Plain Houdini tricks in Softimage style, Volume 00001

Forum for users who have migrated or are migrating to Houdini
User avatar
Daniel Brassard
Posts: 878
Joined: 18 Mar 2010, 23:38
Location: St. Thomas, Ontario
Contact:

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by Daniel Brassard » 20 Feb 2015, 22:02

We are Groot! :D
$ifndef "Softimage"
set "Softimage" "true"
$endif

User avatar
FXDude
Posts: 1129
Joined: 19 Jun 2012, 21:59

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by FXDude » 21 Feb 2015, 15:17

I too find freezing to be pretty important. (among other things like general every-day straight forwardness)

nodeway

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by nodeway » 21 Feb 2015, 17:12

Guys, maybe I'm not getting something right now, but isn't Clean Transforms in Houdini the same what Freeze Transformation in Softimage?

EDIT: You are talking about http://softimage.wiki.softimage.com/xsi ... m#Rbu29906? So you mean cleaning history of operation? If yes, than this is another part of Houdini you don't get. We don't delete history because of the same reasons we don't have script log. Houdini handles things different. But there are good news.

1. You can lock nodes, so the operators before locked node are not executed.
2. You can unload nodes.
3. You put RopSOP and save geometry to disk and use FileSOP to load the same geo by referencing path from RopSOP.

Is this is what you wanted?

User avatar
Mathaeus
Posts: 1778
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia
Contact:

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by Mathaeus » 21 Feb 2015, 18:57

mantragora wrote:Guys, maybe I'm not getting something right now, but isn't Clean Transforms in Houdini the same what Freeze Transformation in Softimage?
Pre - transform operations in Houdini, including Clean Transform are closer to neutral pose in SI. Freeze Transforms in SI is usable more as an modeling operation - equivalent with inverse effect in H, could be Transform SOP.

Anyway, my main concern goes to constraints, for example 'look at' or 'path'. In SI, Maya and Blender (that's what I'm sure), constraint acts as an override of actual SRT, but, you're still able to transfer the constrined transformation, down to function curves of local SRT. In SI or Maya , you just set key, in Blender, there's 'apply visual transform' command. Because constraint is override, you can blend the weight against key-framed SRT, too. This ability alone, imho, is enough to describe complex movements of different centers of rotation, you're using an external object as an 'guide', still having visual control of everything, using only a few more function curves , for weighting.
In Houdini, constraint acts as an sort of parent, I don't see ability to weight (as there is no counterpart). Object moves, but local SRT is still on zero. "dynamic parenting' or 'blend SOP" also acts as an parent, does not affect the 'local' SRT.
In short, while in SI, Maya and Blender, I'm able to describe (relative) complex transforms in 'half-destructive' way (half= external objects are still there, no big deal to repeat the procedure) - in Houdini, controlling of same effect, suddenly spreads around, on many parameters.

I'd said this only as an example. Unfortunately there are other features in H, CHOP for example, very different than SI or Maya practice. Even if they are better, honestly I won't use them, simply because I'm not familiar with them. Maya as an industry standard don't have them, so why I should care about.
On fip side, I like VOPs and SOP, because they are similar to Softimage ICE.

That's it - IF Houdini wants to jump into animation, it has to have almost 1:1 of everything in Maya - OR Maya mirror like SI (it's not), OR something wonderfully better ( I don't see it) - OR, nicely organized simplification like Blender - i don't see that one, neither.
To express this differently, allowed ratios are equal, negate, multiply or remainder, but not completely different class.

Almost the same story as with Modo when it comes to animation, while Modo has a fairly better nodal control of animation than Houdini CHOPs - Modo outputs directly to matrices, not to channels like H, Modo nodes are designed in modern way, with all connectable inputs or outputs.
Modo at least tried to have something unseen in out-of-the-box category, like OOO thing. Houdini... I see only theoretical possibilities, nothing more....
By the way, how old are Houdini CHOPs ? Ten years or more, I think.

nodeway

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by nodeway » 21 Feb 2015, 20:48

Mathaeus wrote:By the way, how old are Houdini CHOPs ? Ten years or more, I think.
For animator Maya will be a lot better place than Houdini, that's for sure. You want to animate and don't bother yourself with other bells and whistles that application can give you. I would look into Houdini only if you need something more than just simple animation, something really phucked up. There is a reason why it's hardcore TD tool.

