PDA

View Full Version : Amazing mograph effect: Bryan Phillips and DPKit



jeric_synergy
09-24-2013, 10:10 AM
Bryphi77 once again amazes with a DPKit mograph effect based on a Houdini tute:

http://www.youtube.com/watch?v=o7qp3qpaUOQ&list=PLkxgHcV1H8PeIPq2JvNBlc3u88or32IXa

It's quite long, but the effect is stunning.

tbagger
09-24-2013, 10:53 AM
The best parts are yet to come... the part I am going to post today is my favorite so far. The amount of functionality that part move and the other transform nodes add is impressive, also the filter and instance nodes. If people were smart they would be throwing money at Denis. We can only hope he continues to develop new tools.

tyrot
09-24-2013, 11:45 AM
Bryan is the master of DP Kits! coolest!!!

AndyUno1
09-24-2013, 12:05 PM
This is fantastic! The more I work with Lightwave, the more I am impressed. Such a great community.

Hail
09-24-2013, 12:15 PM
Good stuff!

jeric_synergy
09-24-2013, 01:32 PM
The best parts are yet to come... the part I am going to post today is my favorite so far. The amount of functionality that part move and the other transform nodes add is impressive, also the filter and instance nodes. If people were smart they would be throwing money at Denis. We can only hope he continues to develop new tools.
Thank god you're exploring what's possible: god bless Denis (and we $hould with our fund$), but the documentation relies on us being as smart as Denis.... and we know how that turns out. (PLUS he's writing in a foreign language, so I'm definitely not complaining*, just saying he's a lot, a LOT, smarter than I am, and some of us need our steaks cut for us. ---And then spoonfed to us.)

So you're doing us a HUGE favor with your tutorials Bryan: not only would I not be able to figure this out, I would never have even known it was even possible.


*multiple times I've inquired about a feature, and Denis has coded it within a couple days. It's freekin' amazing!!!

:thumbsup: :bowdown: :thumbsup: :bowdown: :thumbsup: :bowdown:

tbagger
09-24-2013, 07:00 PM
Thank god you're exploring what's possible: god bless Denis (and we $hould with our fund$), but the documentation relies on us being as smart as Denis.... and we know how that turns out. (PLUS he's writing in a foreign language, so I'm definitely not complaining*, just saying he's a lot, a LOT, smarter than I am, and some of us need our steaks cut for us. ---And then spoonfed to us.)

So you're doing us a HUGE favor with your tutorials Bryan: not only would I not be able to figure this out, I would never have even known it was even possible.


*multiple times I've inquired about a feature, and Denis has coded it within a couple days. It's freekin' amazing!!!

:thumbsup: :bowdown: :thumbsup: :bowdown: :thumbsup: :bowdown:

One of the main reasons I have been doing these tutorials is to show what can be done with these tools, so more people will use them. To overlook these tools is to miss out on a WHOLE lot of functionality. They allow us to do SO much more than what is possible with stock LW.

I do agree that the documentation is like reading braille :D but he can only do so much. Thats another reasons I do these tutorials. I know that Gerardo's tutorial really helped me get a grasp on the filter nodes. If it wasnt for that I would still be trying to decipher that part of the documentation.

I wonder if he has ever considered putting tools/features on kickstarter.com. I know there are a few things that I would be willing to fund. That could be a good way to get him making tools, but I dont know what else he has going on, or if he even has time. I know I would be willing to come off of a few hundred bucks for a good lattice deformer that works like his spline deformer... and I am sure others would as well.

hint hint, Denis :hey:

jeric_synergy
09-24-2013, 07:13 PM
I believe many many LW'ers would kick in money, real actual money not a vague promise, on KickStarter if Denis said he'd code a lattice deformer.

They've certainly whined enough about it. (I've never used any 3d but LW, so I don't know what I'm missing.)(Academically I do, but not practically.)

jeric_synergy
09-24-2013, 08:07 PM
Ok, I'm 16minutes in, and a 'way earlier mention about how 11.x users can use edge tools instead of hypervoxels is still escaping me, although we're nowhere near rendering yet. I trust it will make sense in a bit?

tbagger
09-24-2013, 08:36 PM
Ok, I'm 16minutes in, and a 'way earlier mention about how 11.x users can use edge tools instead of hypervoxels is still escaping me, although we're nowhere near rendering yet. I trust it will make sense in a bit?

I can use render edges as well in 9.6, but in 11.5 they scale with distance properly. In 9.6 they will stay whatever pixel size is set no matter the distance from the object. DWburman has a vid that covers this feature in 11.5.

It is a cleaner set up in 11.5 with being able to render edges properly... but voxels work as well. You just need a bunch of them to make a good line.

jeric_synergy
09-24-2013, 08:54 PM
AH, got it. All Hail 11.n1

jeric_synergy
09-26-2013, 12:00 PM
So, as per the tutorial, I added NODE ITEM MOTION to the 'wires' that get dpmoved, to enable auto-updating so one doesn't have to blip the current frame to get changes to the nodes to update on the UI.

