch, calculated pitch equalling yaw and calculated yaw equalling negative roll. I've never got around to cleaning it up as long as I was getting the right numbers out:embarassed: :embarassed: .

If you want to take the language for a shareware ride I can pass the code onto you.

Cheers

GrumpyOldMan

Re: Animations

Hi alpaca

Quote:

Originally Posted by alpaca

Edit: By the way, does anybody have any resources on bone weigthing?

You could have a look at

http://www.chumba.ch/chumbalum-soft/...84&postcount=1

which is Mete from Milkshape's very basic instruction on using the bone weight editors. Have a look through the forum, I seem to remember there being other posts (by Wes H??) on working with bone weights on Sims2 figures.

Cheers

GrumpyOldMan

Re: Animations

I couldn't see if this was seen earlier, so I post this anyway.
I had a wee look in Verc's script, and the second short in the header is the number of bones.
About the fifth byte, I think he actually skipped it.

Re: Animations

@GOM
Python is being unkind to me tonight and I can't find
my bug yet, but I still wanted to clarify two things:

(1) Went through the quaternion to yaw, pitch, and roll
and I think I'm following the same conventions. Meaning I
followed my quat vs euler matrices and worked through your
QuatToEuler function and I arrive at the same ATan2
relations for the angles. So I think
your code for rotations x, y, z -> roll, pitch, yaw is right
without having to interchange roll->pitch, pitch->yaw, and
yaw->roll. Does Milkshape really require this (excluding the roll
minus sign)?

(2) Byte counts: My example last night was for the CR_spear_charge.cas
which is a small one (7 KB) . If I count bytes this way

Code:

5 - header nframes*NBONES*4*4 - quaternion block nframes*NBONES*4*3 - pose block nframes*4*3 - global block (only bone_pelvis) nframes*4*3 - relative block (only bone_pelvis) 40 final bytes, or 10 float - leftovers

Last 8 bytes always seem to be FF FF 0F 00 00 00 00 00, so maybe a
terminating signature.

This seems to square with the bytelength in the index file so I think
this is ok. For 11 anim frames I get 33+33+10=76 floats that I had last
night. So the global block and the relative block track the movement
of the single bone_pelvis to give the overall traversal of the model, and
the quaternions give the relative animation?

The game then seems to be, merge a .cas with a .ms3d and write the
anim stuff to the joint section, suitably merging the global and relative to
the pose data, since Milkshape doesn't have extra places for this, and
put the transformed quats into the rotation angles part. Reverse process
is separate the global and relative from the position vectors, reconvert
angles back to quaternions and write out a new .cas.

Is the jerky nature of the unpacked .cas files simply that the number of
bones in the standard skeleton for M2TW is different from RTW so the anim
frames aren't coming out right? Caliban has hinted that someone could
simply (simply?) change the MaxScript script to fix this but no one has
stepped to the plate.:dizzy2:

Back to fighting Python, I can see the data I want in xvi32 put it won't print
out for me. This is finally the end for me, it's easier to simply read the data
in hex than it is to try to print it out.

Re: Animations

@KE

I've just got to close down for a while before I throw the mouse through the monitor, getting very sick of looking at hex for RTW cas files.

It's terrible to see two old addicts in the clutches of their vice, isn't it?:laugh4: :laugh4: :laugh4: :laugh4:

The jerky movement is indeed just a matter of trying to cram RTW rotations onto a m2tw skeleton, it just requires a simple renumbering operation I imagine within the plugin/script - disregarding the cloak bones and taking into account the clavicles, jawbone and eyebrow.

Cheers

GrumpyOldMan

Re: Animations

I know the feeling... I think I'll slow down on anims and finish
writing instead. Hate leaving loose strings about. Thanks for
the info re jerky animations, that's what I thought from Caliban's
comments.

Re: Animations

By the way, Verc's animation script, though fantastic, was buggy at least in two respects:

1) it produced flickering animations for certain skeletons, wherein a bone would very quickly rotate 180 degrees back and forth. Sometimes hoplites would flicker their shields, and sometimes horses would flicker with their legs.

2) it could not modify most turn animations. The actual turn often proceeded in-game smooth enough, but the soldier would instantly turn back to his original position, and then turn back to his new direction again, and then back, etc etc. He couldn't complete turning back and forth, and you would have to order the entire unit to go elsewhere, upon which he'd stop and run with the rest of his unit. It's almost as if the script didn't register that the angle of the soldier's orientation was changed to the desired angle, and the engine kept trying to turn him; or it could be something else.

