constant point ID's

[TheDunadan]

I'm having a little problem here; I'm modeling a face and I haven't made up my mind whether I'd like to go for pointy ears or normal ones.. full lips or .. etc.

Ideally I want to model each of those features such as ears, eyes, lips, noise etc. in different versions so I can choose later on which one to take. I know if I actually model each feature seperatly Houdini I won't have any problems joining the different version together...

But sometimes I want to supress a few OP's further up the SOP's network and if these SOP's created / deleted points the inputs on all other following OP's in the chain will get messed up, since some points etc. changed ID's. I know there is little that can be done about it if the required inputs are missing for the following OP's. But in those areas that weren't effected by those OP's I'd like the following OP's to work as they did before I supressed / deleted some OP's further up the SOP network. I think most OP's should do just fine as long as those once used point ID's stay reserved and don't get assigned anymore: is there a way to enforce that those ID's stay constant and there is no renumbering of the existing points / primitves ?

Oh and any other tips & tricks on keeping polygonal models nice and flexible are most welcome as well Or anything that users who worked with Maya or alike might not think of.

Btw. here is my first modeling attempt in Houdini... gonna continue working on it. It's main purpose besides getting used to Houdini's modeling tools is to have some test object for skin shading or alike. For all those who aren't particulary keen modeling (like I do usually ) I'll put it online for download, but first I gotta put it into a decent state and that can take quite some time

Jens

[edward]

Not really. But as I've mentioned several times by now, there are strategies for dealing with this. Off the top of my head:

- If there's certain lists of point ids that you use frequently, then using named groups instead will be better.

- In your specific case, I would branch off of your base head for each specific feature. Create groups out of your base primitives and then delete the non-selected primitives for each group. Now you can model each feature independent of one another. Use a Merge SOP to put them back together.

[DaJuice]

Hmm, I remember there was a discussion about this exact subject on the sidefx board. I should try to find it again as it was very interesting, but I just wanted to say nice model.

[3__]

I was a bit disappointed by this limitation,

given that I was expecting sops to be

procedural and all... but the tools are now

quick and useful enough to just re-edit

(pile on more nodes) and occasionally

collapse (to subnet of file).

btw, nice model! is the wireframe subdivided?

-cpb

[TheDunadan]

@Ed thanks, I expected that seperating the object into the different features and merging them together later would be the best way, yet this requrires some preplaning and this would require my lil brain cell to actually work constantly when modeling . With those switch / merge OP's it works fairly nice though (I remeber those lil 'creature creator' tools for max ... not difficult to do the same with Houdini here).

Since we can branch of the OP network at any time it's possible to at least save the 'good' areas and replace those with the unwanted ones, this way things can be undone fairly well, though it requires a bit manual work. I won't complain maybe ...

@DaJuice I tried finding the thread at sidefx, but failed. If by any chance you come across it, please lemme know.

@cbp I'm fairly happy with the SOP's (there are always a few things we wish for), but I very rarly model at all and therefore more full time models might see things differently; I don't see anytools I miss from Maya and I actually prefer Houdini's tools here, but I know Max and LW modelers often go nuts when having to model in Maya (i.e. maya might not be the best tool to compare to).

I played around with instances and other more classical procedural model tasks and I don't think anysoftware comes anywhere close to Houdini here. And there are as well completly different areas I never expeted modeling tools to be any good for: e.g. I always thought writing Shaders would be all there is to things such as NPR style rendering, but Houdini proved me wrong: amazing what can be done here as well on SOP's level.

Anyhow, the wireframe isn't subdivided, it's what I'd call the rough prototype shape. Once the overall geometry is done I intend to subdivide it further.

Oh, and thanks for the positive comments

Jens

[edward]

I would contend that there doesn't necessarily need to as much preplanning as one would think. I can't argue with needing more brain cells though.

- Groups can always be added after the fact and inserted into the right place. Then named groups can be replaced where things break.

- Branching out can be done at the end of the chain. For those separate features, don't modify upstream. Instead, split them at the end of the chain to further modify them.

What I'm trying to say is that yes, it takes some more work but I don't see why it needs pre-planning. You can always change things at the point you realize that you do.

[stevenong]

Hey Jens,

The model looks great! Keep it up!

Cheers!

steven

[Moose]

Nice looking model so far Jens - any chance of a wire (just to give it some closer scutiny)

Just to add my 2cents concerning Houdini's modeling tools...I gave Silo a whirl a few days back, and although it has a few tools I'd like to see in Houdini, I found it hard to re-adapt back to a system which, compariivly speaking, was devoid of any proper history. There's something to be said for being able to see every move you make and to have the option to go back in and change things. Granted, I only spent a couple of days with it, but even so I did feel pretty quick with it (it's not much to get your head round), it's just the whole experience felt uneasy - like the feeling you get when you're a week swimmer in deep water.

I guess what I'm trying to say is that even for tasks like organic modeling, having your nodes laid out all nice and clear in the network editor is a real security blanket, and easily taken for granted. Though of course I'm sue if I spent longer with Silo this feeling would fade some...but still.

[TheDunadan]

@ Edward: A bit of preplanning doesn't hurt. Adding without much thinking a whole bunch of SOP's does hurt At least my SOP network is hardly controlable anymore, but I know now what I'd do differently next time. And I ordered two new brain cells .. just in case

@ Steve thanks

@ Moose The wire will be up soon and along with the hip file, so everyone who wants to can have a closer look.

I come from Maya and when a friend tried to convince me of lw for modeling I was complaining about the almost complete lack of history. I suppose once I got used to Houdini's 'network history' Maya will drive me nuts just as well when having to work with (like it does know already when faced with a particle effect and my mind goes 'we just need this OP and that one .. and than I awake from my POP dreams and face the MEL script editor )

Speaking of Maya, I dunno who else works with Maya on a regular basis, but I feel like having drunk 10 pints of beer when trying to navigate in the viewport after having worked with Houdini before and this hotbox constantly pop's up and switches the panes. Friend of mine had a good time when he was sitting next to me ....

Jens

[TheDunadan]

Just a small update on the actual model.. doesn't have much to do with the intial topic anymore ... ... here it is

[TheUsualAlex]

Just a small update on the actual model.. doesn't have much to do with the intial topic anymore ...  ... here it is

14804[/snapback]

Cool! Watch out for the corner of the lips, tho.