But, that didn't seem to do it. Am I misunderstanding the limits of auto-update, or what?


Thanks.

jeric_synergy
09-26-2013, 01:22 PM
Have to knock off for the day, but having some issues w/getting scaling to work -- SCENE ATTACHED. 117332

Plus: at a certain point, the viewports started showing radically different things, SEE JPEG.

When I SAVEd PACKAGE, the viewports reverted to their correct (?) views.
117331

tbagger
09-26-2013, 05:28 PM
So, as per the tutorial, I added NODE ITEM MOTION to the 'wires' that get dpmoved, to enable auto-updating so one doesn't have to blip the current frame to get changes to the nodes to update on the UI.

But, that didn't seem to do it. Am I misunderstanding the limits of auto-update, or what?


Thanks.

The auto update only works for things that are controlled by other objects. So, if you have a null that is controlling part of the animation it allows it to update as you are transforming the null. There are some circumstances where it still wont update as you change values in the node editor itself. That trick is only relevant to things that are reliant on the motion of another object in the scene. Not changes made in the node editor.

tbagger
09-26-2013, 05:34 PM
Have to knock off for the day, but having some issues w/getting scaling to work -- SCENE ATTACHED. 117332

Plus: at a certain point, the viewports started showing radically different things, SEE JPEG.

When I SAVEd PACKAGE, the viewports reverted to their correct (?) views.


117331

It is most likely related to going back and forth to modeler... it messes up the point index in some way. Either kill the hub, or only use one app at a time and see if you still run into problems, or it could be related to something that has changed in 11.5.999

prometheus
09-26-2013, 07:31 PM
Cool stuff going on there Bryan...
Darn it that I wont be able to engage in studying this until a couple of weeks I think...to sad since you seem to have gathered some very nice mographs stuff not seen before with the help of
Dennis tools...
Thanks Bryan, and very much thanks to Dennis, and we all need to find ways to support him indeed.

The stuff your working on now might work nicely for Hud sci-fi displays I guess? A thought crossed my mind if you could pull off a type of H.U.D display similar to what was shown in the Prometheus movie?
and on the cryo capsules and also on some other Hudīs in there, those showcase smooth bezier type of connections between data displays, that also expands and contracts depending on motion.
I recall the Vfx team had to resort to a special script done for after effects for this purpose and couldnīt even use standard tricks in after effects.

Michael

jeric_synergy
09-26-2013, 07:42 PM
It is most likely related to going back and forth to modeler... it messes up the point index in some way. Either kill the hub, or only use one app at a time and see if you still run into problems, or it could be related to something that has changed in 11.5.999
I've added "KILL THE HUB" to the 0:00 entry in my timeline transcript....

It's probably some weird vuport updating issue.

tbagger
09-26-2013, 08:07 PM
Cool stuff going on there Bryan...
Darn it that I wont be able to engage in studying this until a couple of weeks I think...to sad since you seem to have gathered some very nice mographs stuff not seen before with the help of
Dennis tools...
Thanks Bryan, and very much thanks to Dennis, and we all need to find ways to support him indeed.

The stuff your working on now might work nicely for Hud sci-fi displays I guess? A thought crossed my mind if you could pull off a type of H.U.D display similar to what was shown in the Prometheus movie?
and on the cryo capsules and also on some other Hudīs in there, those showcase smooth bezier type of connections between data displays, that also expands and contracts depending on motion.
I recall the Vfx team had to resort to a special script done for after effects for this purpose and couldnīt even use standard tricks in after effects.

Michael

I had considered using curved lines as opposed to straight, but decided to go with straight. You could use curved lines just as easily. Pretty much anything that can be rail extruded will work. Not sure exactly what you are going for though... do you have a link to image. I did see Prometheus, but dont remember what the huds looked like, or how they animated. Drawing or scaling curves would not be a problem, but modifying the curve itself could be more complex, but not impossible... Depends on what you are going for.

tbagger
09-26-2013, 08:36 PM
I've added "KILL THE HUB" to the 0:00 entry in my timeline transcript....

It's probably some weird vuport updating issue.

I had a look at your scene... The reason the scale was not working is because you have to drive the position of the turbulence with the same vector that goes into the move. It worked fine after doing that.

I did get a bunch of errors on load about missing "input" something or other, but it still seemed to work as expected.

jeric_synergy
09-26-2013, 09:32 PM
Thanks for checking that, Bryan. Sweet to see it start working.

But....why??? Doesn't the Turbulence texture extend throughout 3D space? I can see how connecting the vector would make a difference, but I'm not understanding why it apparently doesn't work at all if it's not connected: in my mind, doesn't that just move the turbulence around??

Is there some falloff inherent in Turbulence??



(I certainly hope someone besides myself is helped via my egregious display of ignorance.....)

tbagger
09-26-2013, 09:50 PM
Thanks for checking that, Bryan. Sweet to see it start working.

But....why??? Doesn't the Turbulence texture extend throughout 3D space? I can see how connecting the vector would make a difference, but I'm not understanding why it apparently doesn't work at all if it's not connected: in my mind, doesn't that just move the turbulence around??

