PDA

View Full Version : How would you make the LW NODES more Intuitive?



robertoortiz
01-17-2014, 08:20 AM
Hey guys,
I am a long time lw user (since ver 5) but I have to admit that I had to be dragged kicking and screaming into using nodes.
I like them know, but I do find them a bit too abstract for my taste.

Also while playing with them I realized there is a lot of room for improvement for them.
So tell us,
How would improve the LW Nodes. how would you make them easier to use.

I am looking forward to your comments and Ideas.

-R

robertoortiz
01-17-2014, 08:24 AM
OK here are some of my ideas.
They might be in already place, but I got no clue how to activate them.

I would add an ABOUT button for each NODE.

Sade on the NODE MENU fro mthe shader screen
You select
SPOT > Instance Info

And then the NODE control comes up I would have at the end an ABOUT button.
It would show:
NODE NAME
FUNCTION
DESCRIPTION
ATTRIBUTE DESCRIPTIONS

Simple to implement, since most of this info is already part of the documentation, but it would help a LOT.

Also
How the heck do you know the values coming fro ma NODE attribute?
Right now, I take it on faith that a series of values is coming from a NODAL attribute.
This means for me that, there is too much abstration..

How about a DIAGNOSTIC tool that would allow me to READ the values, or a sample of the VALUES coming from a NODE attribute.
Or better yet a SCREEN TEXT NODE that would allow the display a VARIABLE values.



What do you guys think?

-R

Ryan Roye
01-17-2014, 08:32 AM
One of the setbacks of using nodes is that there is little to no bulk management options available (this is an issue in several other areas of Lightwave as well, such as motion modifiers). You can't easily make changes to one node on one object and retroactively have that change apply to other things that might use that exact node setup. This makes the nodal workflow highly impractical in areas that it should not be. I shouldn't have to avoid nodes for the fear of making a mountain of work for myself later on down the road.

hrgiger
01-17-2014, 08:41 AM
One of the setbacks of using nodes is that there is little to no bulk management options available (this is an issue in several other areas of Lightwave as well, such as motion modifiers). You can't easily make changes to one node on one object and retroactively have that change apply to other things that might use that exact node setup. This makes the nodal workflow highly impractical in areas that it should not be. I shouldn't have to avoid nodes for the fear of making a mountain of work for myself later on down the road.

Yes. As far as I know, the node editor only works on a per surface basis or per object basis. There's no way that I'm aware to do any sort of 'global' networking. You could do it in the CORE connection editor. I could drag anything in the scene I wanted into the connection editor, and use one node to control several objects. That functionality in LightWave would be very welcome.

As far as an about interface....I would rather simply have a interactive help menu sort of like Zbrush has. If you didn't know it was there, it would never be in your way. But if you hold the ctrl button down and hover over any tool in Zbrush, it wil give you a pop up window with information about the tool and how it is used.

Sensei
01-17-2014, 09:24 AM
Yes. As far as I know, the node editor only works on a per surface basis or per object basis. There's no way that I'm aware to do any sort of 'global' networking.

???
It's possible for 2 years using Global Materials http://globalmaterials.trueart.eu

Danner
01-17-2014, 10:29 AM
...How about a DIAGNOSTIC tool that would allow me to READ the values, or a sample of the VALUES coming from a NODE attribute.
Or better yet a SCREEN TEXT NODE that would allow the display a VARIABLE values.


There is a probe button on the top left of the nodes window that lets you see the values of outputs if it's active and you mouse over any output. and there is also this:
https://www.lightwave3d.com/assets/plugins/entry/input-spy/

hrgiger
01-17-2014, 10:48 AM
???
It's possible for 2 years using Global Materials http://globalmaterials.trueart.eu

I'm talking about in LightWave, not through a third party. But that's not really a global network shading system anyway. You've just created a node to allow multiple inputs for scene surfaces in which everything must first be converted to a material to make use of. It would just be a lot more useful (and simple) to drag surfaces that we want into a global network nodal interface and make the connections we need on a scene basis. Converting simple textures and colors to materials so that you can globally shade everything in one interface is not ideal.
And just to be clear, I'm not saying that's not useful. It looks very useful actually and I appreciate the third party work you do for LightWave, but we're talking about how to made nodes more intuitive. I just don't feel in this case, this plug-in does that.

