emPolygonizer - metaballs for SOFTIMAGE

[Mootzoid]

Hi again!

It's finally done: emPolygonizer version 1.0 is released

Here's the link to the emPolygonizer web page (very draft, sorry):
http://www.mootzoid.com/html/XsiCorner/ ... nizer.html
The demo version, the documentation and some demo scenes can be downloaded from there.

A few words concerning this version 1.0:
I am aware of the fact that this plug-in is nothing really new and revolutionary, it's "main feature" till now is the simple fact that it is available for XSI ;)
It is frankly "basic", meaning that the metamesher algorithm I implemented is a rather common one. But the main labour was to implement the framework for future updates as well as a (hopefully) good workflow. In other words: there's more to come!
And I am looking forward to get some feedback and/or cool ideas from you guys!

enjoy!
Eric

PS: I will soon start a new emPolygonizer - thread in the forum "Softimage -> Plugins".

[janimatic]

excellent work as usual Eric,

as far as i can tell using the demo, the perf looks pretty good
It will be very usefull
thanks

Julien

[Hirazi Blue]

Haven't had time to test it yet, but congratulation seem in order nonetheless...

BTW: I changed the title of the thread (deleted the word "upcoming") & it might be an idea to move the whole topic to a different location...

[Mootzoid]

@ Hirazi Blue

okay, that is a good idea. Could you move this thread to the "Softimage - Plugins" forum?
Thanks!

[Hirazi Blue]

As you can see, I've moved it to the Plugins-forum and I appended some info to owei's first post...
(As one Moderator to another: it's not so hard!!!! )

[Matic]

I see that it's released, congrats Eric!

I'm going to be interested putting this thru some tests! My first interest is in caching and freezing results, second would be in how it can work in a production pipeline - can I get meaningful motion vectors for blurring etc. Any way you look at you've achieved something a lot of people have been needing and trying for, kudos!

[Mootzoid]

Hi,

just a short post to inform that I added another demo animation to the emPolygonizer web page,
nothing really that exciting (just some "blobby mess" again), but what the hell, it shows that the plug-in works ;)

watch "A very colorful blobby Mess"

cheers,
Eric

[Mootzoid]

I plan to implement a "file caching" option in the next emPolygonizer update that will allow to read/write the generated geometry to disc. There will be a proprietary file format, but I would also like to support a few more common file formats.
Does anybody know what kind of formats are used by other applications for this kind of stuff? I bet there are already some sort of "standards"...
Any information concerning that subject is appreciated !!

[Tekano]

Hi Mootzoid,

Well .PC2 would be the obvious choice being a format both ICE & KP_Pointcache uses. not sure if it supports meshes that change polycount per frame though.

[CiaranM]

just had a play with the demo, looks great! I'm sure I'll have some feature requests to come, but I like the level of control that's already available.

If you're familiar with pWrapper, there's a nice function called 'swrap' that produces concavities and pinching in the mesh, that can give a very exaggerated 'liquid' look. Is there a similar function in emPolygonizer? I can't quite match that look yet. Perhaps the falloff attribute on the pointcloud?

Is the mesher multi-threaded? Either way, it seems quite quick.

second would be in how it can work in a production pipeline - can I get meaningful motion vectors for blurring etc.

I'm sure Eric will have a solution for motion vectors ;) , but in the meantime, this workaround works pretty well for pWrapper and should work for emPolygonizer too.

[Mootzoid]

@ Tekano :

I just looked into the .PC2 format (= PointCache2 format). It indeed seems to be supported by quite some plug-ins available for XSI, Maya, 3D Max, Bender ... I'll continue researching in that direction, thanx!

@ CiaranM :

If you're familiar with pWrapper, there's a nice function called 'swrap' that produces concavities and pinching in the mesh, that can give a very exaggerated 'liquid' look. Is there a similar function in emPolygonizer?

The answer is: "No, not yet". Amongst other things this is one of the more important features that will be available in one of the next updates. I saw pWrapper for the first time about two weeks after I had started coding emPolygonizer. I was amazed by what you call the "exaggerated liquid look", really very well done. The only problem: give pWrapper a few thousand particles, let's say 10.000, and then wait... wait... wait... It is the classic "every particles influences every other particle" problem, and if you try to solve it the naive way (check every particle with all the others) you very quickly have exorbitant calculation times. Thankfully the good ol' "octree" can help solving this problem. I already implemented an octree in emNewton, so it hopefully won't take too long... I am working on it ;)

[Letterbox]

@ Tekano :

I just looked into the .PC2 format (= PointCache2 format). It indeed seems to be supported by quite some plug-ins available for XSI, Maya, 3D Max, Bender ... I'll continue researching in that direction, thanx!

@ CiaranM :

If you're familiar with pWrapper, there's a nice function called 'swrap' that produces concavities and pinching in the mesh, that can give a very exaggerated 'liquid' look. Is there a similar function in emPolygonizer?

The answer is: "No, not yet". Amongst other things this is one of the more important features that will be available in one of the next updates. I saw pWrapper for the first time about two weeks after I had started coding emPolygonizer. I was amazed by what you call the "exaggerated liquid look", really very well done. The only problem: give pWrapper a few thousand particles, let's say 10.000, and then wait... wait... wait... It is the classic "every particles influences every other particle" problem, and if you try to solve it the naive way (check every particle with all the others) you very quickly have exorbitant calculation times. Thankfully the good ol' "octree" can help solving this problem. I already implemented an octree in emNewton, so it hopefully won't take too long... I am working on it ;)

You could also look at MDD format, you might be able to tap into the one thats already there?

I think you might want to take a look again at the free metaballs, cant remember the mans name, but as far as i remember he completely tessellates the object upfront, that might help this and other issues.

Cheers
Terry

[janimatic]

hey!


i don't think pc2 and mdd are well suited , because of the dynamic topology.
The formats supporting this already :
* obj sequence but it won't let you store custom data (such as motion vectors) but a custom op reading cache from disk would be better than nothing...
* houdini bgeo but it's houdini specific
* realflow bin mesh format maybe as there are some plugins already available for xsi 6, max, maya, houdini, etc...

regards !
Julien

[Mootzoid]

Thanks for all the advice, guys
I looked into all the formats and here is what will happen: PC2 and MDD are nice formats, but only for points => those formats aren't suited to cache the geometry (points and polygons).
So for a start I will support an own format (that can also cache the vertex colors) and the well-known and widely supported Wavefront OBJ format. That should be a good start, I think

[Letterbox]

Thanks for all the advice, guys
I looked into all the formats and here is what will happen: PC2 and MDD are nice formats, but only for points => those formats aren't suited to cache the geometry (points and polygons).
So for a start I will support an own format (that can also cache the vertex colors) and the well-known and widely supported Wavefront OBJ format. That should be a good start, I think

What about Collada?

[Mootzoid]

Ah, yes, I hadn't thought about Collada. Definitively a format I should also implement.
Just downloaded the PDF of the Collada file format specifications... it is very detailed:

it has 298 pages!

Good thing I only have points, polygons and vertex colors
=> I will be able to skip 98% of the chapters ;)