Animations

[KnightErrant]

I think its progress... mixed with blowing off steam at finding a lot of
my mistakes. I think there's rhyme to the reason with the
differing cas formats, I'm just missing the RTW experience with the .cas
file format. The exporting skeletons bit is ok I just want to move it to the
new library functions which is why I'm stuck on getting animmerge right,
it seems to be best way to check if the scaling is ok or not. Looking
at animations in game is too hard, no grids to look at to see if the
y-offsets are right or not.

I'm hoping for rain tomorrow, can't be sent out of the house to do yard
stuff. Seriously, thought I'd be done this weekend, but looks like
a couple more days. This is what looks like it would be useful:

animmerge - take a .cas animation and merge with a unit to look at animations
in Milkshape for final tweaking, either in Milkshape for a single
animation or to change scaling values to do a whole family of
animations. This is almost right, I'm just working out a rational
naming scheme: seems the best way to keep track of what
one is doing is to combine the base unit .ms3d name with the
animation .cas file name a la

armored_sergeants_lod0_animby_MTW2_Spear_walk.ms3d

This way the user can know just by looking at the new file name
what he was doing. Also might be useful for animextract to
know what to name the extracted animation, maybe with
"_modified" so it doesn't overwrite something.

animextract- less important, but needed if people want to tweak a few really
important common animations like walk, run, run_to_charge, etc.
Pulls the new animation frames out and makes a new .cas file.

exportskeleton - final name for writing non-standard skeleton to all the .cas
files in a directory and scaling the animation floats. This works
already, just want to use the new library functions.

extractskeleton - little utility for pulling a skeleton out of a .ms3d file after
a modeler has rigged a new skeleton. Already done with the new
library functions, sort of comes for free with the general
.cas reader.

castotxtconverter - Don't know if this is needed but its been helpful to
survey animations. Reads a cas file and converts all the fields
to ASCII and writes out the equivalent file in ASCII. This is done
already.

txttocasconverter - Probably not necessary. Users will probably only modify
animations in Milkshape and not change rotations or animation
translations by hand. If this seemed useful to someone it
could be done but I'll rank it last for now.

[GrumpyOldMan]

Hi KE et al

It sounds like you're doing really well, don't worry about the completion date - when I used to look at IT specs and estimates, I always multiplied the estimated time by 1.5 and the estimated money by 2.5 , it was so accurate it was scary .

Just to add some more angst to your deliberations, I've been looking at the rider figures and they have some variations that don't play ball with your current animmerge either. They seem to have been exported using two different variations on the export screen that Caliban posted a picture of. Some have been saved with full rotation and translation values but others have been saved without any translation frames apart from the pelvis (which is all zeroes anyway). Amongst both these variations some have frames for clavicles and some don't. Also some have two extra bytes in the footer because they have 'CustomAttribNode 1' rather than 'CustomAttribNode'. I shudder to think what we'll find when we look carefully at the camels, I saw one that had 4 'CustomAttribNode' entries.

On an up note, Milkshape will take anims that have different numbers of frames across bones, so if you export an anim with 61 frames for everything except the clavicles which have one 0 frame each, this is still a valid anim. The only point to watch is that you have the same number of keyframes for rotation and translation for each individual bone - ie if you have 61 frames of rotations then you need 61 frames of translations (even if they are all 0 entries).

I've been looking at what has to be done to update the mesh converter so it will work in with new skeletons. I'll have to go to a template system. The converter, currently, is using a vanilla mtw2 human template. I'll have to write a process so that the converter can extract a template from a ms3d file and write the bits and ms3d parts that are required. This way people will have to select which template to use when converting a mesh to ms3d or vice versa. So if you wanted to extract the Hospitaller knight with Bwians Dwarf skeleton and rework the mesh to fit the skeleton it could be done. We'll have to look at adding bones and where they should go ie before or after the weapon/shield bones in the mesh. I did an experiment and extracted these bones from meshes but it crashes the engine. I suspect that any new bones will have to go after these bones because of bone index numbers (I hope not) but this will make writing interactions between mesh and anim processes a bit more difficult.

Cheers

GrumpyOldMan

[KnightErrant]

Yikes, this is getting complicated... but if looked at correctly that means
fun. Good advice, one thing at a time. Leaving camels and others aside,
at least some of the stratmap anims look normal, meaning like the battlemap
ones. I've tried to make the library cas reader read every field since there's
no knowing what field actually means anything until all the variations are seen.
I do remember you saying the adventure would continue, prescience can be
scary.