Sensei
01-17-2014, 11:22 AM
Your solution is far from ideal work flow also.
You have to first open global node editor, add some node, connect it to empty slot.
Then find object and surface in Surface Editor, click Edit Nodes, then add global node, and connect newly added output to some input in that Node Editor.
Then open node editor of 2nd object, other surface, repeat.
Then open node editor of 3rd, ... 10th object, repeat over and over again...
The more objects/surfaces, the more hassle, and most of it is clicking surface and "Edit Nodes"..
All of these just to be able to set color simultaneously for couple objects/surfaces?
We can do it right now, by making null item, and taking null position using item info.. and copying and pasting it over and over again in node editors.. Then set null position will change color..

Not to mention, the most likely it'd impossible to implement it as plugin, if global node would be deeper in tree..


It would just be a lot more useful (and simple) to drag surfaces that we want into a global network nodal interface and make the connections we need on a scene basis.

Now it's sci-fi... LW doesn't have built-in drag'n'drop functionality at all..

hrgiger
01-17-2014, 11:41 AM
Your solution is far from ideal work flow also.
You have to first open global node editor, add some node, connect it to empty slot.
Then find object and surface in Surface Editor, click Edit Nodes, then add global node, and connect newly added output to some input in that Node Editor.
Then open node editor of 2nd object, other surface, repeat.
Then open node editor of 3rd, ... 10th object, repeat over and over again...
The more objects/surfaces, the more hassle, and most of it is clicking surface and "Edit Nodes"..
All of these just to be able to set color simultaneously for couple objects/surfaces?
We can do it right now, by making null item, and taking null position using item info.. and copying and pasting it over and over again in node editors.. Then set null position will change color..

Not to mention, the most likely it'd impossible to implement it as plugin, if global node would be deeper in tree..



Now it's sci-fi... LW doesn't have built-in drag'n'drop functionality at all..

Gee, I wonder how all those other apps do it :)

But in any event, no your summarization of how that might work is not really accurate. You're only thinking in terms of how LW works now and not taking into account changes that would go along with a global network. I'm not sure why you think you would have to be opening up separate instances of node editor for each surface and/or object, that defies the whole point of a global system. Multi-select the objects/surfaces from a scene editor, drag and drop them into nodal, and make your connections all in one go. This isn't about how LW works now, this is about how we'd like it to work. So yes, of course its Sci-fi at this point.

jwiede
01-17-2014, 12:30 PM
Gee, I wonder how all those other apps do it :)

Well, as I'm certain you're aware, in other apps such functionality works pretty much just how you described: Multi-select objects/surfaces, drag to node editor to expose them as nodes, and so forth.

Lack of support for these kinds of modern drag-and-drop scenarios is a strong contributor to why LW's UX feels so outdated and clunky. If we're really talking about how to make LW support for nodes more intuitive, then adding pervasive drag-and-drop support to/from viewports, scene, surface, and node editors; etc. is another (long-overdue) necessity.

VonBon
01-17-2014, 12:34 PM
But if you hold the ctrl button down and hover over any tool in Zbrush, it wil give you a pop up window with information about the tool and how it is used.

Word, did not know this :thumbsup:

I'm scared of Nodes, but it would be nice if they worked like Surface presets in a way.
You could build a node in a main "Setup Environment", and once a Node is changed in
this main "Setup Environment" it will change it globally throughout LW. Could also have
a Check Box for Local and Global controlled nodes if you didn't want a particular surface
of an object to update globally.

I need to work with Nodes more. :foreheads

hrgiger
01-17-2014, 12:36 PM
well to be fair LW3DG (and I think specifically Matt Gorner) added the ability to search and bring up nodes by type which is great. But yes drag and drop is simply a no brainer.

creacon
01-17-2014, 12:46 PM
You can drop your license key and you can drop lwo files in layout, maybe lws too, but that's about it.




Now it's sci-fi... LW doesn't have built-in drag'n'drop functionality at all..