Is there some falloff inherent in Turbulence??

What you have to remember is that all those parts are in one spot... at the origin. The turbulence texture still assumes thats where they are. It has no idea of the motion after part move takes place. If you had all parts pre scattered in modeler and then used the turbulence like you were it would work how you expected. What you are doing by plunging the vector into the position of turbulence is telling it that this is where the parts currently are. Otherwise it would assume they were still in the position as they were loaded from modeler. You will notice that if you change the turbulence parameters the way you had it all parts will react the same because it is still assuming they are all right on top of one another at the origin.

To see what I mean... make a few boxes in modeler that are not right on top of one another, and then scale them in same way like you were... you will see that it works like you would expect.

jeric_synergy
09-26-2013, 11:14 PM
AHHHHHhhhhhh, got it, that was a very good explanation: the parts 'real self' are all overlapping in the same spot, so any effect the texture would have on them would be identical.

Man, it's a very different mindset one has to maintain to even comprehend what is going on.

There's some other process that's very analogous to this inside LW, but I can't remember off the top of my head what it is: the upshot is that even though something looks like it's one place, as far as the math goes is in its original place, very much the situation here.

Thanks man!

tbagger
09-27-2013, 09:23 AM
AHHHHHhhhhhh, got it, that was a very good explanation: the parts 'real self' are all overlapping in the same spot, so any effect the texture would have on them would be identical.

Man, it's a very different mindset one has to maintain to even comprehend what is going on.

There's some other process that's very analogous to this inside LW, but I can't remember off the top of my head what it is: the upshot is that even though something looks like it's one place, as far as the math goes is in its original place, very much the situation here.

Thanks man!

Yes, it is limitation do to not having proper modifier stack. All nodal displacements take place on original point positions and use original point normals. This is huge limitation of LW. Many of DPs nodes do try to get around this by having the delta input, but its not perfect solution. In this example we are able to work around it. This is actually one of the key concepts that make part move as useful as it is... being able to use the procedurals like that.

It would be nice if nodal displacements could be stacked, and previous displacement in the stack would output new normal and new positions. This is how Houdini does it... but in Houdini displacements are a node that can be strung along, but a stack could work as well. Similar to how we have regular texture displacement that can be applied before the nodal displacement.

One solution would be to have nodal displacements be displacement plug that could be added and stacked like the other displacement plugs.

phillydee
09-27-2013, 10:24 AM
This is great, thanks for the effort!! (posting to subscribe to the thread)

Cheers

jeric_synergy
09-27-2013, 04:19 PM
Today's "Takeaway Idea" from the tutorial:

There's all sorts of gems for nodal idiots like me in Bryan's work, stuff he doesn't even think worth mentioning. For me today, it was re: the LOGIC node.

Specifically: despite the name, the OUTPUT of the logic node can be any value, not just zero or 1.

So, if you need a value to just pop on and off, you just enter that value into the IF TRUE/IF FALSE outputs of the Logic node.

Sweet.

raw-m
09-29-2013, 12:23 PM
Really enjoying your tutorials, Bryan, another thanks for putting the time in!

Can I hijack to ask a quick question? A pet hate of mine in LW is the way that Color Space works when you you're plugging vector values in to color input nodes - if working sRGB you'll get a value shift to sRGB, giving unpredictable results with your vector values. Got a work around (other than working in Linear CS!)?

dwburman
09-29-2013, 01:03 PM
db&w tools has a color space node.
http://www.db-w.com/products/dbwtools/docs
I haven't tried it in production, but in theory it should work to correct the issue. :)

jeric_synergy
09-29-2013, 02:34 PM
OK, I figured this out, 'cuz it was simple stupidity, but perhaps others will enjoy the process/be amused. Also, found some fun stuff, and some mysterious stuff unrelated to Bryan's tutorial:

A) (Just noting that changing the object in LWM, in any way and even with the HUB off, makes Layout display things nutzo --infinite lines and whatnot-- and quickly crash. In this case all I did was add a default Surface (i.e. only a new Surface), which was enough to make Layout roll over and die.)

B) Still working on BP's tutorial:
I'm confused, so I'll just relate my confusion, and somebody correct me (I know my reasoning is wrong, I'm just not sure where):

To place the rectangles, at 37:23 Bryan adds an ADD node and enters 64001 into the B field as a point index offset. This will place the rectangles at the LAST point in every wire.

To me this implies that when numbering points along the wires, LWM makes the first point in the 32 wires points 1 thru 32 (or possibly 0-31). Then the SECOND POINT in each wire is indexed as 33-64. (?? ahhh, the joys of fence-post errors...) etc, until the last wire will be 32x(number of points in wire+1) (where'd the "+1" come from?).

MY wires have 201 points (his have 200) so the last point in my first wire should be 201x32 = 6,432. +1 to get to 6,433. (ED: hmmm, may have spotted the problem....)....

