Face.classify_point - compare results?

thomthom

I'm running into some pretty weird problems with the classify_point method. Here's the code I'm running:

puts ii.classify_point(curPoint).to_s
> if ii.classify_point(curPoint)&8
> ...

The value 32 is always sent to the console, yet the if statement executes, even though 32&8 = 0. Can someone help me out with this?

Take note of the description of that method: http://code.google.com/apis/sketchup/docs/ourdoc/face.html#classify_point
Don't use hard-coded numbers, but instead use the provided constants, as the values are different between versions.

Dan Rathbun

Ok, so what he's doing should read as:

if ii.classify_point(curPoint) & Sketchup::Face::PointOnFace != Sketchup::Face::PointUnknown

TIG

Or perhaps more transparently the full list from the API is

* 0: Sketchup::Face::PointUnknown (indicates an error),
* 1: Sketchup::Face::PointInside (point is on the face, not in a hole),
* 2: Sketchup::Face::PointOnVertex (point touches a vertex),
* 4: Sketchup::Face::PointOnEdge (point is on an edge),
* 16: Sketchup::Face::PointOutside (point outside the face or in a hole),
* 32: Sketchup::Face::PointNotOnPlane (point off the face's plane). So you can test for several possibilities by Constant - for 'on the face itself' ...

if ii.classify_point(curPoint)==Sketchup::Face::PointInside etc...
Or if you wanted it to be only on a vertex you'd use ...
if ii.classify_point(curPoint)==Sketchup::Face::PointOnVertex etc...

TIG

Correction: the API doc's list is completely foobar...
the proper list [by Brad] is [probably ]

0: Sketchup::Face::PointUnknown (indicates an error)
1: Sketchup::Face::PointInside (point is on the face, not in a hole)
2: Sketchup::Face::PointOnEdge (point is on an edge)
4: Sketchup::Face::PointOnVertex (point touches a vertex)
8: Sketchup::Face::PointOutside (point is outside the face or in a hole)
16: Sketchup::Face::PointNotOnPlane (point is off the face's plane) There is no 'PointOnFace' ??

thomthom

@tig said:

Correction: the API doc's list is completely foobar...
the proper list [by Brad] is [probably ]

0: Sketchup::Face::PointUnknown (indicates an error)

1: Sketchup::Face::PointInside (point is on the face, not in a hole)

2: Sketchup::Face::PointOnEdge (point is on an edge)

4: Sketchup::Face::PointOnVertex (point touches a vertex)

8: Sketchup::Face::PointOutside (point is outside the face or in a hole)

16: Sketchup::Face::PointNotOnPlane (point is off the face's plane) There is no 'PointOnFace' ??

His list of constants and values is not relevant for SU8M1

TIG

So... thomthom what IS the list ?
Why the secrecy?
I now think it should be from Sketchup::Face.constants - which returns
["PointOutside", "PointInside", "PointOnFace", "PointUnknown", "PointOnEdge", "PointNotOnPlane", "PointOnVertex"]
I assume in the order they list here and not related to any 'integer' value they represent I THINK they are now...

PointOutside= Outside the outer loop edges of the face, but perhaps still on its plane, BUT does a hole count as 'outside' ?
PointInside= Inside the outer loop edges of the face, but perhaps in a hole so not 'onface' ?
PointOnFace= On the face - different from PointInside - as it cannot be in a hole - it is therefore within all loops ?
PointUnknown= An Error - invalid point
PointOnEdge= On an edge belonging to the face
PointNotOnPlane= A valid point, but it's NOT on the plane of the face
PointOnVertex= On a vertex belonging to the face
This really is a complete mess ! Bitwise was much more logical that f'ing 'constant names'

thomthom

I think the list is the one listed in the docs now. There was an old value listed before, which I think where never returned. And SU8M1 introduced a new value.

What appear to have changed in the docs is that before it said that the comparisons had to be done bitwise.
I am a bit confused about best practice myself. I only found out about the constants when they released M1. The info has been sparse and I've been doing a bit of trial and error to get things working.

thomthom

You can use the WayBack Machine to see the older versions of the docs:
http://waybackmachine.org/*/http://code.google.com/apis/sketchup/docs/ourdoc/face.html#classify_point

I had basically no info....

Not sure where I got the bitwise comparison from...