Sensei
01-17-2014, 12:53 PM
Dragging and dropping files to the main window is handled by OS.
Programmer has just to add one call-back function to code.
I have made d'n'd of images a long time prior NT added it.
You can see it here
http://www2.trueart.pl/?URIType=Directory&URI=Products/Plug-Ins/DragDropImage

It's different thing than dragging and dropping between GUI objects-controls.. It must be implemented in GUI at so low level, to work properly, I am not sure how it would be added now..

ps. you forgot d'n'd images. It works too.

jeric_synergy
01-17-2014, 03:11 PM
Multiple LAYERS of help, including notes to oneself. IE

1-hover gives you Node Type/Name (in case you've changed it)(pretty much instantly)
2- longer hover gives you paragraph, editable by user to make more (personal) sense
3- oh, you could just fold this into #2-- link to appropriate page in manual (appears)
4- allow the user to adjust the hover times

Global toggle for native names versus user names of nodes (ala AE Timeline Layers)(as described here on the forum)

As mentioned, d-n-d : that trips me every time.

User adjustable font sizes (personally, a bit small for my eyes)

Visual adjustments: user defined backdrop color, line thickness, arrowhead size, and as above, font size

"Favorites" panel, holding most of one's most commonly used nodes and COMPOUND node arrangements. And entire networks (which you can currently SAVE intact, but since we are WISHING here...)

+++EDIT:

Here's a whole NEW thing/feature: A "Spot Info Crosshairs"-- a lot of time the "logic probe" and/or Input Spy Node is not really useful because the network is EVERY SPOT, ALL THE TIME (paraphrasing Rebel Hill here). If you could isolate ONE spot, that would sometimes come in useful in debugging your network. Well, how about a crosshairs that isolate the network input to ONE particular "spot" in the render, so that users could say "what type of output am I getting HERE?"

Davewriter
01-17-2014, 08:13 PM
I agree with VonBon - I'm scared and I need to work more with it.
There is a certain level of frustration (as a practicing dunderhead) when I watch someone who really understands Nodes work. All of those things that make perfect sense to them are all things that I would never had dreamed of using together. I take them for their word when they say that all you need do is take an Astrial Envirnnment node, link it to an Atomic Syntax and the outcome will be Red to the third level.
Which brings me back to "I don't get the basic how's" so I stop trying.
Yes there is much power there... but I feel rather the chimp with an infinate number of lighters, standing in a fireworks factory.

jeric_synergy
01-18-2014, 01:24 AM
Fortunately, this thread is about "MORE intuitive", so it's all uphill from here.

Rebel Hill's first part of his node tutorials are now FREE online. Start there.

raw-m
01-18-2014, 02:33 AM
For me, there's far too much clicking! I'd like to see more sliders at the default view with input fields - or a least functionality used in the color node where you can drag values left/right. Simple feature but would have a great impact on workflow speed.

Polygert
01-18-2014, 06:40 AM
I personally would love to be able to open up a material inside the node editor itself, now i have to open up my surface editor choose the material and click the "Edit Notes" to open up the node editor and for each material i have to do the same over and over again....to many clicks

1). If only it would be possible to open up the node editor itself, have all materials presented inside, clicking activates them as nodes and makes you enter the node setups, clicking home button puts you back in the materials main node shelf.....
2). inside each material node you should be able to put lesser needed nodes into a container node, hiding some nodes if wanted to unclutter the overall view of the node.
3). each material in the main node shelf as every container node should have the ability to put some parameters set up by the user accessible outside, so user can change these parameters (example mix value) without going in to the node itself so changing the appearance off a material...
4). maybe some kind off preset shelf inside the main material shelf itself so you can pick and drag them to assign ?

etc...


Just an idee

High regards,
Polygert

jeric_synergy
01-18-2014, 11:31 AM
Good ideas here....

raw-m: "Click Reduction" is a noble goal, you are right. I'd like for the devs to sit and WATCH users use the Neditor (yeah, I went there) and see what can be done in that respect.

Another overarching goal or principle should be MOUSE MILEAGE REDUCTION. Time spent waving the fracking mouse around is time wasted. When possible, esp for smallish dialogs (eg Scaler & Vector values), dblclking the node should open up the dialog DIRECTLY UNDER the mouse.

