ICE Topo - Adaptative Subdivision

[origin]

Hello guys,
my first simple ICE topo compound for adaptive subdivisions
let me know if it works for you

[Maximus]

thanks Origin, that works perfect! nice compound !
thanks for sharing!

[Memag]

Thanks.
Works here as well (didn't work on grid object though).

[guillaume]

Very nice one . I like how you handle the adaptivity at each iteration.

I took a quick look at your graph and made a 1.1 version. I removed the Build Array from Set, and some Find in Array nodes. Instead I'm using a custom attribute to decide if the polygon must be subdivided or not. This way it is 1000% faster ;).

Here is the compound : http://dl.dropbox.com/u/5533643/Softima ... sicompound

Cheers,

Guillaume Laforge

[origin]

Thanks a ton!
I suppose I should avoid repeat node at all cost (no multithreading?) ?

[guillaume]

Thanks a ton!
I suppose I should avoid repeat node at all cost (no multithreading?) ?

Your welcome .

About the Repeat node, it depends. For example in your compound, the main repeat node will not repeat too many time (the user will keep it at one or two times) and so its not a big problem. But in the "Fill Poly IDs Array to subdivide" the repeat time is dependent of the point count. So on big mesh it will be very expensive. Sometime it is not possible to avoid it, but if possible, avoid it ;). The repeat node is an execution type node. That means that it will output some ICE evaluations to Softimage (using the Set Data node). Each time a set data is executed, the attribute value stored in the polygon mesh (or point cloud or null) will be updated. This mechanism is optimized but can't be as fast as just doing some pure multi-treaded ICE evaluations between two nodes.

To answer your question about the multi-threading of the repeat node, it is half yes and half no . The repeat node will just ... repeat the same execution branch several times. Of course it can't distribute each repetition in its own thread as next repetition is dependent of the previous one and so can't be done in parallel, but inside each repetition, multi-threading will be handle if possible.

Guillaume

[gfxman]

That's great, you're here "helping" people with their compound!
I've read that you're the main "ice modelling developer" is that right?

I have a couple of question and suggestion, i hope you don't mind ( really big OT here, mods!)

The whole ice / compound idea,as it was in the first iteration was very community - user generated tools - related; and it could have it worked a lot to let the user develop for other people, and make xsi grow by itself.
The " upload to sofimage " idea was simply genius, and i can't really understand why, they rip it away....
All the compounds, soon get lost in the cloud, and that's a fact.
A central site, to upload / download, compounds and revision, would be so great, for people like me, that can't really "code in ice" and would multiply the value of xsi as multipurpose tool.

The second thing a would like to ask is why, there is so little support to the new ice nodes.
Just 5 scenes.
Just the videos you made ( they're not official by Softimage i suppose ) no real tutorial.
So little default compounds given to the user to understand what's going on.

That's the big mistake you're making: too much potential but so scarce support.

I so easy for some of you, to code new tools in ice, why dont' you just integrate them in the software? Even without a proper documentation, they would be better then nothing; and olso these user compound, would be very usefull, as a full integrated tool by default, or not?

Second thing:

The ice idea, turned softimage from an animation/modelling station, to a special effect / phisic simulator; no real love was put in the modelling tools, animation mixer ( are them perfect to you ? ) or fx tree in the last 3 years.

But it would be so easy for you and 2 or 3 other people like heldge, to boost the modelling set in less then a year.
I hoped even to see a lot of user modelling plugin ( a lot of witch are not working any more..) converted and integrated in xsi: i think you can make a better taper or thinkness tool in a second, with your skill.. maybe even a cap tool.... witch i'm waiting as a basic tool in xsi for 10 years now....

Third thing:
Xsibase is dying. The people who used to run it, they're lost interest, as life goes on. si-community is an answer to that, but the community is too litlle to be so splitted. We need a modo-like, support site.

The forum
The tutorial
The central compound uploda/download
The info about the tecnology is going to come in the future

It cant' be don in the area, you know that.
Regars,

gfxman


Moderator edit: As this truly is a really big OT,
I will try to follow this & move your post and the following ones
to their own thread if the situation warrants this... - HB
;)

[guillaume]

That's great, you're here "helping" people with their compound!
I've read that you're the main "ice modelling developer" is that right?