TIG

The current API lists 6 Constants [Brad has 7] BUT the .constants method returns 7 that aren't the same as Brad's 7 so something is wrong soemwhere?? aaargh! My post http://forums.sketchucation.com/viewtopic.php?p=316191#p316191 list what I think might be right [today at least ] ???

thomthom

Cheers TheDro

TIG

TheDro
Thanks, could you perhaps publish a definitive list in the Face API note explaining which is what and that there is [possibly] an obsolete constant too... Comparing and contrasting with the current text and the Brad's follow up notes too...

TheDro

Alright, let's try to put this one to rest . I just tested it with a face on the xy plane with a hole in it and here are the results(and code):

puts "On Face(Inside); " + face.classify_point([0.25,0.25,0]).to_s #1
puts "On Vertex; " + face.classify_point([0,0,0]).to_s             #2
puts "On Edge; " + face.classify_point([0.25,0,0]).to_s            #4
puts "Outside; " + face.classify_point([-0.5,0,0]).to_s            #16
puts "In Hole; " + face.classify_point([0.6,0.6,0]).to_s           #16
puts "NotOnPlane; " + face.classify_point([0,0,1]).to_s            #32

So this means that whether the point is in a hole or outside the face, the return value is the same (16). I think that answers TIG's question about the matter. The OnFace constant (which is is not in the API nor am I able to choose a point that returns the value. It could therefore be obsolete. If someone can think of another case that I missed, let me know (other than an error (0) of course).

Oh and just a quick note. The return value of the classify_point method will be one of the above values and NOT a sum of multiple values. I mention this because I expected a point on an vertex to return 6 or 7 because a point on a vertex is on the face and an edge at the same time, but this is not the case.

As for how I'll write my code, well there are two ways of doing and I think this would be the shortest way (syntactically) of doing it:

if (face.classify_point(point)&(Sketchup;;Face;;PointInside+Sketchup;;Face;;PointOnVertex+Sketchup;;Face;;PointOnEdge))!=0

As long as those constants remain powers of 2, the code should work although to be safe I might just go with if((p==a) or (p==b) or (p==c)), which doesn't require any bitwise operations (I just find the bitwise way of doing cool since I've come to understand it recently ).

EDIT:
As requested, here is the definitive list of the constants:

0: Sketchup::Face::PointUnknown (error)
1: Sketchup::Face::PointInside (somewhere on the face)
2: Skethcup::Face::PointOnVertex (on one of the vertices of the face)
4: Skethcup::Face::PointOnEdge (on one of the edges of the face, vertices do not count since they return 2)
8: Sketchup::Face::PointOnFace (likely obsolete)
16:Sketchup::Face::PointOutside (on plane of the face but not on face; this includes holes)
32:Sketchup::Face::PointNotOnPlane (not on the face's plane)

Remember that this is for version 8.0 and that the constants should be used instead of the numerical values for forward compatibility.

TIG

TheDro has edited the list provided a few posts back with 'all know data'...
Unexpectedly it agrees exactly with the list in the main API -[except for that omitting '8 PointOnFace' as it is obsolete anyway - why is it in the constants still ? Have Google just forgotten to remove it ?]
Brad's list in the API notes is therefore wrong [probably].
I've suggested to TheDro to add another note to the API docs to state what seems to be the correct situation with v8M1

thomthom

I digged in my emails - here's a quote from a Googler:

@unknownuser said:

Here is a comment from our defect database about this issue

"..I learned that the Sketchup::Face::PointOnFace constant is never used, so I just removed it from the docs.."

TIG

Out of interest I wrote this snippet
Sketchup.constants.sort.each{|e|puts e;eval("Sketchup::"+e+".constants").each{|c|puts "\t"+c};puts}
which prints out all Sketchup 'constants' - as this on my system at least [e.g. I expect Geodesic is an extra type added by a script...] There aren't that many 'constants' found this way...

AppObserver

ArcCurve

AttributeDictionaries

AttributeDictionary

Behavior

Camera

Color

ComponentDefinition

ComponentInstance

Console

ConstructionLine

ConstructionPoint

Curve

DefinitionList

DefinitionObserver

DefinitionsObserver

Drawingelement

Edge

EdgeUse

Entities

EntitiesObserver

Entity

EntityObserver

Face
	PointOutside
	PointInside
	PointOnFace
	PointUnknown
	PointOnEdge
	PointNotOnPlane
	PointOnVertex

Geodesic
	TETRA_S
	SQRT2
	PRIMITIVES
	TETRA_R
	GOLDEN_MEAN
	TETRA_Q
	TETRA_T
	SQRT3

Group

HLR

Image

Importer
	ImportFail
	ImportFileNotFound
	ImportSuccess
	ImporterNotFound
	ImportCanceled