I'm not sure this would be useful, but one idea for CLICK REDUCTION would be a feature that displays the dialog panel on rollover/hover. This should probably be optional, and might work best if the Neditor Sidepanel is open. This would allow the user to scan the values w/o clicking nodes separately.

"Clicky" interfaces get tiring fast.

++++
Polygert, I think you are using "Materials" in non-LW fashion-- I think you mean "Surfaces". Materials in LW are something else.

jwiede
01-18-2014, 04:17 PM
Good ideas here....

raw-m: "Click Reduction" is a noble goal, you are right. I'd like for the devs to sit and WATCH users use the Neditor (yeah, I went there) and see what can be done in that respect.

Another overarching goal or principle should be MOUSE MILEAGE REDUCTION. Time spent waving the fracking mouse around is time wasted. When possible, esp for smallish dialogs (eg Scaler & Vector values), dblclking the node should open up the dialog DIRECTLY UNDER the mouse.

I'm not sure this would be useful, but one idea for CLICK REDUCTION would be a feature that displays the dialog panel on rollover/hover. This should probably be optional, and might work best if the Neditor Sidepanel is open. This would allow the user to scan the values w/o clicking nodes separately.

"Clicky" interfaces get tiring fast..

QFA. LW can feel like a sea of dialogs -- move, click/open, move, click/open, repeat ad nauseum down through the dialog strata. Other pkgs have become a LOT smarter about anticipating users' intents, and either automatically displaying attributes where expected, or just not hiding them from display in the first place.

As others have already suggested, either putting a permanent "node attributes" panel at the bottom of the editor (ideal), or having it appear when the user hovers over the node for a bit, would prevent the endless clicking and mousing of nodes.

Here's an even better idea: Open up SDK access for _official_ programmatically access the contents of node flows, with specs and all! That way at least third-party developers can try to address the node editor shortcomings by writing replacements, and can experiment with new UX designs. Nodes still remaining closed off from both SDK and scripting interfaces (practically, anyway, it can be done but without specs it's a fragile trial-and-error dev sinkhole) this long after their pervasive integration is kind of silly. I can't see it happening any time soon, but hope springs eternal.

jeric_synergy
01-18-2014, 05:03 PM
Other pkgs have become a LOT smarter about anticipating users' intents,
I got a LOT of pushback from Lightwolf when I suggested that, if a user is attempting to adjust a Null's color, the user most likely wants to adjust that color.


119409
But noooooooooooooooooooooo. :devil:

Instead an extraneous click, no, TWO extraneous clicks are forced upon the user when s/he falls afoul of this 'law'. (Just one if s/he remembers.)

One of a myriad of examples. The software is FIGHTING the user in these cases. It's RESISTING. That's probably not what one desires from the software.

Davewriter
01-18-2014, 08:48 PM
Some sort of consistancy.
This node is a 1 - 100, but you can get an interesting look if you bump to 2000... while this one is 1 - 10 and this one is 0 - 1 but you would rarely ever use it higher than .25

Riff_Masteroff
01-19-2014, 04:01 AM
Nodes are powerful. More so than layers. Much more so than layers. Get used to it.

I have been comping in Fusion lately. Compositing passes from LWx64. In essence, Fusion is all nodes and NO layers.
Its workflow is powerful and intuitive. So many choices, so many ways to do something better. Complexity is a 'coin' with two sides:
Whine about needing nothing that will help you stay on top of what is do-able. Or learn. The team at eyeon does a marvelous job.

Like Roberto, I have been working with LW mostly since version 5x. My suggestions are as follows:

A) Bite the bullet and get rid of layers altogether. Retro apps, Adobe apps, are history . . . . stop working with them.
Trying to imitate those retro apps is not good.
B) The NT team 'should' view, concentrate and incorporate what other apps have done with nodes.
Really, not much is needed. NT is not behind. Could use some finesse.
C) Am talking about layout, don't even try nodes on Modeler geometry. I am OK with nodes on Modeler surfaces.

Polygert
01-19-2014, 04:10 AM
Nodes are powerful. More so than layers. Much more so than layers. Get used to it.