Even with Halfdan on board, I don't think Houdini will end-up anywhere close to just simple animation package you would like anytime soon, if ever.

Houdini Engine is the way to go for them right now, so they will rather focus on making it possible to use it in real-time too, without baking stuff, than on making happy Maya/SI animators. VFX is a dead end for them, they have to expand.

Yeah, CHOPS are old. They are on TODO list, if my memory doesn't fail me.

User avatar
FXDude
Posts: 1129
Joined: 19 Jun 2012, 21:59

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by FXDude » 24 Feb 2015, 02:19

Hi Mantagora, just to say thanks for your input.

it goes without saying that Houdini is also definitely unique, (not unlike XSI can be pretty unique) and indeed very different.

There are things you can only do in Houdini, and it's non-destrucive proceduralism compared to soft is perhaps what nuke is to after effects.

But what I meant in terms of Houdini was as a general purpose XSI replacement conditate at this point in time.


While procedualism in XSI is of course one if it's main point with either ICE or 'just' with basic construction stack operators for when modeling, when animating or "deforming", unlike Houdini it's not it's main focus.

In it's most basic usage, the construction stack acts as like local construction history or undo stack (in itself being a pretty big thing) that you otherwise only know it's there when you have to remind yourself to clear it once in a while, or when yo see how godsend it is when needing to go dig in some object's stack.

But sometimes everything in the construction stack is like maintained and "controlled".

Many times did I personally have things that depended on other things, that themselves depended on other things, in the end making-up quite a chain, and by changing/animating a few things at the source would change everything in the final result (with or without ICE)

But while that is very much a main point to XSI when making pretty much anything or all sorts of dynamic setups,
the very large majority of scenes consists of most elements ending-up having very little if not nothing in their stacks, except for really just a few things (in focus) that are kept dynamic.




Otherwise you were suggesting Maya, but while Maya definitely also has it's place, particularly for larger facilities that typically would want to have their own software (and humanizations) built on top of it...

... if Houdini, from an XSI point of view, may perhaps for now be lacking in everyday straight-forwardness,
(be it to a certain extent having to build your own tools out of low level ones, or the dependency on expressions)

Apart that Maya generaly fairs as one of the "the worse" in terms of "intuitivity"

In Maya, say for controlling particles (for instance... not unlike many or most things in Maya),
does involve fair amounts of scripting to have things behave in particular or specific ways.


But you know, when setups more or less heavily depends on scripts...
(when overviewing, editing your own stuff the next day (or the next year), or inheriting work of others....)
while you can find scripts or help on the net,
fair amounts of deciphering needs to be done to figure out how various setups work, from setup to setup.

And it goes without saying that you have to be, or become the type that doesn't mind debugging/dealing with syntax every day to make /edit your (more elaborate) things.

While sometimes even in more visual applications like Soft(ICE) or Houdini, it can already sometimes become pretty hard to follow the flow.


But I think a good way to describe XSI...
(if you pardon the blatant reductions to off-the-top-of-head percentages)


imagine something with 70% (or 90% with ICE) the proceduralism/flexibility of Houdini
(which is super important part of what makes XSI, XSI)

85% the customizability, and 90% to 150% the (immediate feedback) performance, (depending on what)
and a hard to ignore ~200% "production performance" of Maya (depending on what but for many things)
(performance also being a super important part of XSI)

and with just about as much upfront straight-forward simplicity and **stability** as c4D
for both new users and every day for the most technical "power users"
(intuitive workflow & easy to manage complexity, most definitely also a super important part of XSI)

And for each of those aspects, it's the fact that they are all together that can really top everything off.


Because from the point of view of a typical XSI user,
it could be said that what Houdini could be missing is both Maya performance & C4d simplicity

Maya can be missing both Houdini flexibiluity & C4d simplicity

and what C4d or Modo can be missing is the very customizable industrial grade nature of Maya and the incredible flexibility of Houdini.

