MoI discussion forum
MoI discussion forum

Full Version: Node Wish List

Show messages:  1-15  …  116-135  136-155  156-175  176-195  196-215  216-235  236-255  …  416-425

From: Frenchy Pilou (PILOU)
17 Apr 2020   [#176]
That's magical thing!
From: James (JFH)
29 Apr 2020   [#177]
BranchStructure node

The proposed node would generate venation patterns from seed points branching out to leaf points (attractors) using the "space colonization" algorithm.

Daniel Shiffman has done a coding challenge that serves as a good introduction to the method:
https://www.youtube.com/watch?v=kKT0v3qhIQY

Another good resource is Jason Webb's "Modeling organic branching structures with the space colonization algorithm and JavaScript" :
https://medium.com/@jason.webb/space-colonization-algorithm-in-javascript-6f683b743dc5

He demonstrates the difference between "Open versus closed venation" which is an option I have include in the info panel. The other option I have included is 2D vs 3D. 3D would work similarly, however with branches attracted to leaf points in 3d space, making possible the generation of natural forms such as tree, coral etc. This may be something that is left for later development.





James
https://www.instagram.com/nodeology/

Image Attachments:
branchStructures.jpg 


From: James (JFH)
24 May 2020   [#178]
Hi All,


I have already alluded to this here:http://moi3d.com/forum/index.php?webtag=MOI&msg=9810.10 but in terms of implementation, rather than the generalised approach it may be better to have a purpose built "Reaction Diffusion" node.

React-Diffusion node

I am proposing a node that operates analogous to online app by @pmneila https://pmneila.github.io/jsexp/grayscott/

For our purposes the output would only need to be black and white, which hopefully would reduce the complexity of the coding. The proposed node would, similarly to Wayne's imgSampler convert greyscale to heightmaps, however instead of importing an image, it would be generated via an algorithm (grayscott) within the node.

Since this algorithms produce an evolving animated image there is a need to pause evolution at a desirable point so I have included an input for 1 to play & 0 to pause. There may be a more elegant solution of the inclusion of menu option in the info panel, or perhaps ideally a button on the node, itself.

I'm not sure if this is even possible without the addition of external js libraries. (I note that in the code for the online app link above that it imports three.js)
James
https://www.instagram.com/nodeology/

Image Attachments:
react-Diffusion.jpg 


From: bemfarmer
24 May 2020   [#179] In reply to [#178]
Viewing page source yields: https://github.com/pmneila/jsexp

- Brian
From: wayne hill (WAYNEHILL5202)
13 Jun 2020   [#180]
Playing with a new node conversion of Max Smirnov's cloth script that he converted from the ThreeJS engine.

http://moi.maxsm.net/files/scripts/

My first attempt in animation on the Node Editor. Much to learn from Max's script.






From: Frenchy Pilou (PILOU)
13 Jun 2020   [#181] In reply to [#180]
Does it works now with any volumes and not only a sphere ?
From: wayne hill (WAYNEHILL5202)
13 Jun 2020   [#182] In reply to [#181]
Only spheres are in the original code. Will be testing other geometry types.
From: wayne hill (WAYNEHILL5202)
13 Jun 2020   [#183]
Added node form and javascript code to message:

http://moi3d.com/forum/index.php?webtag=MOI&msg=9581.180
From: wayne hill (WAYNEHILL5202)
13 Jun 2020   [#184]
Finally releasing my ImgRGB first mentioned here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=9581.165

The rgbquant.js library limits the style layers on the image.

ImgRGB send an RGB connection to OutRGB to reassemble the image to the style layers.
The OutRGB node only works with the ImgRGB node.

Thank you to James for the on going testing of the nodes I keep sending him... :-)

Wayne

Message 9581.185 was deleted


From: James (JFH)
14 Jun 2020   [#186] In reply to [#182]
Wayne,

The cloth node is fantastic.......and mesmerising to watch.

"Only spheres are in the original code.
Will be testing other geometry types".


Although no doubt, working with other shapes would be usualful, I think your time would be better spent with the inputting of anchor points. This facility is included within Max's cloth script.

Anchoring points makes it possible to generate tent structure by setting gravity to a negative number:



I wonder if it would be easier to define, if instead of a rectangle input as you now have, it worked with a point array input (ie with MathsPoints or Points node input) In this way, the cloth point density is externally defined and anchor points selected with idSelect node.