OKAY, found the problem: I was calculating the total number of points, and of course what I want for an offset is TOTAL-32. Worked perfectly after that.

C) OK, while I was investigating this I put an envelope on the B field of the index offset (the ADD node), and it is cool to slide the rectangles along the wires by incrementing the offset.

D) prepping for render, I decided to add a green outline to the rectangles. Surprisingly, TO ME, "Silhouette Edges" did not do the trick, indeed nothing happened at all (the rectangle Surface is double-sided). Unshared Edges worked fine, but now I'm worried about Silhouette Edges.

E) Attempting to Package the scene caused Layout to crash. (ED: twice! Zipped the whole directory, use scene 007)

Moving on to Part 2 soon...

tbagger
09-29-2013, 05:25 PM
OK, I figured this out, 'cuz it was simple stupidity, but perhaps others will enjoy the process/be amused. Also, found some fun stuff, and some mysterious stuff unrelated to Bryan's tutorial:

A) (Just noting that changing the object in LWM, in any way and even with the HUB off, makes Layout display things nutzo --infinite lines and whatnot-- and quickly crash. In this case all I did was add a default Surface (i.e. only a new Surface), which was enough to make Layout roll over and die.)

B) Still working on BP's tutorial:
I'm confused, so I'll just relate my confusion, and somebody correct me (I know my reasoning is wrong, I'm just not sure where):

To place the rectangles, at 37:23 Bryan adds an ADD node and enters 64001 into the B field as a point index offset. This will place the rectangles at the LAST point in every wire.

To me this implies that when numbering points along the wires, LWM makes the first point in the 32 wires points 1 thru 32 (or possibly 0-31). Then the SECOND POINT in each wire is indexed as 33-64. (?? ahhh, the joys of fence-post errors...) etc, until the last wire will be 32x(number of points in wire+1) (where'd the "+1" come from?).

MY wires have 201 points (his have 200) so the last point in my first wire should be 201x32 = 6,432. +1 to get to 6,433. (ED: hmmm, may have spotted the problem....)....

OKAY, found the problem: I was calculating the total number of points, and of course what I want for an offset is TOTAL-32. Worked perfectly after that.

C) OK, while I was investigating this I put an envelope on the B field of the index offset (the ADD node), and it is cool to slide the rectangles along the wires by incrementing the offset.

D) prepping for render, I decided to add a green outline to the rectangles. Surprisingly, TO ME, "Silhouette Edges" did not do the trick, indeed nothing happened at all (the rectangle Surface is double-sided). Unshared Edges worked fine, but now I'm worried about Silhouette Edges.

E) Attempting to Package the scene caused Layout to crash. (ED: twice! Zipped the whole directory, use scene 007)

Moving on to Part 2 soon...

I knew I should have spent more time explaining the part about adding 64001.

The plus one is because the part index starts at 0 but the point index starts at 1. The point ordering is the way it is because all lines were rail extruded at the same time. If you had made 1 line first and cloned it, it would give completely different result.

What it comes down to is the last points in each line are in order because of the way they were created. So the second to last points are also in order. If you wanted to target the boxes to the lines path instead of the camera what you could do is use another point info but add 64001-32 instead, then you could align the boxes to the point before the one its on.

I dont get a crash with package scene, I dont get many crashes at all with these tools. That is main reason I am still using 9.6. Until Denis stops supporting 9.6 I will continue to use it. As far as the jumbled lines you should be able to revert scene to last saved to fix that error. You can even save it when they are jumbled, and it will still revert to correct positions. The best way to not have this happen is to close layout, edit the object in modeler and save, then reopen layout and load the scene again, and it will load with the updated model.

I will have a look at your file in a bit, cant get to it right this moment, but I will, unless you already worked it out?

tbagger
09-29-2013, 05:26 PM
Really enjoying your tutorials, Bryan, another thanks for putting the time in!

Can I hijack to ask a quick question? A pet hate of mine in LW is the way that Color Space works when you you're plugging vector values in to color input nodes - if working sRGB you'll get a value shift to sRGB, giving unpredictable results with your vector values. Got a work around (other than working in Linear CS!)?

Yes, DW is correct, and XswampyX has a vid on his youtube channel on how to do it.

jeric_synergy
09-29-2013, 08:34 PM
I will have a look at your file in a bit, cant get to it right this moment, but I will, unless you already worked it out?
No need: it works just like yours now. THANKS!

On a tutorial level, it'd be educational to explicitly point out why you might choose one modeling approach over another: in hindsight, I can see that this method allows a simpler (but how much simpler?) network when you want to distribute other parts, versus cloning -- it's possible, but I'm sure it takes more arithmetic.

And the difference in point/part index start is one of those classic "gotchas!" that make things so much fun..... grrrr......

Working thru this made me realize I couldn't remember the name of the math node that throws away the decimal portion of a division. *sigh*

SAVE PACKAGE crashing is most likely a 11.5+ bug. I suppose I should send it in...

dwburman
09-29-2013, 10:07 PM
You might be thinking of Ciel (for ceiling?), but that might push the value up to the next integer/whole number if there's a decimal rather than just dropping the decimal or rounding up/down.