Re: Animations

C'mon guys, no turning back now:whip:

Re: Animations

Naw, not turning back, just finishing up things before the plunge.
Found out a little more today. The 8 floats before the final 8 signature
bytes follow this form:

first is the length of time of the animation, e.g. an 11 frame animation
is 0.5 seconds (1/20 of a second per frame, first frame starts at 0.0), a
91 frame animation is 4.5 seconds, etc.

second SEEMS to be the maximum excursion of all the entries in the relative
block or, same thing, the maximum excursion of bone_pelvis.

third, fourth, and fifth are just the last frame of the relative positions,
sort of like a registration position.

sixth, seventh, and eighth: no clue. They're not the same as any other
entry when they're non-zero. For the stationary animations (non walking,
running) they're usually zeros.

Anyway, just finishing up a little utility to convert binary .cas files to an
editable text form that can be converted back to .cas by the same
converter. Not the way I would want to do it, but it might be fun for
people to play with until the Milkshape stuff is sorted out. Just need to
add a script to turn Euler angles into quaternions that people can use
to change entries.

Re: Animations

Hi KE

Here is some more code to work with quats and eulers:-

Code:

; Change these constants if you notice slips in accuracy Const QuatToEulerAccuracy# = 0.001 Const QuatSlerpAccuracy# = 0.0001 ;convert a Rotation to a Quat Function EulerToQuat(pitch#,yaw#,roll#) ;NB roll is inverted due to change in handedness of coordinate systems Local cr# = Cos(-roll/2) Local cp# = Cos(pitch/2) Local cy# = Cos(yaw/2) Local sr# = Sin(-roll/2) Local sp# = Sin(pitch/2) Local sy# = Sin(yaw/2) ;These variables are only here to cut down on the number of multiplications Local cpcy# = cp * cy Local spsy# = sp * sy Local spcy# = sp * cy Local cpsy# = cp * sy ;Generate the output quat w# = cr * cpcy + sr * spsy vectorx = sr * cpcy - cr * spsy vectory = cr * spcy + sr * cpsy vectorz = cr * cpsy - sr * spcy End Function ; convert a Quat to a Rotation Function QuatToEuler(quatx#,quaty#,quatz#,quatw#) Local sint#, cost#, sinv#, cosv#, sinf#, cosf# Local cost_temp# sint = (2 * quatw * quaty) - (2 * quatx * quatz) cost_temp = 1.0 - (sint * sint) If Abs(cost_temp) > QuatToEulerAccuracy cost = Sqr(cost_temp) Else cost = 0 EndIf If Abs(cost) > QuatToEulerAccuracy sinv = ((2 * quaty * quatz) + (2 * quatw * quatx)) / cost cosv = (1 - (2 * quatx * quatx) - (2 * quaty * quaty)) / cost sinf = ((2 * quatx * quaty) + (2 * quatw * quatz)) / cost cosf = (1 - (2 * quaty * quaty) - (2 * quatz * quatz)) / cost Else sinv = (2 * quatw * quatx) - (2 * quaty * quatz) cosv = 1 - (2 * quatx * quatx) - (2 * quatz * quatz) sinf = 0 cosf = 1 EndIf ;Generate the output rotation vectorx = -ATan2(sinv, cosv) ; inverted due to change in handedness of coordinate system vectory = ATan2(sint, cost) vectorz = ATan2(sinf, cosf) End Function ;use this to interpolate between quaternions Function QuatSlerp(Ax#,Ay#,Az#,Aw#,Bx#,By#,Bz#,Bw#,t#) Local scaler_w#, scaler_x#, scaler_y#, scaler_z# Local omega#, cosom#, sinom#, scale0#, scale1# cosom = Ax * Bx + Ay * By + Az * Bz + Aw * Bw If cosom <= 0.0 cosom = -cosom scaler_w = -Bw scaler_x = -Bx scaler_y = -By scaler_z = -Bz Else scaler_w = Bw scaler_x = Bx scaler_y = By scaler_z = Bz EndIf If (1 - cosom) > QuatSlerpAccuracy omega = ACos(cosom) sinom = Sin(omega) scale0 = Sin((1 - t) * omega) / sinom scale1 = Sin(t * omega) / sinom Else ;Angle too small: use linear interpolation instead scale0 = 1 - t scale1 = t EndIf vectorx# = scale0 * start\x + scale1 * scaler_x vectory# = scale0 * start\y + scale1 * scaler_y vectorz# = scale0 * start\z + scale1 * scaler_z w# = scale0 * start\w + scale1 * scaler_w Return w# End Function ; result will be the same rotation as doing q1 then q2 (order matters!) Function MultiplyQuat#(Ax#,Ay#,Az#,Aw#,Bx#,By#,Bz#,Bw#) Local a#, b#, c#, d#, e#, f#, g#, h# a = (Aw + Ax) * (Bw + Bx) b = (Az - Ay) * (By - Bz) c = (Aw - Ax) * (By + Bz) d = (Ay + Az) * (Bw - Bx) e = (Ax + Az) * (Bx + By) f = (Ax - Az) * (Bx - By) g = (Aw + Ay) * (Bw - Bz) h = (Aw - Ay) * (Bw + Bz) w# = b + (-e - f + g + h) / 2 vectorx# = a - ( e + f + g + h) / 2 vectory# = c + ( e - f + g - h) / 2 vectorz# = d + ( e - f - g + h) / 2 Return w# End Function