I have been comping in Fusion lately. Compositing passes from LWx64. In essence, Fusion is all nodes and NO layers.
Its workflow is powerful and intuitive. So many choices, so many ways to do something better. Complexity is a 'coin' with two sides:
Whine about needing nothing that will help you stay on top of what is do-able. Or learn. The team at eyeon does a marvelous job.

Like Roberto, I have been working with LW mostly since version 5x. My suggestions are as follows:

A) Bite the bullet and get rid of layers altogether. Retro apps, Adobe apps, are history . . . . stop working with them.
Trying to imitate those retro apps is not good.
B) The NT team 'should' view, concentrate and incorporate what other apps have done with nodes.
Really, not much is needed. NT is not behind. Could use some finesse.
.

That's kinda what i meant, for me the surface editor should reside inside the node editor...like a single work spot for surface editing.

Regards,
Polytek

Lewis
01-19-2014, 04:55 AM
- make them visible in OGL (that's no1 killer for me currently 'coz LW don't see any of nodal inputs in OGL (not even color or first level depth of textures)
- make them update nodal flow if UV name is changed (layers update all materials accordinglly if you change UV name while nodes don't so it's breaking complete materials if you do that = bad :()
- defenitily need global acess and drag and drop so we can reuse same nodal aproach for different surfaces/materials
- nodal search filters should stop require ENTER key to search, it shoudl just filter as long as user types letters instead waiting for Enter key to actually filter (slow and too many clicks ;)).
- needs some kind of navigation panel so we can see how zoomed in node flow we are.
- some nodes don't work with side panel but still use their own popup (like gradient node and few more) = bad workflow (different panels options probabl yin GUI code). If we have side panel which can be collapsed/expanded then ALL nodes should worh in it and not just some so user can use popup on all or panel on all depending what he/she prefers.

djwaterman
01-19-2014, 08:41 AM
How about if the little Activate Nodes check box also appeared within the node editor as well, so you can activate the nodes on or off while still in the node editor, currently if you haven't already activated it and you don't see your surface update in VPR, you have to navigate back to the surface editor to turn it on. It's a tiny thing.

robertoortiz
01-19-2014, 07:24 PM
Ok about if we use main layout window as an uber control center for the nodes.

Let say you have three characters on screen.
It would be cool if you could RIGHT CLICK on any of them and see the nodal map (if any) assigned to them.
the the nodal control window that shows up
you could group the nodes under :

OBJECT xxx.lwo NODES (all nodes, for every category for the object)
LOCAL MOTION (Motion nodes for the object)
LOCAL SHADING
LOCAL DEFORMATION
LOCAL SIMULATION
GLOBAL NODES(all SCENE nodes,organized by the object)

or you could make your attributes GLOBAL and share them to the other two objects.

Hell you could even see the GENOMA controls assigned to EACH object.

And to make things more cool , you could copy a LOCAL category, and paste on another objects.

jeric_synergy
01-19-2014, 09:37 PM
Need mockups. Sounds busy.

robertoortiz
01-20-2014, 09:18 AM
Need mockups. Sounds busy.

You got a point, I am working on something.

But please guys, do keep the conversation going.

3DGFXStudios
01-20-2014, 09:24 AM
Navigation in the node editor could be a bit more intuitive. For example, I like the eyeon fusion navigator. It lets you move to any place in one "click and drag" motion.

http://forums.newtek.com/showthread.php?139709-Node-editor-Navigator-Feature-Request

Simon-S
01-20-2014, 04:00 PM
Not read this whole thread so this may have been mentioned before but I'd love to see a short description for each node type, maybe by right clicking in the node list.

There are lots of nodes there which I have no idea what they are or how to use them, so some kind of help/description would certainly be an improvement for a nodal novice like myself.

jeric_synergy
01-20-2014, 04:11 PM
I sense a theme here....


Here's another documentation idea: a (switchable) MODE that when a node is added, the QUICK DOX sentence is automagiclally inserted in the node comment field.

Since this occurs at creation, it will not overwrite comments a user has added/altered.


"Hand-Holding Mode", or "Training-Wheels Mode".

robertoortiz
01-24-2014, 10:55 AM
Here is something I came up today

I decided to start simple.How about adding an about to both the main Node Editor and the individual node attributes?
I was thinking that it would be cool if you could right click each attribute?
119559

jeric_synergy
01-24-2014, 11:19 AM
By 'attribute' do you mean 'output'?