I'm not a math guy either.

tbagger
09-29-2013, 10:49 PM
You might be thinking of Ciel (for ceiling?), but that might push the value up to the next integer/whole number if there's a decimal rather than just dropping the decimal or rounding up/down.

I'm not a math guy either.


Don't forget about its counterpart "floor" rounds down...

jeric_synergy
09-30-2013, 12:57 AM
IIRC, "modulo" in programmer-speak as in "25 modulo 8=3".... or is it "1"??

There IS a 'mod' function node which is modulo, but I think I have it backwards.... but the examples in the LW10 dox seem like a spectacularly bad set of data to try to explain this function with. Take a look, it's awful. Geeze.

Anyway, I think I'm looking for the opposite of modulo. RH?

jwiede
09-30-2013, 07:29 AM
IIRC, "modulo" in programmer-speak as in "25 modulo 8=3".... or is it "1"??

Modulo operation yields the "remainder" of integer division. 25 modulo 8 = 1. What you want is just the result of the division with floor() applied, floor( 25 / 8 ) = 3.

jeric_synergy
09-30-2013, 10:41 AM
"Floor" it is!

Random question: if one wanted the modulo of a division, but always wanted it 'normalized' to an integer value between 0 and 10, what would that look like?

stevenpalomino
09-30-2013, 12:19 PM
This is really great. On top of that... now that we can compound nodes they can be exported and shared for common motion graphics things. Is there a place to upload compounds to share for these types of things? Reminds me of XSIs ICE.

jeric_synergy
09-30-2013, 01:00 PM
(I typed most of this before I started doubting "Wait, CAN we save subnetworks?" --away from LW right now--, but I think we can, and I think I've done it before. If not, consider it a feature request....)

We ALREADY could save sections of networks, but I've never read/heard of anybody saving a library of networks to avoid repetitive labor-- I'm sure some people MUST have been doing so, but just haven't seen it brought forward as a technique.

My point here is : we didn't have to wait for Compound Nodes to have prefab functions-- saving subnetworks would've been much of the same effect. Not as slick, for sure, but better than nothing.

For instance, Bryan seems to load manually, repeatedly, the same 5 nodes to calculate offset. He could easily create and Save a subnetwork, with or without connections, that would alleviate all that mousing in the Node Editor menu, which seems to take quite a bit of time itself.

(Gahd I hope I'm not misremembering that.....)

tbagger
09-30-2013, 01:27 PM
(I typed most of this before I started doubting "Wait, CAN we save subnetworks?" --away from LW right now--, but I think we can, and I think I've done it before. If not, consider it a feature request....)

We ALREADY could save sections of networks, but I've never read/heard of anybody saving a library of networks to avoid repetitive labor-- I'm sure some people MUST have been doing so, but just haven't seen it brought forward as a technique.

My point here is : we didn't have to wait for Compound Nodes to have prefab functions-- saving subnetworks would've been much of the same effect. Not as slick, for sure, but better than nothing.

For instance, Bryan seems to load manually, repeatedly, the same 5 nodes to calculate offset. He could easily create and Save a subnetwork, with or without connections, that would alleviate all that mousing in the Node Editor menu, which seems to take quite a bit of time itself.

(Gahd I hope I'm not misremembering that.....)

:hey:

http://www.youtube.com/watch?feature=player_detailpage&v=Dy2d1-7iV2s#t=1554

:D

I could do this more, but I figure repetition helps others learn easier than just having me load premade networks all the time. Also, if someone is just watching one vid out of the bunch it would be harder to follow along.

jeric_synergy
09-30-2013, 02:19 PM
I could do this more, but I figure repetition helps others learn easier than just having me load premade networks all the time. Also, if someone is just watching one vid out of the bunch it would be harder to follow along.
Pedagogically, for a tutorial you're absolutely right. As my friend says, "Repetition makes the master."

But it just seemed a bit odd that I've never even heard people SUGGEST building up a library of common subnetworks so as to lessen both ERROR and labor. Maybe I don't get around enough. I stick mine in the premade folder I duplicate to make new project folders-- while this results in a lot of duplication, it makes it easier to find when I want one.

And now with Compound Nodes it should be even easier!
+++
Now, just in case you weren't sick of me, time to start Part 2!

cue: flood of my [email protected] questions


I'll attach my time transcript of Part1 for anybody who might find it useful, at the least to add to for their own use. It only goes up to 39:01 for some reason, but that's the meat of the tutorial-- it's quite handy to have a transcript, no matter how opaque/sketchy, to navigate the video: gahd knows I had to watch bits multiple times and it's pretty nice to jump directly to the part you need:

117387

stevenpalomino
09-30-2013, 02:40 PM
Yeah. Saving nodes was still possible. I'm just saying it's easier now with just one node instead of a network.

jeric_synergy
09-30-2013, 02:42 PM
Maybe everybody was already doing it, but I never heard 'em talking about it.

stevenpalomino
09-30-2013, 02:45 PM
Maybe everybody was already doing it, but I never heard 'em talking about it.