Should be enough to fill in your weekend :laugh4: :laugh4:

edit :- If you haven't looked already

http://www.gamasutra.com/features/19...ernions_01.htm ,

most of the code is based on this.

Cheers

GrumpyOldMan

Re: Animations

Many thanks for the references, most of the quaternion stuff I've
got is from memos at work. There's a group that uses them for attitude
control for reentry vehicles in simulations which doesn't really
translate easily to moving bones around. Oh, it did hit me yesterday about
the cyclic permutation for the yaw, pitch, and roll, of course Mete's
convention has the y-axis as the vertical. D'oh!

Edit: Went and joined, lots of refs on quaternions in gaming. Even a few
physics ones!

Re: Animations

Well, I tested unpacking all the cas files and found examples
violating almost all of the rules from a couple posts back.
The first of the last eight floats is the animation time in
seconds. The header in descr_skeletons.dat say there will always
be an odd number of frames so this number will always be a multiple
of a tenth of a second. Otherwise the remaining seven floats I have
no rules for.

The header bytes aren't always 20 0 20 (14 00 14 hex), they have lots
of variations especially for the stratmap anims. The final 8 bytes aren't
always FF FF 0F 00 00 00 00 00 they have lots of variations too.
Here's a chunk with some stratmap examples from the extraction log:

Code:

31 20 20 data/animations/MTW2_Swordsman/MTW2_Swordsman_at_mid_c_stab_v1_s1_slashrl_fail.cas 39586546 18149 255 255 15 0 0 0 0 0 47 20 20 data/animations/MTW2_Swordsman/MTW2_Swordsman_at_mid_c_stab_v1_s1_slashrl_success.cas 39604695 27493 255 255 15 0 0 0 0 0 23 20 20 data/animations/MTW2_Swordsman/MTW2_Swordsman_at_overhead_c_stab_fail.cas 39632188 13477 255 255 15 0 0 0 0 0 29 20 20 data/animations/MTW2_Swordsman/MTW2_Swordsman_at_overhead_c_stab_success.cas 39645665 16981 255 255 15 0 0 0 0 0 55 20 1 data/animations/MTW2_Swordsman/MTW2_Swordsman_basepose.cas 39662646 19625 1 0 0 0 0 0 0 0 41 20 20 data/animations/MTW2_Swordsman/MTW2_Swordsman_charge_jump_attack.cas 39682271 23989 255 255 15 0 0 0 0 0 99 20 1 data/animations/New_Crew/Crew_Stand_to_Wide_Push.CAS 39706260 35289 1 0 0 0 0 0 0 0 25 20 1 data/animations/New_Crew/Crew_Wide_Push.CAS 39741549 8945 1 0 0 0 0 0 0 0 39 20 1 data/animations/New_Crew/Crew_Wide_Push_to_Crew_Stand.CAS 39750494 13929 1 0 0 0 0 0 0 0 67 2 1 data/animations/New_Swordman/Weapon/Sword_Default.CAS 39764423 4601 1 0 0 0 0 0 0 0 59 19 1 data/animations/SM_D01 Diplomat unselected idle 01.cas 39769024 20105 1 0 0 0 0 0 0 0 51 19 1 data/animations/SM_D01 Diplomat unselected idle 02.cas 39789129 17385 1 0 0 0 0 0 0 0 45 19 1 data/animations/SM_D01 Diplomat unselected idle 03.cas 39806514 15345 1 0 0 0 0 0 0 0 21 19 1 data/animations/SM_D03 Diplomat selected idle.cas 39821859 7185 1 0 0 0 0 0 0 0 25 19 1 data/animations/SM_D06 Diplomat walk loop.cas 39829044 8545 1 0 0 0 0 0 0 0 35 19 1 data/animations/SM_D11 Diplomat standing to stepping backwards.cas 39837589 11945 1 0 0 0 0 0 0 0 35 19 1 data/animations/SM_D12 Diplomat stepping backwards loop.cas 39849534 11945 1 0 0 0 0 0 0 0 25 19 1 data/animations/SM_D13 Diplomat stepping backwards to stand.cas 39861479 8545 1 0 0 0 0 0 0 0 23 19 1 data/animations/STRAT LID_08_Stand 2 90 right.cas 39870024 7865 1 0 0 0 0 0 0 0 23 19 1 data/animations/STRAT LID_09_Stand 2 90 left.cas 39877889 7865 1 0 0 0 0 0 0 0 107 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_assassinate.cas 39885754 59537 255 255 7 0 0 0 0 0 25 19 1 data/animations/Stratmap_Assassin/Strat_Assassin_basepose.cas 39945291 8545 1 0 0 0 0 0 0 0 39 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_die_backward_1.cas 39953836 21729 255 255 7 0 0 0 0 0 83 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_die_forward_1.cas 39975565 46193 255 255 7 0 0 0 0 0 41 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_no_mp_idle.cas 40021758 22841 255 255 7 0 0 0 0 0 15 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_no_mp_to_stand_A.cas 40044599 8385 255 255 7 0 0 0 0 0 131 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_sabotage.cas 40052984 72881 255 255 7 0 0 0 0 0 33 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_selected.cas 40125865 18393 255 255 7 0 0 0 0 0 41 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_idle.cas 40144258 22841 255 255 7 0 0 0 0 0 11 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_idle_selected.cas 40167099 6161 255 255 7 0 0 0 0 0 111 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_lf_idle_1.cas 40173260 61761 255 255 7 0 0 0 0 0 81 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_lf_idle_2.cas 40235021 45081 255 255 7 0 0 0 0 0 85 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_lf_idle_3.cas 40280102 47305 255 255 7 0 0 0 0 0 29 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_to_no_mp.cas 40327407 16169 255 255 7 0 0 0 0 0 25 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_to_walk.cas 40343576 13945 255 255 7 0 0 0 0 0 59 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_to_walk_backward.cas 40357521 32849 255 255 7 0 0 0 0 0 33 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_turn_90_ccw.cas 40390370 18393 255 255 7 0 0 0 0 0 33 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_stand_A_turn_90_cw.cas 40408763 18393 255 255 7 0 0 0 0 0 35 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_unselected.cas 40427156 19505 255 255 7 0 0 0 0 0 33 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_walk.cas 40446661 18393 255 255 7 0 0 0 0 0 27 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_walk_backward.cas 40465054 15057 255 255 7 0 0 0 0 0 29 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_walk_backward_to_stand_A.cas 40480111 16169 255 255 7 0 0 0 0 0 25 19 19 data/animations/Stratmap_Assassin/Strat_Assassin_walk_to_stand_A.cas 40496280 13945 255 255 7 0 0 0 0 0 55 19 19 data/animations/Stratmap_Diplomat/Strat_Diplomat_backward_1.cas 40510225 30625 255 255 7 0 0 0 0 0 21 19 1 data/animations/Stratmap_Diplomat/Strat_Diplomat_basepose.cas 40540850 7185 1 0 0 0 0 0 0 0 105 19 19 data/animations/Stratmap_Diplomat/Strat_Diplomat_conduct_diplomacy.cas 40548035 58425 255 255 7 0 0 0 0 0 67 19 19 data/animations/Stratmap_Diplomat/Strat_Diplomat_forward_1.cas 40606460 37297 255 255 7 0 0 0 0 0 41 19 19 data/animations/Stratmap_Diplomat/Strat_Diplomat_no_mp_idle.cas 40643757 22841 255 255 7 0 0 0 0 0

The header and trailers are in decimal and I'm showing the three header
bytes as a short and a byte so 20 0 20 is shown as 20 20.

