Call to Jeff Gomez

I just read the article on the New World Notes about your god-like importation of .bvh files. Any chances you would post the thing here so we can all play with it?


http://secondlife.blogs.com/nwn/2005/03/the_shape_of_pr.html

I've been found out!

Err... yes. Sometime later today? It's pretty much done - I just need to comment everything so people will understand what to do. It works pretty well though - that much I'll say.

This is what I get for bringing Hamlet in a day early.

Edit: And the "minor bug" referred to in the article has been squashed. As it turns out, I uncovered a bit of an LSL bug with that - and removing a small "on_rez" call fixed it.

I'de dance to celebrate, but i don`t have my dancing bracelet at the moment.

Very impressive, Jeffrey. I'd long said that it would be too difficult to implement, but I was evidently very wrong. I anticipate some very nice mecha-centric applications for this idea.

Just a question, though: how does your script handle scaling? Can I apply the same animation to a 20 m tall av as to a 2 meter tall av without any major corrections or changes to the ani? Do the joints move as expected? Are you an incarnation of God?

Thanks.

Jeffrey, congratulations on your New World Notes article! I would like to see this inworld, as it sounds cool but firsthand is quite a different experience, of course.

I hope to see... giant tentacles... space arms... Asimo-like robots with more of a Shaft attitude...

This could -- and should -- open up a whole realm of new possibilities for primotion in SL.

As an aside, though, joints are going to have to be "light" in terms of mass due to weight limits of llMoveToTarget. NPCs and stuffs can be done, but they're going to need to be reasonibly sized or hollowed out - and max, 31 prims per joint. Just an FYI.

So, while Hamlet gave me (way too much) credit - there are a few limitations to be mindful of. I'll try to detail them in the script itself.

EEEEEEEEEHHHH OMG OMG OMG OMG OMG OMG OMG OMG <breathes> Wow!

1) 20m tall avatars? Maybe. Respect the limits of llMoveToTarget
2) Joints move fairly as expected, but I may need some more testing on other animations, to be safe.
3) No, I'm not God in scripter form.

Edit: What am I thinking? There are mass commands for stuff like this...

Will this be the same setup as your .obj importer, i.e. you have to paste in the contents of the bvh file? If so, with the 100 item limit in the list, will this cramp how many frames, etc can be used at once?

Yes, same format. Yes, because of the potential to abuse.

As for cramping how many frames can be used at once, realistically this method can only take so many - roughly 50-60 frames or so with four moving joints, or 30ish with 8-12. This is due to the cells being greater than the script buffer and, again, I personally feel that if you're piping in 50 frames for an animation (as I have with a tested "ballerina" downloaded animation), that should be enough. If you want more, you can write your own multi-script interpretter.

Well, I'm going to get no sleep.

For those of you that wish to jump the gun, here's the code! I'll be posting it to the script library as fast as the monitoring Lindens will let me.

NOTE: The "Heirarchial Data Code" and "Raw Data Reader" MUST use be named thusly to work, unless you edit the code itself. Bear this in mind. The recomended name for each script is included at the top of the text itself. Usage instructions included in the Heirarchial Data Code.

Happy importing!

PS: The example I showed Hamlet is included in this code.

I gave the setup a shot this morning, and though it worked well for the bvh file which you supplied, I had a hellish time getting my own to work. After various memory errors, I finally produced a simple 10 fram walking ani. However, after I prepared it and applied it to the scripts, the ani motion was lost. Instead, the script caused the joints to oscillate between 0 offset (moving into the same position as the root prim) and the pose in frame 1 of the ani. I'm probably importing something wrong, but I don't have the time to play with it any longer this morning.

I find that doing a search and replace is infinitely easier than manually preparing the BVH files. In Microsoft Word, you can use the following search and replace strings to prepare the files:
search: ^p
replace: ",^p"

followed by a couple of runs of
search: "^t
replace: ^t"

Then manually clean up the few stray commas, or missing quotes.

Right now, each joint is generated on the fly. Though I think I'll be able to edit the script so that it rezzes premade objects, I won't have a chance to find out until later.

Yeah - any holes in documentation, code, etc, would be nice to refer to me - since I am kinda listing the script as "open beta."

I actually use llParseString2List for nuking stray commas, spaces, and extraneous code from the segments. I spent a fair deal of time yesterday doing that for precisely this reason. Instead of "search and replace," it simply gets rid of all the elements that we aren't interested in... so if you're still having a problem, I would be interested in why. One possibility is Case-Sensitivity.

So far, I've only had a chance to test with Blender-exported BVH files (and looked into Poser ones), so if you're using another program, I would be very interested in whether the syntax is different. I did my best to accomodate for varied syntax with what I knew (for example, X Y and Z data may be declared Z-Y-X, X-Z-Y, Y-Z-X, etc).

Otherwise, I have had the chance to test with a second import (a ballerina from a site I found) which I actually brought in in about three pieces (similar to my model importing method). It ran fairly well. If you find yourself rapidly running out of script memory, this is the way to go - and I'd be glad to show you how to "break up" existing BVH motion files in this manner. I was able to get in roughly 50 frames from it, and it ran fairly well.

Anyway, IM me in-world if you have any further problems with it.

Edit: Oh, and pre-made objects can just be rezzed from inventory and they'll still run fine. That's the beauty of LSL.

This is wonderful and magical. I just tried it out and am now carrying a piece fo <i>magic</i> in my pockets.

Very cool.

Glad to know you liked it.

What's the proper syntax for splitting a .bvh across the two lists in the hierarchy script? a full-body anim of 10-12 frames fits into the number cruncher script, but one obviously can't fit the whole hierarchy bit into the first list of the appropriate script. When I try to split the hier into the second list, it compiles, but the anim borks badly. I can't wait to get this properly working...

I split them up directly with Blender. Just import 'em to Blender and start deleting, save, repeat.

Then the comma-list-copy-paste syntax should work from there.

Ok, either I am very blind or I just don't know where to look. Where can I find all of Jeffrey Gomez's codes to help with importing prims into SL? Did I miss a link somewhere?

I am similary confused. I saw the "code is here!" part, and there's supposed to be 5 attachments but I don't see them nor any way to get at them...

Just to let everyone know that the link to the zip file is dead. Can this be fixed?

Yep. Nada is going to fix it shortly.

Jeff

This seems like an amazing piece of work from what I have read.

will it need any scripting knowledge or can any of us numbskulls work out how to use it without months of study?

I looked at the offline builder but was hopelessly lost within minutes.lol

Keep up the good work though. Its great to see someone doing something without profit in mind.

I can't get this to work. I must be missing something fundamental.

I did as the manual in "Heirarchical Data Code" said:
Created a prim and named it the_root_prim, put all three scripts in it.
Created a new prim, put the other two scripts in it ("Heirarchical Data Code" speaks of "List Interpreter" and "Rez Code", but there are no scripts with these names so i assumed them to be "Raw Data Reader" and "Base Code", since their meaning is somewhat similar).
Put the new prim inside the_root_prim.

Then I have to explicitly recompile the scripts in the prim or nothing would happen.
Then I can touch it, and i get error messages telling me that object "Joint" is not found.

So I assume that the new prim that I put into the_root_prim should be named "Joint".
I couldn't find any notice of this in the manual.

So I do that and recompile and touch again.

Many instances of the Joint-prim are rezzed! One for every BVH node i suppose.
But then, nothing happens, no matter how long I wait.

There must be more to it... can you please help me?

Thanks,
Stukas

Can you send me the files? I'd like to look at them myself.

Do anybody have a sample of something using this in-world that I can see?
Would be nice if a fully functional copy with mod rights was made available also.