I hope people are noticing that a LOT of tips boil down to "more convenient documentation". It's not so much that the dox are lacking (although I often have to stop and think WHICH pdf they are in) but that accessing them is a speedbump when you are in the thick of nodal creation.

As usual, I think "on-demand" dox (eg balloon-help, et al) should be user-editable, so that users can add specific, personal reminders to their own help, and hopefully, share them with the LW community.

robertoortiz
01-24-2014, 11:29 AM
By 'attribute' do you mean 'output'?
.

Yes I would do this for both Input and Outputs.


I hope people are noticing that a LOT of tips boil down to "more convenient documentation". It's not so much that the dox are lacking (although I often have to stop and think WHICH pdf they are in) but that accessing them is a speedbump when you are in the thick of nodal creation.


Agreed. The amount of infromation that we need track when using the nodes is quite high. It would be less for a hassle if the control itself (and the Inputs and the outputs) gave you a bit blurb of what the heck it is supposded to do.





As usual, I think "on-demand" dox (eg balloon-help, et al) should be user-editable, so that users can add specific, personal reminders to their own help, and hopefully, share them with the LW community.

This is a fantastic idea. Also I would add the ability to be able to preview values coming from an input node by "RIGHT CLICKING" the connector between nodes.

EDIT

One thing I added was the Possible Input Nodes option.

For me it would be cool if after clicking it the node options menu on the right EXPANDED and all the possible input nodes you can use for an Input
(lets say the Color input node) were highlighted in RED.

jeric_synergy
01-24-2014, 12:00 PM
I think the new-ish "data probe" feature gives you values as per your suggestion, similar to "Input-Spy" (which SHOULD be named "Output-Spy", sheesh) but w/o needing to place a node.

I'd be happy to coordinate all these crowd-sourced documentation efforts, for a reasonable retainer. ;)

geo_n
01-24-2014, 07:12 PM
Am talking about layout, don't even try nodes on Modeler geometry.

There are some advantages for model geometry to be nodal or using modifier stacks for sure.

Here's an array node tool which shows how a mesh can be distributed with a lot of options non-destructively because of its nodal nature. Edit the model as you wish but the array is non-destructive. You can even "ground" the objects to a hill or plane which is a feature a lot of lwvers have been asking for years.


http://www.youtube.com/watch?v=a4uhxbSKCn4&feature=youtu.be

If lightwave can start to make some modelling tools nodal(bend, loft, etc) that would be great!
Then it could be animatable in the future.

robertoortiz
01-25-2014, 08:54 AM
If lightwave can start to make some modelling tools nodal(bend, loft, etc) that would be great!
Then it could be animatable in the future.

Agree 100%.
this is something that you can start slowly, right now, in layout.

raw-m
01-25-2014, 09:25 AM
This can already be done with DP Kit. This will only work more effectively if there is proper UI widgets onscreen to play with rather than second guessing what values do what I the Node Editor. Totally agree, though.

jeric_synergy
01-25-2014, 09:38 AM
Tools like "bend and loft" are the LAST things that need to be added to Layout piecemeal. However, a procedural/nodal Array, Matrix make a lot of sense. --Although, Instancer & Flocking already satisfies both of those.

Waitaminit-- you guys are "requesting" something WE ALREADY HAVE. You want a Matrix? Make a flock. You want items distributed across an area? Use Instancer. You don't want to see them? Parent them and collapse the null.

Topic drift.

--Sheesh.

geo_n
01-25-2014, 10:17 AM
Afaik there is no proper mesh/model editing tools in layout so an array would be an instance that you can't edit in the mesh level. Has that changed with the new instancing tools?

jeric - you haven't used bend and loft tools as modifiers? Not destructive tools as we have now. You're missing out the potential of having them as modifiers. And animatable, too :D

jeric_synergy
01-25-2014, 10:42 AM
In my limited experience in dealing w/procedural arrays, they aren't editable at the mesh level until you 'freeze' them, removing the procedural aspect. (Unless you are referring to the instanced object.) In Layout, this would be equivalent to "freezing instances", which I believe is possible. You wind up w/a ton of individual objects.

