Someone provided me a case class I could test on yesterday. I've found and fixed a bug where extra subgeometries were added. Explains why the MeshHelper recenter was failing and also why using apply method in a loop was that slow...
Fix is uploaded. Fabrice On Aug 2, 2011, at 6:36 PM, DBird wrote: > I thougt is 1 by default: > > Cube (material:MaterialBase = null, width:Number = 100, height:Number > = 100, depth:Number = 100, segmentsW:uint = 1, segmentsH:uint = 1, > segmentsD:uint = 1, tile6:Boolean = true) … > segmentsW: The number of segments that make up the cube along the X- > axis. Defaults to 1. > > ...but to test it, I used this piece of code: > > var cube1:Cube = new Cube(colMat, 100, 200, 100, 1, 1, 1); > var cube2:Cube = new Cube(colMat, 100, 200, 100, 1 ,1 ,1); > > 48 Polys after the merge... :( > > Cheers, > Dirk > > On 2 Aug., 16:07, Fabrice3D <[email protected]> wrote: >> The Cube primitive class doesn't have the same default subdivisions in both >> engines. >> and in your Cube constructors code you do not set them. >> >> Fabrice >> >> On Aug 2, 2011, at 2:41 PM, DBird wrote: >> >> >> >> >> >> >> >>> Hi Fabrice, >> >>> I'm still on the merge-problem. Perhaps you could give me a quick info >>> about that. >> >>> I'm using the merge-class to put cubes together, something like this: >> >>> // Creating my "word-Cube" witch persists >>> var myCube01 = new Cube( { material:"green#lime" } ); >>> myCube01.x = -300; >>> view.scene.addChild(myCube01); >> >>> // Creating the source-cube, for adding cubes as often as needed >>> var myCube02 = new Cube( { material:"green#lime" } ); >>> myCube02.x = -150; >>> // and not adding it to the scene. the merge-object will do this. >> >>> // create MergeTool >>> var merge:Merge = new Merge(false, true, false); >> >>> // Merge them baby >>> merge.apply(myCube01, myCube02); >> >>> // After that we have 3 Cubes, because i put myCube02 again in >>> myCube01. is like a copy of it. >>> myCube02.x = 150; >>> merge.apply(myCube01, myCube02); >> >>> After testing this with Away3.6 and Broomstick I got these results >>> from the debug-panel: >> >>> Test with Molehill & Broomstick >> >>> Cubes Polys normal Polys after merge >>> 2 24 48/0 >>> 3 36 144/0 >>> 4 48 384/0 >>> 5 60 960/0 >> >>> Test with FP10.2 & Away3.6 >> >>> Cubes Polys normal Polys after merge >>> 2 07/24 07/12 >>> 3 11/36 11/12 >>> 4 15/48 15/12 >>> 5 19/60 19/12 >> >>> What is happening with the molehill-test? A bit to many polys... >> >>> I thing, because of this I can't compile your example >>> http://www.closier.nl/broomstick/mergetest.html >> >>> Any idea? >> >>> Thanks again in advance! >>> Dirk >> >>> On 30 Jun., 15:23, Fabrice3D <[email protected]> wrote: >>>>> Yes they are, but this sounds a bit more tricky... >> >>>> not really trickky, just a bit more work, a brick = from / to >>>> so if you loop over the mesh indices vector, you simply increase count + = >>>> mybrickfacecount >> >>>>> To get one "brick" out (to delete it), i have do some vertice-operation >>>>> or a >>>>> complete rebuild without that deleted brick. is this correct? >> >>>> yes and no. >> >>>> yes it is correct you can do this way. >>>> no, you could ask FaceHelper class for this dirty job ;) >> >>>> Fabrice >> >>>> On Jun 30, 2011, at 2:08 PM,DBirdwrote: >> >>>>> Wow, that's really good news! Thank you very much. >> >>>>> Ah, and for all who what to watch it, here is an example from Fabrice: >>>>> http://www.closier.nl/broomstick/mergetest.html >> >>>>> ... and some addition background information: >>>>> http://www.mail-archive.com/[email protected]/msg21091.html >> >>>>>> if your bricks have exact same face count, it would be also easy to >>>>>> alter them vertice wize into the buffers. >>>>> Yes they are, but this sounds a bit more tricky... >> >>>>> After some code-reading I found out, that after the merge, there is no >>>>> access to all my objects I have merged in my master-mesh. To get one >>>>> "brick" out (to delete it), i have do some vertice-operation or a >>>>> complete rebuild without that deleted brick. is this correct? >> >>>>> Dirk
