Login

buy downloads forums docs

Summary

Title:	Barrycentric interpolation on hair producing incorrect result
Category:	Workflow
Status:	Open
Posted By:	baothebuff ( )
Date Created:	1 June 2015

Problem

Description:	Barrycentric interpolation on hair producing incorrect result with strand clustering. You would expect the hairs to clump around the existing guides, but it's not doing that correctly when looking at the roots. Every other interpolations seems to be working fine with it as it is producting even root distribution. I don't know if it just does this with the strand clustering or if barrycentric interpolation is broken in general but this makes it real apparent. - Tom
Steps to Reproduce:	create plane. 4x4 segments add oxGuidefromSurface, set to vertext distribution add oxHairfromGuide add oxStrandClustering, making the tip very pointy. if you look at the roots, they are not in distinct "zone" and kind of all over the place. If you switch the interpolation in the oxHairfromGuide to something else, it produces more correct and pleasing results. I'm using the latest 4.0 beta.
	Attached Files:

1 June 2015 #6169
baothebuff	here's the other interpolations for reference. Attached Files:
1 June 2015 #6171
EPHERE	Thank you Tom, I will let you know when I perform more investigation on this. Marsel Khadiyev (Software Developer, EPHERE Inc.)
18 July 2015 #6341
baothebuff	Any news on this? I'm still seeing it in the lastest 4.x build. - Tom
18 July 2015 #6342
EPHERE	Haven't gotten to this yet. Something tells me that this is just a side effect of barycentric interpolation which heavily depends on the distribution mesh topology as it calculates the closest guides based on vertices rather than physical proximity to the guides. Marsel Khadiyev (Software Developer, EPHERE Inc.)
18 July 2015 #6343
EPHERE	A couple of notes: It seems you are referring to the "Ox Strand Clustering", not "Ox Hair Clustering" modifier As suspected, this is an artifact of barycentrci interpolation. Try increasing the number of plane segments in the underlying distribution object and you should see improvement in clustering. Unfortunately this is how this interpolation type works so not much can be done besides making the topology denser or using a different interpolation method. Marsel Khadiyev (Software Developer, EPHERE Inc.)