My point is: we already have a nodal approach to getting a slew of (identical) objects in Layout. But this is not the subject of this thread, which is about making node operation/use more 'intuitive'.

IMO we can get "more", but it'll never be all that intuitive. Just better.

devin
01-25-2014, 11:22 AM
I would like to see practical examples included in the documentation of how certain nodes may be utilized. Many of the textbook definitions just confuse me further and tell me nothing.

On that note, I really need to revisit the RH Node videos.

jeric_synergy
01-25-2014, 11:48 AM
I would like to see practical examples included in the documentation of how certain nodes may be utilized. Many of the textbook definitions just confuse me further and tell me nothing.
+1 Absolutely. Indeed. One stinkin' example (minimum) for every node-- is that too much to ask?

And for the REALLY simple ones we could probably do without (add, subtract, mult, etc)

geo_n
01-26-2014, 12:03 AM
In my limited experience in dealing w/procedural arrays, they aren't editable at the mesh level until you 'freeze' them, removing the procedural aspect. (Unless you are referring to the instanced object.) In Layout, this would be equivalent to "freezing instances", which I believe is possible. You wind up w/a ton of individual objects.

My point is: we already have a nodal approach to getting a slew of (identical) objects in Layout. But this is not the subject of this thread, which is about making node operation/use more 'intuitive'.

IMO we can get "more", but it'll never be all that intuitive. Just better.

Yep my reply was not to neglect modelling tools as node tools from the other user post.
They are nodes that have clear functions and directly beneficial. Not confusing to use, when or where to use it not like some of the nodes we have now that users have no idea what its for. I'm sure one can create a node flow that can achieve a bend function that is useful for secondary animation. I think I've seen one already somewhere with super complex flow. But why not just create a bend node to make it useful from the start. Its intuitive and understandable "bend" node :D. Just an example btw to make it atleast simple and direct.
Imho nodes can't be second nature to all users, it just clicks for some and doesn't for others. Watching node tutorials and trial and error and actual testing is the only way.

jwiede
01-26-2014, 04:35 AM
Yep my reply was not to neglect modelling tools as node tools from the other user post.

Fair enough, but Jeric's point is valid about that discussion being orthogonal to the existing thread on how to make nodes more "intuitive". OTOH, this thread seems to be going down the same path as most "idea" threads. Without LW3DG participation as to which (if any) seem interesting or even feasible, nor any push-back to help guide unrealistic asks to be more feasible, the existing thread seems doomed to slide into idle speculation anyway.

jeric_synergy
01-26-2014, 09:35 AM
geo_n, i would say "With Compound Nodes, we do have such a capability-- now who is going to produce these nodes, and how do users find them?"

dwburman
01-26-2014, 05:12 PM
I'll second the idea of having a way to add compound nodes to permanent list like with all the other nodes.

I first got comfortable with nodal workflows by compositing with Fusion (well, the DFX+ that was bundled with LW one upgrade a long time ago) and Shake. It might not be a bad way to ease into using nodes in LW.

I there are several UI bits from compositing apps that'd make working in the node editor more pleasant if not more intuitive. Some of these concepts will probably be harder to implement in LW because some nodes have more data lines than most compositing nodes.

1. The ability to pass-through/disable/ignore a node. Basically, a quick way (keyboard shortcut and/or visible switch) to pass the data through the node so you can compare with and without the node without actually changing the settings or detaching it. It might get a little tricky for nodes with multiple inputs, but I guess you could add a selector to choose which input to pass through.

2. In shake, if you grab a node and shake it, it gets detached from the wires. The wires stay connected. In LightWave, if you delete a node, all the connections go away so you either have to move the connections before deleting the node or reconnect them after.

3. The ability to look at/preview an upstream part of a node flow. In compositing apps, you simply click on a widget/switch thing on a node and you can see the output of that node in the viewer. In LightWave, for surfacing, it'd be nice if we had a preview node with generic UV or planar mapping that you could plug a node into to see what is going on visually. I think an easier way to implement this, however, would be to add a preview switch (or two) onto the node that would cause everything downstream of it to be muted in the results. In terms of texturing, it'd be like plugging the output of the node into Diffuse Shaded, only it wouldn't require the extra attaching and detaching of the wire, disconnecting a wire that's already there, and would work if something is plugged into materials.