Like all doing what they do very well, but also while sorely missing ~2/3 of what makes XSI, XSI,
having a fair part of those things but at the same time (it's all-in-one-ness)


So like Houdini, XSI also pretty much has it's own thing, and what could replace Houdini?. Maya? C4d?
Just not the same thing. Right?

And XSI similarly could'nt replace Maya or Houdini where they are best suited.

But in any event, perhaps and surely many things would change at some point in the future,
but I personally remain convinced we don't have to wait for "a new XSI",
compensating with what can make-up for parts of it at a time,
which -can- be sometimes better for their respective parts, BUT sorely missing other equally important parts to say the least.

No waiting because there is an XSI right now, and not only does it work, but I mean... Boy! can it really work (!) which is the entire point I believe.

nodeway

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by nodeway » 24 Feb 2015, 04:48

:)

Yeah, but it seems that Matheus wants animation package only. Somehow he thinks that writing ton of MEL/Python will be easier for him.

I myself will never touch Maya ever again. Last time I tried it, I quit 3D for 3 years. Even if everybody on earth will use it I will be the only person that will not.
FXDude wrote: imagine something with 70% (or 90% with ICE) the proceduralism/flexibility of Houdini
(which is super important part of what makes XSI, XSI)
ICE in nowhere close to 90% of Houdini flexibility. ICE may equal only to SOP part of Houdini, and then there is DOP land which is as big as SOP land. And of course there is COPS/CHOP. Each of them can be mixed with other.

BTW. Have you guys tried rigging in H14? Do you feel that new viewport is faster than old one with heavier rigs?

User avatar
FXDude
Posts: 1129
Joined: 19 Jun 2012, 21:59

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by FXDude » 24 Feb 2015, 05:37

mantragora wrote: ICE in nowhere close to 90% of Houdini flexibility. ICE may equal only to SOP part of Houdini, and then there is DOP land which is as big as SOP land. And of course there is COPS/CHOP. Each of them can be mixed with other.
Fair enough :ymsmug:

User avatar
Mathaeus
Posts: 1778
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia
Contact:

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by Mathaeus » 24 Feb 2015, 10:42

mantragora wrote::)


BTW. Have you guys tried rigging in H14? Do you feel that new viewport is faster than old one with heavier rigs?
A lot faster than before, but still nowhere close to XSI performance, especially not nowhere close to Maya. All that with something that looks like only as a brutal black-boxing, perhaps they tried to get some small piece of speed, on every price. Example: Capture SOP returns one (1) attribute.
While in ICE pre - kinematics age ( that is, since XSI 7.01 from 2008) - XSI envelope operator, in weighting stage, returns four, two of three of them more than enough for building the custom envelope, on top of one of (still) best weighting system on the planet. All that without going into loops in ICE tree. That is, for those who don't know, Houdini got nodes for array manipulation, only in v14 - while plain ICE 'array average" still does not exist in H ( can manipulate with indices, but still not with values from array members - actually it can do some kind of sorting).
That's about flexibility. However, ICE superiority over Houdini, doesn't matter that much, when it comes to performance of characters. SI users will know, that, even in best case of plain linear blend envelope, ICE variance is still slower, around 20 percent, than standard Si envelope.
In other word, flexibility and performance does not like each other. My small impression is, that, *any* of these flexible system, like ICE or entire Houdini, simple can not compete against 'hard coded', from scratch solution. Or in yet another word, once Side Effects, as developer, decide to compete against Maya, something will happen, to move Houdini rigging from zero state. For now, Houdini rigging is just not here, yet. In meantime, flexible system could be used for *exceptions*, interesting in some dark room of some company, or in head of some fanatic - but irrelevant anywhere out of this zoo, not able to show any wider influence.

nodeway

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by nodeway » 24 Feb 2015, 11:59

Mathaeus wrote:My small impression is, that, *any* of these flexible system, like ICE or entire Houdini, simple can not compete against 'hard coded', from scratch solution.
I wouldn't call it impression, Fact is more appropriate word. Maybe if there was possibility to compile network to pure code, once you have everything what you want, maybe then it would give comparable performance.

I rig once in a while and I prefer flexibility.

Mathaeus, have you tried Fabric Engine? In theory they should have faster solution (that's what they say), so maybe using it for rigging would be a good choice?