I'm one of the main dev yes.

The second thing a would like to ask is why, there is so little support to the new ice nodes.
Just 5 scenes.
Just the videos you made ( they're not official by Softimage i suppose ) no real tutorial.
So little default compounds given to the user to understand what's going on.

That's the big mistake you're making: too much potential but so scarce support.

Will try to do better next time.

I so easy for some of you, to code new tools in ice, why dont' you just integrate them in the software? Even without a proper documentation, they would be better then nothing; and olso these user compound, would be very usefull, as a full integrated tool by default, or not?

Because it is not always so easy.

Now my turn to ask some questions.

Did you take a look at the few sample scenes and compounds. Do you understand how they are working ? If yes, that's great, but if not feel free to ask question on public forums. I'm sure some people will help you.

Before being a dev at Autodesk, I was a user trying to help others (on my spare time) on various forums. I'm keeping the same behaviour, but if every time I say something I've got this kind of answer, it will be hard for me to continue this way. I hope you understand it.

Sincerly,

Guillaume Laforge

[gfxman]

Before being a dev at Autodesk, I was a user trying to help others (on my spare time) on various forums. I'm keeping the same behaviour, but if every time I say something I've got this kind of answer, it will be hard for me to continue this way. I hope you understand it.

Sincerly,

Guillaume Laforge

I'm very sorry, maybe i sounded harsh, but i didn't want to be, at all.
Really.

The softimage devolping team is like a mythological thing, and it's such an event i can get in touch to them, directly, that i took my chace to tell you what i thought would be a great help to softimage, as a product, not only to me, like a user.

Sorry if i bodered you. I hoped for a different answer but i suppose that the only one you can / want to give me.
My apologies, again.

p.s.
there's noway we're getting back the centralized compound upload/download thing he ?
Ok ok, i'm leaving!

[Hirazi Blue]

I don't think there is any need for apologies...
To (hopefully) conclude this OT side-track of this thread:
as long as Autodesk sincerely (and to state the obvious: wrongly IMHO) believes
the download section at the Area is sufficient,
nothing much will change in that respect, I'm afraid...
;)

[guillaume]

No problems gfxman !

@ Hirazi Blue : Maybe you could conclude this OT by creating a specific thread for a centralized and good place for ICE compounds (I 100% agree to this request) ?

Cheers

Guillaume

[ActionArt]

Thanks Guillaume, your insight and help on compounds is greatly appreciated.

BTW, I have to say ICE is incredibly complex and well made! Great work to say the least!!!!!

And origin, thanks for starting this, it's a VERY useful tool!

[Tekano]

yes - this place is as good as any, viewforum.php?f=19 its just not uptodate or has *all* available submitted compounds but its still online right now, is usable and can accept user attachments which is better than xsibase or 'the area'. I recon a cross pollination from rray.de and a dig out of old ICE compounds backup folder then we may be getting it into a better shape than it currently is. I too miss the old softimage website for centralised ICE compounds, it had a lot of initial momentum that AD simply, for want of a better word, *destroyed* but yes it is very apparent that AD are not interested in or have the resourses to deliver what was already in place, so we will have to make do with our own home grown efforts. next week Ive some free time so I promise to raid my backups folder and try and beef up the ICE downloads section with already downloaded ICE compounds

[gustavoeb]

Hi Piotrek,

Due to current project needs I´ve downloaded your compound and modified it to support subdivision based on distance to camera and geometry. For camera I have nothing fancy like considering the frustum, rasterized polygon size or anything like that. But its good enough for many cases.

I´m sharing the compound here, if you dont mind... If you do, just warn me and I shall remove it:
https://dl.dropbox.com/u/39280733/ooAda ... sicompound

[AlexanderM]

It would be good to add adaptive poly reduction mode, in addition. Or as a separate tool

[gustavoeb]

it would be interesting... probably would only be useful for someone using data from scanned 3d models I guess