Lets test DissolveSOP

[Guest mantragora]

Hi.

Here is video:
http://1drv.ms/1Nmzrw3

Make the same steps and see if you get all the edges extracted when set to Delete Non-Selected. On my machine if I place SphereSOP set to Polygon Mesh and select edge between p2-26, I get extracted only points, not edges.

Especially test edges that starts in point . Some are working, but others not. For example, if I select edges between p0-23-47-71-95, I get only points, but If I do the same selection and just remove first edge that starts at , I get extracted edges correctly.


Thanks in advance!

[Guest mantragora]

What's even worse, on a TorusSOP, if I try to DissolveSOP on p0-17 in Delete Non-Selected, I get no points, no primitives, no vertices. Completely blank GDP. 

 

Anyone else?

[anim]

for the Sphere case it seems that the prim is there it's just closed

if you append Ends SOP and set to Open, it will magically appear

 

 

for Torus, I don't see points 0 and 17 being neighbours, so it makes sense it doesn't produce anything, maybe I have different defaults to torus though

[Guest mantragora]

for Torus, I don't see points 0 and 17 being neighbours, so it makes sense it doesn't produce anything, maybe I have different defaults to torus though

 

Rows: 12

Columns: 24

and select p0-23

 

or

 

Rows: 12

Columns: 18

and select p0-17

 

or

 

Rows: 12

Columns: any

and select p0-`ch("../torus/cols") - 1`

Edited September 28, 2015 by fântastîque Mântragorîè

[anim]

yup, the same results here

[Guest mantragora]

yup, the same results here

 

 

Ehh... that means blank geometry? or EndsSOP set to Open? or what? 

 

I need confirmation before I attach this topic to my second response for a BUG I logged on DissolveSOP.

Edited September 28, 2015 by fântastîque Mântragorîè

[anim]

sorry, same as you, empty geo

H14.0.367

Edited September 28, 2015 by anim

[Guest mantragora]

Submitted as Ticked: #71431

[Guest mantragora]

OK, I'm officially at war with this sucker.

 

1. Place GridSOP, set 10 on SizeX/SizeY/Rows/Columns.

2. Append BlastSOP to GridSOP and set to Primitives and in group put 40-41 50 49

3. Append DissolveSOP to BlastSOP, set it to Delete Non-Selected and in group field put p44-43-53-62-63

 

Q1: How many primitives and vertices you get?

Q2: How many primitives and vertices you would expect to get?

 

PS.

I get 3 primitives and I would expect to get only 1. I get 11 vertices and I would expect to get 8.

Edited October 5, 2015 by fântastîque Mântragorîè

[Guest mantragora]

Guys, say something!

Are you getting similar results?

Or does it work correctly on your machine?

 

I think that Delete Non-Selected mode is totally broken.

[f1480187]

Same bug for me. Why you expect 1 primitive, not 4? And if 1 primitive expected, why 8 vertices, not 5? There also third option to get two L-shaped primitives and 6 vertices of them.

dissolve_expectations.hipnc

Edited October 10, 2015 by f1480187

[Guest mantragora]

Same bug for me. Why you expect 1 primitive, not 4? And if 1 primitive expected, why 8 vertices, not 5? There also third option to get two L-shaped primitives and 6 vertices of them.

 

Because if I do the same selection in two different places, I get each time different outcome.

 

 

And 8 vertices because middle points, each contain two vertices, from each edge that shares this point. Only begin and last point shouldn't contain additional vertices.

On top of it, there is no possibility to sort those points correctly, so you woul get 0 on begin and biggest point number on end of shape. You have to rebuild whole extracted edge yourself, and if you have to do this, what's the point of using DissolveSOP at all?

 

BTW.

Bug I logged was already expanded with those insights.

Edited October 11, 2015 by fântastîque Mântragorîè

[f1480187]

Isn't it degenerate case when point shares two vertices within one primitive? I can't figure how to get it in sop without manual adding vertices in cvex.

[Guest mantragora]

In those cases degenerate is because some primitives end up with more than one vertex in points that are not shared by multiple edges. And that's what breaks possibility to sort it correctly. Otherwise you could probably just use FuseSOP followed by JoinSOP and SortSOP set to By Vertex. 

 

The biggest problem is that, if you don't go the way of rebuilding extracted edge yourself, you will end up with additional couple nodes to fix what DissolveSOP brakes , and even then you are not sure that you will get what you would expect each time.

 

With some trickery (to have first and last point) and FindShortestPathSOP it's very easy to rebuild edge in cases where you have just simple edge extraction from between just two ends. But then again, what's the point of existence of DissolveSOP if I have to do this myself?

Edited October 11, 2015 by fântastîque Mântragorîè

[f1480187]

As a temporal solution it can be fixed using Carve (0..1 U, "Break at All Internal U Breakpoints"), Fuse and PolyDoctor at default settings. However, it will produce one prim per edge.

Edited October 11, 2015 by f1480187

[Guest mantragora]

If you add two more nodes (JoinSOP followed by SortSOP) you will get what I want. That's 5 additional nodes.

I can do the same with 2, PythonSOP (to find last and first point) + FindShortestPathSOP (which will output it as one primitive and correctly sorted).

Edited October 11, 2015 by fântastîque Mântragorîè