Anyway, I know that this is no easy task, & what you done is just brilliant, I am just think about how this functionality may be integrated into the node ecosystem for procedural modelling.

Great work
James
https://www.instagram.com/nodeology/

Image Attachments:
tentStructure.jpg 


From: wayne hill (WAYNEHILL5202)
14 Jun 2020   [#187] In reply to [#186]
James,

Yes, the anchor points are a part of the code. I was excited to see the first conversion drop cloth running, it was released for testing before setting up anchor points option. Negative numbers, great tip.

A point array would work since Max is converting the plane to a grid point array. That would skip a step.

Wayne
From: James (JFH)
14 Jun 2020   [#188] In reply to [#187]
Wayne,

"A point array would work since Max is converting the plane to a grid point array. That would skip a step."

Ideally both the geometric data input and output would be point arrays, so that they delineate curves or polylines in U&V directions. Whereas curves may be lofted or networked, polylines could be polylofted. I know that this has presented difficulties with other nodes (ImgSampler), so if it is only possible to output curves, this would still be fantastic.

"Negative numbers, great tip."

Ahh! Negative gravity: this was the underlying methodology behind Gaudi's Sagrada Familia

http://www.ipernity.com/doc/jacquespfjc/13188615

Looking forward to the next beta
James
https://www.instagram.com/nodeology/
From: wayne hill (WAYNEHILL5202)
18 Jun 2020   [#189] In reply to [#188]
James,

Added the anchor index point numbers of the grid array. No error checking...
You have negative gravity :-)

Left in the error checking "file.writefile()" outputs I use to watch the program flow.

Still working on the grid array input.

Wayne
From: Frenchy Pilou (PILOU)
18 Jun 2020   [#190] In reply to [#189]
Cloth2.js must be inside the folder
Moi\nodeeditor\nodes\extensions\ ?
From: wayne hill (WAYNEHILL5202)
18 Jun 2020   [#191] In reply to [#190]
Pilou,

That is correct. It is still a beta program.

Wayne
From: James (JFH)
18 Jun 2020   [#192] In reply to [#189]
Thanks Wayne,

The addition of anchor selection is a great advancement.
Did you have a look at using a point array instead of a plane
as the input?

I don't know if Max'x code would even allow it, but it would
really be something, if anchor points could be dynamically
manipulated (re-positioned), with the effect of stretching &
tugging the fabric.

It hints at great promise.....
very exciting

James
From: wayne hill (WAYNEHILL5202)
19 Jun 2020   [#193] In reply to [#192]
James,

Cleaned up the inputs.
The input are mostly on the node.

Still working on moving points and setting array inputs.

Wayne
From: James (JFH)
20 Jun 2020   [#194] In reply to [#193]
UPDATED
________

Wayne,

Although the decluttering of the node interface makes it more user-friendly, I think you may have pushed it too far.

It would be advantageous to dynamically set gravity (as the principle force).
The other material behaviours would in most cases only be set at the outset, so it is appropriate to have their adjustments made on the info panel.

A numerical value could be input for EndFrame, which could serve both as setting the maximum number of frames in the animation, while only outputting the geometry defined by the final frame. The user could start with a high number, and then dial it back to the frame they want to output.

Also I see in Max's code that wind is treated as a vector, so there would be interface advantage in inputting as a vector, as then a single input gives direction & magnitude. I imagine, in most cases wind would not be utilised, anyway.

This is just my 2 cents,
You are doing a fabulous job.

James
https://www.instagram.com/nodeology/


Updated: I removed net width & height inputs from node because they are, of course, already inherent within the pointArray input.

PS. the inclusion of XY , X , Y , P pulldown menu is great, & although a small thing, should more appropriately be labelled UV , U , V , Pts

Image Attachments:
FabricSim.gif 


From: wayne hill (WAYNEHILL5202)
22 Jun 2020   [#195] In reply to [#194]
James,

The latest work in progress beta version. Renamed to FabricSim. Added WindVector input. Wind calculations with vector indicator. Did come across that error in the vector code where the indicator points up when it should point down on X0 Y0 Z- values . Pilou detailed it in another thread.

Wayne

Show messages:  1-15  …  116-135  136-155  156-175  176-195  196-215  216-235  236-255  …  416-425