4. Auto switching of inputs. Say you want swap the A and B input (or Diffuse and Specular). Now you have to detach one, move the second one and then go back to the original node for the first one and reattach it... or plug the 1st one into another input temporarily until you moved the second one into its place. In compositing apps, you can swap inputs by dragging the 1st wire onto the 2nd wire's input and poof. They are swapped. Of course, sometimes you don't want things to swap and you just want them overridden, so you might want to have the swapping happen after you hover for a while.

5. Also in compositing apps I've worked with, there is a little light on the node telling you when that node is working so you can see how the data is moving through the flow... I don't know if that'd be useful or possible in LW, though.

jeric_synergy
01-26-2014, 07:52 PM
Those are all good ideas, and I'm consumed with envy. However, i'm thinking composting nodes MUST be much more similar to each other than LW/3d nodes, because these items dana mentions imply a great similarity-- for instance, if you delete a LW Gradient node, where are the many many inputs going to get passed thru to?

Re-swapping: I'd make it: Drag the CIRCLE to swap inputs, click the arrow-head to disconnect.

re-pass-thru: I'd have a DEFAULT Pass thru configuration for each node- devs' best guesses, but allow users to re-wire which inputs get passed thru. A hotkey click/hover would graphically depict, via 'wires' which inputs get routed where.

re-previews: there's a number (infinity) of approaches to previews-- multiple ones IN the flow itself, for one ---but I think a single largish preview window that has "channels" that the user assigns at various points would save screen real-estate. That is: the user could drag a 'wire' from numbered slots on the preview window to various points in the flow, and by clicking on the number could see what's happening at different points in the flow. AND, there could be multiple preview windows. (Big ones and little ones would complement each other, functionally.)

robertoortiz
01-28-2014, 07:14 PM
Hey guys, how about if we expand the functionality of controls we already have...

I would propose using the current schematic view at the global NODE controller for a scene.
When you click an schematic item on the scene I would allow to see ALL the nodes associated with that item.

And hell allow to establish connection with nodes from other objects on the scene.

ncr100
07-29-2014, 01:38 AM
May be OT:

Is there any SDK control over how the Node Editor works? E.g. can we decorate nodes with our own GUI elements? It's not hard to conceive of a HELP function: taking the LW Documentation, programmatically stripping out the sections whose headers start with a given Node Name, and making a generic button on the base Node element that displays a scrollable text pop-up containing the text section.

Seems easy to me.

Great ideas in this thread, lots of great ideas.

OlaHaldor
07-29-2014, 06:18 AM
I haven't read the entire thread, so this may have been mentioned before.

I've started playing with Unreal Engine, and in their "simple-to-use" so call Blueprint editor (instead of programming in C++) you basically have a node tree, just as with any surface (or material as they're named in UE4).

There's a simple, yet huge feature I really love there. Click any input or output connection, and drag it where you'd like it to connect to something new. Let go of the button, and a search field and list of available nodes installed shows up. You can search or just browse manually. Hit enter or use the mouse to select the one you're after. The search box/list goes away, and you're left with a noodle going from the initial node to the new node you added. Ready to connect to another node in one of the input values if needed.

Ok... Probably hard to understand a word of that in writing. I recorded it. Just as easy to show what I mean by doing that.

http://youtu.be/A8kCnTPLUDc

OlaHaldor
07-29-2014, 06:23 AM
Seems like I couldn't include two videos in the same post. So I'll continue my post here..


I also wish grouping a bunch of nodes would be possible. This way you can get a better idea of what controls are located where. They also push each other around if you happen to make them collide. Keeps the node tree tidy.

http://youtu.be/5RlumxnTj_o

jeric_synergy
07-29-2014, 07:58 AM
Both of those are cool. THAAAAANNNNKS for making me even more unimpressed w/the LW node editor. ;)

The grouping is esp. cool.

CaptainMarlowe
07-29-2014, 09:35 AM
Grouping exists in substance designer, and it's clearly a considerable help to keep complicated nodes workflows organized. The same feature in our node editor would be most welcome.