Yeah. Me neither.

tbagger
09-30-2013, 02:50 PM
Part 2 is pretty easy... you should be able to work through it much easier... especially after doing part1.

The compound node is huge step in right direction. I will download the trial if they have one to see what its all about. I do have a question in regards to the compound node...

Is the only way to control an input you create with another node? Or does it create an envelopeable channel that can be edited directly in the node properties panel? Like many of the other nodes.

For example, like the make vector node. You can plug into the input to set the value, or you can open up its properties and envelope the value direc, and if there is a node pluged into that value the envolope becomes uneditable.

jeric_synergy
09-30-2013, 03:02 PM
I suspect you could hack your way around that (Dana?) by including inside the compound node a {scalar node+an add node} that would serve the envelope functionality, but not have any effect on the input with which it would always be combined.

tbagger
09-30-2013, 03:09 PM
I suspect you could hack your way around that (Dana?) by including inside the compound node a {scalar node+an add node} that would serve the envelope functionality, but not have any effect on the input with which it would always be combined.

What I am getting at is... it would be real nice if all the inputs you create also created an envolopable counterpart that could be envoloped without needing to drive it with another node. So all inputs/ parameters could be controled without need to plug in nodes. Its kinda hard to explain what I mean.

jeric_synergy
09-30-2013, 03:12 PM
Understood. Well, it's still pre-release: maybe that is already included, or maybe there's time to add it. --anyway, I should quit speculating.

tbagger
09-30-2013, 03:22 PM
Understood. Well, it's still pre-release: maybe that is already included, or maybe there's time to add it. --anyway, I should quit speculating.

Its not really a huge deal, but it would be a way to make consolidated tools that did not rely on other nodes, just edit parameters directly on the group node itself. This is how houdini does it... could be usefull.

It will still be extremely usefull without... and it won't lose any functiinality by not having this, but will always need to use other nodes to control parameters.

pinkmouse
09-30-2013, 03:51 PM
Brian, I have a completely self contained compound node here (http://forums.newtek.com/showthread.php?136524-house-building&p=1337603&viewfull=1#post1337603). Is that what you were talking about?

Oh, and great tutorial BTW, watched through the first couple of parts, and will give it a go sometime soon. :)

jeric_synergy
09-30-2013, 04:03 PM
I just watched Dana's intro to the Compound node and AFAICT, no, BUT, my scheme (jpg) would seem to offer pretty much a near identical functionality. It adds one more doubleclick, because you have to doubleclick on the Compound node itself to get to the interior node, which you have to do anyway to adjust node envelopes. Of course, the interior "default nodes" should be set to values that have no effect on the input values if the the input nodes ARE connected.
117391

tbagger
09-30-2013, 04:12 PM
Brian, I have a completely self contained compound node here (http://forums.newtek.com/showthread.php?136524-house-building&p=1337603&viewfull=1#post1337603). Is that what you were talking about?

Oh, and great tutorial BTW, watched through the first couple of parts, and will give it a go sometime soon. :)

Nice animation you have there... I had seen that a few weeks ago, and thought it was nice effect.
What I am talking about with the compound node is... do you need to control the inputs that you create with other nodes, or does the compound node also create envolopable channels for each input that is created? So if I create one scalar and two color inputs is there a way for me to envolope or change the value of those channels without having to plug a node into it.

tbagger
09-30-2013, 04:16 PM
But in that set up jeric you need to go into the compoud itself in order to edit it.

pinkmouse
09-30-2013, 04:32 PM
Nice animation you have there... I had seen that a few weeks ago, and thought it was nice effect.

Ta! :)



What I am talking about with the compound node is... do you need to control the inputs that you create with other nodes, or does the compound node also create envolopable channels for each input that is created? So if I create one scalar and two color inputs is there a way for me to envolope or change the value of those channels without having to plug a node into it.

Just trying to clarify with an example: You would have a self contained compound node that, say, moves an object up 1m and then spins it around clockwise. You would like connections to the outside that if connected, overide the default behaviour, so it could instead move downward and spin anticlockwise. Is that what you're thinking of?

I'll have a play tomorrow, I'm working on a nodal motion scene at the moment. Still lots to do, but apart from the keyframed motion of the UFO and camera, everything is driven nodally.


http://www.youtube.com/watch?v=kmV-sCnN7X0&feature=youtu.be

jeric_synergy
09-30-2013, 04:40 PM
But in that set up jeric you need to go into the compoud itself in order to edit it.
Yeah, but even with a regular node, you have to open it to Envelope it. So, it's one more doubleclick. Regretable, but there it is.

It'd be slick fer sure, but I don't think it's in the cards this time around.

Since inputs are only, what, 5 different types, it should be doable-- man, I can see the UI in my mind's eye right now. It'd just be a simple panel w/all the inputs, plus Envelope buttons. Any click in the Input side of the Compound node icon would open this panel-- one would have numerical access to all the inputs simultaneously.