User avatar
Mathaeus
Posts: 1778
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia
Contact:

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by Mathaeus » 24 Feb 2015, 13:57

mantragora wrote:
I rig once in a while and I prefer flexibility.

Mathaeus, have you tried Fabric Engine? In theory they should have faster solution (that's what they say), so maybe using it for rigging would be a good choice?
Didn't tried Fabric. Always been small dog, now I'm too old to change that. That's my point of view, nothing else. Houdini got into focus because of upcoming c4d connection, finally giving to Houdini authors the best what could happen to them (imho) : feedback of huge community. Occasionally, small hair system I've running, still is mentioned here on forums, sometimes. While this thing, created with ICE, gave me a tons of headache in around five years, feedback was also a motivation. Otherwise it wont go so far.
Should I say that features, I've considered as best part, usually are not considered in the same way, by users ( nicely to say).
One nice advantage of exactly c4d people, imho, is that they aren't influenced that much by practice of big players, like Maya people, perhaps it's easier for them to adapt something new and unconventional, as long this fits into their everyday work. For now, really don't know what this is, but... never say never.

Maya is unavoidable. As an Softimage kid, I still hate it - but I have to admit, Maya community always been a huge resource of ideas and experiments - so perhaps not bad idea to dive into that ocean, too.

FabricPaul
Posts: 188
Joined: 21 Mar 2012, 15:17

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by FabricPaul » 25 Feb 2015, 20:58

http://fabricengine.com/double-negative-customer-story/

Hope that helps :)

I can't comment on what we're working on right now, but you're going to start seeing some very interesting news from us in March/April.

User avatar
Mathaeus
Posts: 1778
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia
Contact:

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by Mathaeus » 25 Feb 2015, 22:11

FabricPaul wrote:http://fabricengine.com/double-negative-customer-story/

Hope that helps :)

I can't comment on what we're working on right now, but you're going to start seeing some very interesting news from us in March/April.
Thank you for posting this - anyway I'm pretty sure, anyone who animated anything more complex than bouncing ball, will thinking exactly the same as Double Negative people. Waiting for news about Fabric.

nodeway

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by nodeway » 25 Feb 2015, 23:05

FabricPaul wrote: I can't comment on what we're working on right now, but you're going to start seeing some very interesting news from us in March/April.
1. How it's going with node based interface for Fabric?
2. When we will see Houdini integration?

FabricPaul
Posts: 188
Joined: 21 Mar 2012, 15:17

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by FabricPaul » 25 Feb 2015, 23:54

My apologies - I just realised this was the Houdini section (I had a google alert on Fabric and I went straight to the thread). I thought this was in the main section.

Anyway - to answer the questions:

1. The visual programming system is coming along extremely well and is currently in closed alpha with our customers. We're giving a presentation at GTC that shows a bit more: http://fabricengine.com/gtc-2015/ We will start showing public videos in March and we'll be giving presentation sessions every day at FMX.

2. Houdini integration will be part of 2.0. We decided to hold back rather than do the work for FE1.x and then push people to migrate. It makes much more sense to hook up with the visual programming system since that should feel quite natural to the Houdini crowd.

Cheers,

Paul

User avatar
SamHowell
Posts: 364
Joined: 09 Jun 2009, 14:09
Location: Birmingham
Contact:

Re: Plain Houdini tricks in Softimage style, Volume 00001

Post by SamHowell » 26 Feb 2015, 10:48

FabricPaul wrote:My apologies - I just realised this was the Houdini section (I had a google alert on Fabric and I went straight to the thread). I thought this was in the main section.

Anyway - to answer the questions:

1. The visual programming system is coming along extremely well and is currently in closed alpha with our customers. We're giving a presentation at GTC that shows a bit more: http://fabricengine.com/gtc-2015/ We will start showing public videos in March and we'll be giving presentation sessions every day at FMX.

2. Houdini integration will be part of 2.0. We decided to hold back rather than do the work for FE1.x and then push people to migrate. It makes much more sense to hook up with the visual programming system since that should feel quite natural to the Houdini crowd.

Cheers,

Paul
Image

Post Reply

Who is online

Users browsing this forum: No registered users and 18 guests