InputPoint

InstanceObserver

Layer

Layers

LayersObserver

Loop

Material

Materials

MaterialsObserver

Menu

Model

ModelObserver

OptionsManager

OptionsProvider

OptionsProviderObserver

Page

Pages
	UnitsPixels
	ImageLinked
	UnitsNormalizedY
	ImageEmbeddedAndLinked
	UnitsNormalizedX
	ImageEmbedded

PagesObserver

PickHelper

RenderingOptions
	ROPAssign
	ROPSetDisplayInstanceAxes
	ROPSetFogHint
	ROPSetDisplaySketchAxes
	ROPSetGroundColor
	ROPSetExtendEdges
	ROPSetLineExtension
	ROPSetDepthQueWidth
	ROPSetSectionDefaultCutColor
	ROPSetFogUseBkColor
	ROPSetTexture
	ROPSetHighlightColor
	ROPEditComponent
	ROPSetModelTransparency
	ROPSetDrawUnderground
	ROPSetFaceColorMode
	ROPSetFogDist
	ROPSetLockedColor
	ROPSetSkyColor
	ROPSetJitterEdges
	ROPSetExtendLines
	ROPSetLineEndEdges
	ROPSetSectionInactiveColor
	ROPSetDisplayDims
	ROPSetTransparencyObsolete
	ROPSetForegroundColor
	ROPDrawHidden
	ROPSetEdgeType
	ROPSetDrawGround
	ROPSetProfileWidth
	ROPSetFogColor
	ROPSetProfilesOnlyEdges
	ROPSetHideConstructionGeometry
	ROPSetEdgeColorMode
	ROPSetDisplayColorByLayer
	ROPSetDepthQueEdges
	ROPSetSectionActiveColor
	ROPSetDisplayText
	ROPSetRenderMode
	ROPSetBackgroundColor
	ROPSetSectionDisplayMode
	ROPSetFaceColor
	ROPSetDrawHorizon
	ROPSetProfileEdges
	ROPSetDisplayFog
	ROPSetLineEndWidth
	ROPSetSectionCutWidth
	ROPSetEdgeDisplayMode
	ROPSetConstructionColor
	ROPTransparencySortMethod
	ROPSetMaterialTransparency
	ROPSetGroundTransparency

RenderingOptionsObserver

SectionPlane

Selection

SelectionObserver

ShadowInfo

ShadowInfoObserver

Style

Styles

Text

Texture

TextureWriter

Tools

ToolsObserver

UVHelper

Vertex

View

ViewObserver

["AppObserver", "ArcCurve", "AttributeDictionaries", "AttributeDictionary", "Behavior", "Camera", "Color", "ComponentDefinition", "ComponentInstance", "Console", "ConstructionLine", "ConstructionPoint", "Curve", "DefinitionList", "DefinitionObserver", "DefinitionsObserver", "Drawingelement", "Edge", "EdgeUse", "Entities", "EntitiesObserver", "Entity", "EntityObserver", "Face", "Geodesic", "Group", "HLR", "Image", "Importer", "InputPoint", "InstanceObserver", "Layer", "Layers", "LayersObserver", "Loop", "Material", "Materials", "MaterialsObserver", "Menu", "Model", "ModelObserver", "OptionsManager", "OptionsProvider", "OptionsProviderObserver", "Page", "Pages", "PagesObserver", "PickHelper", "RenderingOptions", "RenderingOptionsObserver", "SectionPlane", "Selection", "SelectionObserver", "ShadowInfo", "ShadowInfoObserver", "Style", "Styles", "Text", "Texture", "TextureWriter", "Tools", "ToolsObserver", "UVHelper", "Vertex", "View", "ViewObserver"]

thomthom

RenderingOptions got contants!!

One can actually use constants - that refer to integers (I assume indexes) to refer to options. That should be more efficient than strings.

TIG

@thomthom said:

RenderingOptions got contants!!
One can actually use constants - that refer to integers (I assume indexes) to refer to options. That should be more efficient than strings.

Yes it's weird ?
I can't really see what half of these do...

thomthom

hm... yea - the names are odd. Might just be internal constants. Which would explain why they are not documented.

TheDro

Alright so I posted my results on the api but is it possible to edit it in case there is a mistake?(not that I've found one) Oh and it's posted as anonymous.... whatever.