I never get anything done on weekends anymore. Just another push this week
and maybe things come together. I think you're right about the mesh converter;
only a template system seems to make sense to deal with the variations already
in game and those that people with mods would want to work with. I think
the system for this can't be done with little utils here and there, someone is
going to have to pull it together to make a standard interface for meshes and
anims since they are so closely tied together. (I nominate you of course and
volunteer for research assistant.) I have read the e-mails re
chariots and testudo anims and did not realize just how rich this could be.
Really got me thinking now, animations could be so much more interesting
than I thought when I started.

[SigniferOne]

Man you guys are amazing. GOM, just don't put a bucket on your head and start saying "I am your father" You two rock.

[Bwian]

Mk 1 Guinea Pig takes the plunge!

Step 1 .... Animation folder copied and renamed ...check. Didn' rename any files in the folder...just the folder itself.

Step 2 .... Renamed the two 'bad' files that are not in the descr_skeleton

Step 3 .... bottled this one, and used the supplied skeleton! Will try to extract my own when I have this working with yours.

Step 4 .... hit the export skeleton button ( using your skeleton ) and I get exactly what you said I would. A sub directory 'convertedfiles' with a copy of all the animations in it. No evt files. feeling comfy so far.

Step 5 .... run the anim-merge and got an MS3D file OK. Ran it in MS3D onthe keyframer, and the model animated correctly. I noticed that the 'floor' was not where the guy walked... but I wasn't overly worried at this stage. What I could clearly see, though, was thet the bones were in the correct place for my model, and not the stock one. Also, the movement was correct and the mesh was not deforming in any unexpected way. So..I press on!

Step 6 .... skipped the convert to txt effort ....wasn't sure I would know what I was looking at, so I pressed on. I will come back to these steps if I have a failure, or feel I need to understand a bit more about this!

Step 7 .... Converted Dwarf ( used the one I had previousely zeroed...seemed to work fine. On comfortable ground here, done this bit many times!

Step 8 .... Slotted the mesh into my Warhammer mod directory ( made sure, while I was at it that I had removed my current animation files so it would generate a fresh one ) and updated the modelsdb and unit files. Again...on safe ground here!

Step 9 .... pasted in the chunk of animation from your file, rather than just use yours. Getting brave now!

Step 10 .... RUN THE GAME! Ahhh...it fails. Despair. Then....shame... I had forgotten to put the textures in the mod folder. CA...I curse your 'unspecified error' .

Step 11 .... run the game again with the files in the right place! All goes well, and a custom battle is created with Dwarf v Skeleton. Loads fine ....

But there is a snag. The Dwarfs have learnt to levitate. When I loaded the animerge product in Milkshape, I noted that the model was not sitting ON the baseline lile I expected it to. It looked like it was higher....and that is exactly how it has come out in game.

On the plus side, every step worked smoothly, and exactly as predicted. Nothing crashed, failed or screwed up. No nasty errors. The only thing that has not worked correctly is the vertical position. This has not altered.

Did I miss something?

Superb stuff otherwise, and easier to use than I had expected!

[Bwian]

to help out a bit ... here's a picture of the thing in animerge:

[KnightErrant]

Excellent news! No, I don't think you missed anything. I may not
have set the exact scaling values I used to get them on the y=0
plane. Open up animationutilities.py in your favorite editor and search
for SCALEY. This should be at the top of the function exportskeleton.
I thought 0.64 put them on the plane, but try making this smaller,
save, then do the exportskeleton step again. Rather than copy the
files and run the game, just do an animmerge with like the walk animation
in convertedfiles and see how it looks. Might have to iterate just a
couple of times to nail it, but it shouldn't take too long.

When it looks good in Milkshape, then copy the cas files to your
dwarf animations directory, and, (silent prayer) they should look allright
in game.

Edit: Leaving from work, I'll check back when I get home.

[Bwian]

SCALEY was indeed set to 0.64. Tried changing it to 0.14 .... which should have a noticable difference. Ran the steps through, but the animerge file still showed the animation in EXACTLY the same place as before. Too high. I was expecting it to be wrong, but not the same. I will give it another shot with some different values in there tomorrow. Right now, it's 23:30 UK time, and I need to get some shut-eye! It's a 6am wakeup on work days.

Soon as I get in tomorrow, I'll do some more testing to see what I can change with different values in there. I'll post back to let you know how I get on.

I am running in 1.2 ( if that makes any difference, I don't know ) so this is valuable test data whether I succeed or fail here. Main thing is, the process steps all did pretty much what theywere expected to do. Just seems hung up on the y axis positioning.

Biggest worry for me was that the model seemed to end up stuck to the origin. I had exactly the same issue when I tried to scale animations......hope it's just coincidence. Will know more tomorrow.