EDIT: I misspoke: if one is using the side panel, one doesn't have to explicitly open the node, IIRC.

tbagger
09-30-2013, 05:03 PM
Yeah, but even with a regular node, you have to open it to Envelope it. So, it's one more doubleclick. Regretable, but there it is.

It'd be slick fer sure, but I don't think it's in the cards this time around.

Since inputs are only, what, 5 different types, it should be doable-- man, I can see the UI in my mind's eye right now. It'd just be a simple panel w/all the inputs, plus Envelope buttons. Any click in the Input side of the Compound node icon would open this panel-- one would have numerical access to all the inputs simultaneously.

EDIT: I misspoke: if one is using the side panel, one doesn't have to explicitly open the node, IIRC.

This is what I am getting at... Like I said its not that big a deal, and current method doesn't cause you to lose any functionality, but I think this would be good way to do it.

http://www.youtube.com/watch?v=eQFpz0bOKLE

dwburman
09-30-2013, 05:04 PM
That's not a bad feature request.... a way to add input channels without having to have nodes connected.... if a node was connected then it could override the internal channel you set.

pinkmouse
09-30-2013, 05:13 PM
Ah, now I see where you're coming from. Not at the moment unfortunately:

117392

I'd definitely second it as a feature request. ;)

jeric_synergy
09-30-2013, 06:26 PM
Got up to 8:05 in the Part Two, when I encountered the same difficulties a commenter on YouTube encountered, but I can attach a Packaged Scene (and LW still locking up after a Package, but successfully exported the directories):
117393

Please test this Bryan and see whether it's A) user incompetence, or B) if something in 11.6 has been changed in DPKit compatibility.

tbagger
09-30-2013, 06:30 PM
Got up to 8:05 in the Part Two, when I encountered the same difficulties a commenter on YouTube encountered, but I can attach a Packaged Scene (and LW still locking up after a Package, but successfully exported the directories):
117393

Please test this Bryan and see whether it's A) user incompetence, or B) if something in 11.6 has been changed in DPKit compatibility.
Ill have a look at it now...

tbagger
09-30-2013, 06:34 PM
Are you getting an error in its current state? If so, something has changed... works as expected here.

tbagger
09-30-2013, 06:41 PM
Yeah, there must be an issue. It could also be an error in the 11 version of dpkit. Can you try it in 11.5?

jeric_synergy
09-30-2013, 07:25 PM
The version of DPKit that I have installed w/my copy of 11.5 has an substantially different RANDOM node that currently.... should I change to the current DPKit?

(Well, I will whatever...)
clicka clicka clickaclicka...

OK, here's LW11.5, with the latest DPKit (dl'd ten minutes ago)

It looks like it's displacing the POINTS separate from than the PARTS....
117397

tbagger
09-30-2013, 07:38 PM
The version of DPKit that I have installed w/my copy of 11.5 has an substantially different RANDOM node that currently.... should I change to the current DPKit?

(Well, I will whatever...)
Use newest version of kit... I am just trying to figure out if it is issue with 11.6 or issue with dpkit for 11.5. I hope it is issue with dpkit... cause it will actually get fixed in timely manner. If it is problem or change in 11.6 it may be out of his control.

jeric_synergy
09-30-2013, 07:43 PM
Done, see above.

tbagger
09-30-2013, 07:51 PM
Done, see above.

Ill send test scene over to Denis... and see what he can do.

jeric_synergy
09-30-2013, 08:18 PM
I note that the 11.6 Move node is substantially different, at the very least that it has Sample ball. So it looks like NewTek has had its hand on that node, and maybe it got borked in the process.

++++++
While we're waiting for that: Would it have been a viable tactic to get the vertical spacing less overlappy to use a narrower Random min/max and a multiplier?

I note that the "data probe" thingy isn't reporting the proper total number of Parts on the output, although the node label is correctly reporting it.

tbagger
09-30-2013, 08:27 PM
I note that the 11.6 Move node is substantially different, at the very least that it has Sample ball. So it looks like NewTek has had its hand on that node, and maybe it got borked in the process.

I sent it over to him...

Are you saying that 11.6 has native move node? Are you using the native one or Denis's?

tbagger
09-30-2013, 08:32 PM
I mention in a later part that you get much better distribution if you multiply by a number instead of add before you go into the seed. So, add 1 or whatever, then multiply by a number. This gives much better result.

jeric_synergy
09-30-2013, 08:42 PM
I thought I was using the native one (away from machine).

Does the seed make a difference in the distribution, uhhhh, scale? I understand that it would be different set of random numbers, but I'm surprised if it would make affect how it might bunch up.

tbagger
09-30-2013, 08:54 PM
I thought I was using the native one (away from machine).

Does the seed make difference in the distribution, uhhhh, scale? I understand that it would be different set of random numbers, but I'm surprised if it would make affect how it might bunch up.

Yes, it makes a difference. I didn't think it should either, but when using add and having the seed be consecutive numbers it seems that 2 of the parts are always very close to the same number, but when using the multiply this does not happen.

jeric_synergy
09-30-2013, 09:06 PM
I was wrong: the Move node is DPKit, but as I said, now it has a sample ball.