I modified two .cas files for MTW2_Spear and tried putting them in game
using the no_animdb=1 and no_animdb=true under [util] in my cfg file
under the assumption it was like the io_filefirst switch, i.e. only put in
modified files under /data/animations. No joy. Thinking it might be an
all or nothing I put the whole unpacked animations directory and
renamed pack.dat and pack.idx. This just crashed. That's all the experiments
so far.

Edit: The .evt event files referred to in descr_skeleton.txt aren't in pack.dat
so is it crashing because they're not there? Couldn't find them by hexing through
events.dat, that's related to other sound events.

Re: Animations

Erf, the .evt files caused a major problem in RTW also, and ultimately led to downfall of all attempts by CA helpers to allow us to keep animations unpacked. They always had to be packed back into the files, or else the game would crash. (Editing the animations and then stuffing them back into pack files was only way of modding them.) Maybe Caliban can help us there.

Re: Animations

Ahh, thanks for that info. Reading zxiang1983's thread has me thinking
re-packing is the only way to get this to work too. Kind of ugly if that's
the only way, though.:no:

Re: Animations

OK, documentation is out of the way. Back to the fun stuff.
Wrote an animmerge script to combine a stock ms3d mesh with
a cas file. Here's the walk animation in Milkshape:

https://img248.imageshack.us/img248/...hapeap1.th.jpg

EDIT: Forgot the significant part, the shields held up is done by changing the animations in Milkshape.

and also the celebration .cas in Milkshape:

https://img248.imageshack.us/img248/...rateit1.th.jpg

From here you use the animextract utility to extract to a .cas to put in
game as:

https://img441.imageshack.us/img441/...mentbz4.th.jpg

EDIT: These are the shields held up in-game.

NOTE This ain't right, everybody has become left-handed. Not that I mind, I'm
left-handed myself.

Unresolved Problems:

(1) Does no_animdb=1 work? PM into Caliban hasn't been answered.
Can't unpack skeleton.dat into the /animations directory because OS constraints don't allow a file without an extension to have the same
name as a directory. (Repeating xziang1983 here, been down this road.)

(2) Unpack in a higher directory? Maybe, haven't tried this.

(3) Unpack in a lower directory? Same answer, try later.

(4) Repacking the pack isn't that bad. Live with it.

(5) Maybe the TYPE files have an implied extension we just don't know
about. This would be nice but we need information.

(6) What the heck is the global data? I just saved it in the materials comment
section but need to know what it really is. Also the mystery 8 floats at
the end, also written into the materials comment section.

(7) Outside the bounds, but the TYPE files (using the language of
descr_skeleton.txt) have bone info in them, can these be modded
to change the sizes of units re Bwian's question in the mesh thread?

Re: Animations

Haha those sergeants are really screwed :laugh4:
The arms look weird... hmm, maybe the game saves some bone positions you didn't export correctly?

Re: Animations

I just don't know how to use Milkshape. The verts didn't
move with the joints when I edited the keyframes. Anyway,
just an experiment for repacking and putting modded cas's
back in game.

Got a reply from Caliban. He showed me a .jpg of his /animations
directory. Could not see from that if the type files were unpacked
there or not but the .evt files are definately there so they need to
be found and unpacked.

I asked for two experiments: try searching for navy_strat*.* and
fs_camel*.*. If these don't turn up any files then skeletons.dat
isn't meant to be unpacked and the question should be settled.

I've been looking at the data at the top of the type files
before it gets to the animation families and the standard skeleton rig is
in there. These could probably be modded to make custom
animations for re-rigged figures, i.e. like dwarves etc., if the engine
will accept more entries in skeletons.dat and skeletons.idx.

Re: Animations

Search in the skeleton.dat? Not sure if I follow you right here but if your looking for animations in there with this as part of the filename you wont find them, however the animations in the descr_skeletons file under the fs_camel and strat_navy all reference cas's are in the dat. Sorry if I'm not with you here, having trouble following this

About the evt files, looking at one of the ones that's not packed theres reference to anim_cannon_creak or similar, events.dat has the same text followed by paths for relevant sounds. Looking around a bit in there theres stuff like anim_archer_aim which in turn corresponds to text in the skeleton.dat at the archers animations. Possible they're packed in here?

Bytes	Variable type	Explanation
0-3	long	strlen+10 of the tag (below)
4-7	long	byte offset in pack.dat
8-11	long	byte length in pack.dat
12-15	float	Unclear so far, often 1.0 (first four additional bytes of the strlen+10)
16-25	-	These are the other 6 additional bytes in the strlen+10*
strlen	String	tag (qualified path for the animation)