Re: Schematic View

["Thiago Costa" <thiagocosta3d(at)gmail.com>]

Luc-Eric,
That's a final decision that the Schematic will not change much... right?

A new view?... I don't know if I like that but anyway... You said to keep sending suggestions, so here I go:

It's called the "Generic Expansible Super-Ultra-Master-Blaster Node Based View" - or just GESUMBNBV (for marketing purposes). Isn't good? :)

Simplifying... it would be a nodal representation of the explorer, with editable connections. bam!

- View Modes
It use Scopes (like the explorer does), where we can see the Scene, Passes, Materials, Application... All of it would be represented by nodes and it connections.

It could have more specialized modes, like "Rigging" where you can hang a toolbar for rigging in the same view... so specific modes would have specific menu context and tools... and specialized visualization modes. "Actions" for example could show external actions, mixers and it connections. And let you easily edit it and copy/connect nodes from one mixer to another.
So it's basically let you do everything you can already do in the interface, but dragging connections. Allowing you do quick interaction with the represented data.

eg: Passes could be represented as the explorer does...:

[Default_Pass]
   \--[shared render options]
   \--[partition]--->[override]--->[objects]
   \--[partition]--->[shader override]-->[objects]

Each node would have a self-explanatory style... With specific color/size. Nodes that represent Passes could be a little bigger than others... all those design/usability things, you know.

I vote for specific modes on this view because this way you don't need to show every possible graph connection, like hypergraph does. It would be more specific, so this view set to show "passes" would permit connections related to that. Exactly like the explorer lets you "add partition" with a right click when viewing Passes, but don't care at all with passes if you are viewing in Scene mode.

It's a way to do not scare the user with a gazillion options that you don't need to show. It's an extension of the (so good) workflow we have in the explorer.

- Project/Application representation.
Something that goes beyond the limits of the scene, this view could represent your project structure, just using nodes.

Example - I already suggested that but here I go again:
Would be great if we could have 100% feedback of refmodels in the new "Schematic":
I wish we could see files that each delta is outputing, and what file is feeding the refmodel... In a scene populated with 100 refmodels, it's painful do not have an overview of all your refmodels/where they come from or what .delta they send out.

I wish we could put floating nodes in the graph that represent files of your project... I mean drag and drop models/clips/actions/deltas on your schematic view, and be able to connect them to models.

- Group nodes

I wish we could see groups in the schematic. 
I mean:
Char_ctrls_hand
   \
    \------------------
    |  left_hand    |
    |  right_hand  |
      -----------------

It's not the group_node itself, but a big node hosting all the objects inside.

- Visual representation
You know those boring diagrams/flowchart you see in economy classes? So it should NOT look like that, in my opinion.
It must be efficient and easy to read. Right now a bad example is the schematic. With 50+ nodes in your scene you can't recognize what is what. I mean make a biped, open the schematic and zoom out. Then tell me what is a model, a chain or a polymesh. It's impossible.

A "node" is a very technical guy by nature, but it doesn't need to be represented as a little bar with a text on it and 50 arrows going to different directions. 
It can be more organic/mutable, hosting it relations instead of just indicating it. Of course it's necessary that the user can rearrange the graph manually, but if a node changes itself depending on what it's doing. It would be way more "informative" by it nature. 
eg: Simulations enviroments could be represented in the graph, and all
the objects that act inside of this enviroment could stay right next to
it by default.
a refmodel could have a specific type of node on top of it that indicate how many resolutions it have, you put your mouse on it and it says what file it's. So you query parameters values but it happens by levels.

This way "nodes" are entities with a very specific drawing style, where you can identify what it do at first sight.
Icons are friends but they get small in a huge graph, and sometimes is hard to make an icon say the same thing to everybody... So I believe the shape and color of each node would help to make it clear... Example: a group could be a circle with other nodes inside. a Model could be a long bar hosting all the hierarchy under it. A skeleton chain could be a sequence of connections where you could "stick" nodes in a row...

To make it clear... The disposition of each type of node and it drawing style, is
what will stay on your mind. Looking a graph with 300 nodes is easy to
identify circles, squares, chains, triangles... even if you are 
looking from very far, graphic primitives will always be abstract/easy
to read. There's no reason to do not use it.