Sometimes I'll put in an envelope on something to examine what gradual increases do to whatever.

tbagger
10-01-2013, 04:09 AM
I was wrong: the Move node is DPKit, but as I said, now it has a sample ball.

Sometimes I'll put in an envelope on something to examine what gradual increases do to whatever.

This error has been fixed in new version of dpkit. Was caused by threaded mesh eval error. Thanks a bunch Denis...

jeric_synergy
10-01-2013, 08:32 AM
This error has been fixed in new version of dpkit. Was caused by threaded mesh eval error. Thanks a bunch Denis...
SWEET! Thank You Denis!!!! (runs off to dl...)



ED: at first glance, looks like it's working! :thumbsup:

tbagger
10-01-2013, 09:25 AM
SWEET! Thank You Denis!!!! (runs off to dl...)



ED: at first glance, looks like it's working! :thumbsup:
May have also fixed your package scene issue as well...

jeric_synergy
10-01-2013, 11:19 AM
Jumped into LWM, decided to make the "distributor" points into 1 pt polys for easier visibility, as expected this borked Layout, Reloaded Scene, BANG, Layout dead.

OK on reLaunch, but obviously something is corrupting enough data, between LOAD SCENEs, to crash Layout. Or perhaps this means that the initialization routines of LOAD SCENE are inadequate to return the internal data structures to a pristine state. (I'm assuming that a 'virtual NEW SCENE' is executed before attempting to LOAD SCENE. It'd be nice if a dev could pop in and say if this is so.)

tbagger
10-01-2013, 11:54 AM
Jumped into LWM, decided to make the "distributor" points into 1 pt polys for easier visibility, as expected this borked Layout, Reloaded Scene, BANG, Layout dead.

OK on reLaunch, but obviously something is corrupting enough data, between LOAD SCENEs, to crash Layout. Or perhaps this means that the initialization routines of LOAD SCENE are inadequate to return the internal data structures to a pristine state. (I'm assuming that a 'virtual NEW SCENE' is executed before attempting to LOAD SCENE. It'd be nice if a dev could pop in and say if this is so.)

You are right that something is not being put back at its original state when scene is cleared. The best way to work around this is to just close layout before you make edits in modeler. Then reload the scene. I don't get crash like you, but I get point reordering if layout is not a fresh open. Layout is indeed remebering something that it shouldn't.

jeric_synergy
10-01-2013, 12:23 PM
Oh, I can always work around it, BUT WE SHOULDN'T NEED TO, of course. :devil:

To me, this has got to be "low hanging fruit" -- LW already has initialization routines. How hard is it to call them AGAIN? :twak:

+++

BTW, afaict, the "add Node Item Motion" trick doesn't work in 11.6. I've added it to both the deformed items and the controller items (nulls) and it doesn't seem to auto-update.

That is, I can alter a (eg) rotation, set the key, and deformation doesn't change until I blip the frame count. :(

+++
As our UKers would say, it's bloody brilliant that Denis fixed that threading problem so promptly! Thanks again, Denis!

tbagger
10-01-2013, 01:09 PM
Are you using native node motion or Denis's node item motion? I dont know that it will make a difference, but it may. It is only needed on the deforming mesh, not the control. I do recall someone mentioning another way to get things to update, but cant remember what it is.

jeric_synergy
10-01-2013, 01:38 PM
Both Node Item Motion and (native?) Nodal Motion are failing at auto updating (in 11.6, for this Scene).

Can anybody confirm?


++++

This has been confusing me for years:

In the menu you see:

Node Item Motion

Once you've applied it, it is displayed as:


Item Motion Node Editor


Sure, it's a tiny teensy thing, but it's Yet Another Speed Bump. Another little grain of sand of confusion. I'd hear in video tutorials "Node Item Motion", but in the modifier display (not the menu) I'd SEE "Item Motion Node Editor". I wish they were both the same.

jeric_synergy
10-02-2013, 02:08 PM
Bryan;
you seem to be very concerned about applying a different Seed to all the Random nodes. Is it that noticeable if you do not?

tbagger
10-02-2013, 02:16 PM
Bryan;
you seem to be very concerned about applying a different Seed to all the Random nodes. Is it that noticeable if you do not?

Depends on the situation. If you dont all parts will always be using the same random number for every time the random node is used. It may not be to big a deal in some situation, but others it will. Say you wanted to move and scale parts randomly if you use the same seed it will use the same number for the move and the scale. For example in part 2... if you used the same seed for both rotation functions the speed of the spinning would be tied to the size of the circle segments. The biggest segments would spin the fastest, and smallest the slowest. That may not be a bad effect either though ;)

jeric_synergy
10-02-2013, 03:14 PM
If you could invert that, I think it would make good visual 'sense' -- as if they were gaining weight as they grew.

Julez4001
01-21-2014, 11:27 PM
Very Impressive from you Bryan and Jeric, thanks for keeping up with him.