I know that some Technical Directors like the most simple straight thing with an ugly arrow and a text. But interaction/visual feedback can save you crazy amount of time if it's done right. And this is why tools in XSI were always fun when I first started on it...

The primary objective of this view should still be represent objects----->relations. So It could be all driven by the same way/engine that display the explorer data.
I mean, this new view it's just how you show data you already have, and how you'll let the user interact with it...

So this is what I think about have a new "interactive" view... maybe 50% useless, maybe not. point it out 

Thanks for read.

Thiago Costa
thiagocosta.net

On 03/03/2008, Matt Lind <mlind(at)carbinestudios.com> wrote:

The last thing we need is yet another view.  I'd rather see consolidation.  If this new view comes online, then I'd prefer to see more intelligent display filters so it could be dumbed down to the existing schematic view, then remove the old schematic view.  For example, a rendertree like nodal graph for operator connections as displayed in the moondust demos, but for those who want a straight schematic, perhaps just hide all the operator nodes and operator connections.  It's getting very difficult to train new users on XSI from the kubillion options already on the table.  It's becoming more an excercise of what not to show them because early on in their experiences, there's so many ways for users to click and get into trouble.  XSI could use some streamlining in the UI, this is a good place to start.

 

Matt

 

 

 

---

From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of Luc-Eric Rousseau
Sent: Monday, March 03, 2008 8:55 AM
To: XSI(at)Softimage.COM
Subject: RE: Schematic View

Well ,  I wrote this to explained the current Schematic view in response to comments that it fails at being a nodal editor à la Houdini.  It's not what we were going for in that view, so it's not a failure of the view that it doesn't do this, or generic relationships à la Hypergraph.  That sort of workflow would probably go in new view, so keep the suggestions coming.   I've been thinking that another view may better reflects the dual-world of TDs and animator, for example.  For the Schematic itself, we are more cautious and conservative.

 

From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of kim aldis
Posted At: February-29-08 6:03 PM
Posted To: xsi
Conversation: Schematic View
Subject: RE: Schematic View

 

I so don't understand this reluctance to create a decent hierarchy and relationship viewer in what is, after all, a hierarchical data set. You desperately need a node view for viewing relationships and you desperately need to improve the usability of the schematic. Surely it's not impossible to add that functionality without breaking it for those that use it as is.  I'd be willing to bet most of those would be ecstatic to be able to see and manipulate relationships in the schematic and those that don't use it for the most part don't use it because it needs badly to be improved.

 

Your argument makes no sense and if there's anyone on this list you should be listening to  seriously about this it's Brad. He has a better understanding of this and articulates it better than probably any other user you have.

 

 

From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of Luc-Eric Rousseau
Sent: 29 February 2008 20:06
To: XSI(at)Softimage.COM
Subject: RE: Schematic View

 

Actually the schematic in XSI isn't driven by XSI's architecture, but rather by the workflow for a specific group of Softimage|3D users.  It's not meant to show operators or construction history, and as you can see moving nodes around is a secondary action (i.e. you can't just click and move around nodes by default), and it remembers user positions of the nodes you modified.  We do not really want to show operator in that view or complicate that view with more features that then "breaks" it for these users, it's a conscious decision.

 

XSI's Schematic was also a good way to view and edit shared materials, but the size of the scenes and practices have evolved.  The Material Manager is the best place to view and edit these relations today, with features and including "Who Users?" tab.  SI|3D also didn't have the powerful material inheritance feature that XSI has, simple things like materials-on-groups.

 

 

From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of Bradley Gabe
Posted At: February-29-08 2:34 PM
Posted To: xsi
Conversation: Schematic View
Subject: Re: Schematic View

 

[…] As far as improvements go, it also seems there are two areas, aesthetic improvements such as node size and layout, and functionality improvements such as the ability to rewire operator connections. Aesthetic changes don't require updates to XSI's core the way many functionality improvements do, which is why we've seen a few changes in aesthetics of the schematic in the past few XSI updates. We haven't seen any major udates to functionality such as the ability to rewire operators, not because Soft doesn't want to, but because what we already have in the schematic is likely a manifestation of XSI's current core architecture.