Macha Posted May 20, 2011 Share Posted May 20, 2011 I have displaced geometry by a tiny amount. It's not a true displacement so occasionally I get white noise just on the edge of an object. I know this must be because of displacement because if I reduce it to 0 or use true displacement they go away. Any idea whether there is another way of getting rid of these? Quote Link to comment Share on other sites More sharing options...
kustaa Posted May 20, 2011 Share Posted May 20, 2011 its a wild guess but if your artifacts are about the size of your pixel filter it might be that your shader is returning some wild numbers or not a numbers at all. and this tiny displacement is causing them on edges of the object. what rgb values you have on your artifacts? K Quote Link to comment Share on other sites More sharing options...
Macha Posted May 20, 2011 Author Share Posted May 20, 2011 I get reasonable RBG values and it blends across a few pixels. I have a feeling it's related to DOF as well. Quote Link to comment Share on other sites More sharing options...
Jason Posted May 26, 2011 Share Posted May 26, 2011 I get reasonable RBG values and it blends across a few pixels. I have a feeling it's related to DOF as well. What is the primitive type you're rendering? Polygons? Primitives? Patches? Quote Link to comment Share on other sites More sharing options...
Macha Posted May 26, 2011 Author Share Posted May 26, 2011 They are polygons with mantra rendertime subdivision. Quote Link to comment Share on other sites More sharing options...
Serg Posted May 26, 2011 Share Posted May 26, 2011 I find that rendertime subdivision often returns pixels with invalid normals. Quote Link to comment Share on other sites More sharing options...
thekenny Posted May 26, 2011 Share Posted May 26, 2011 does your geometry have a normal attribute when before it is rendered? It could have been the shader i was using at the time, in another life, but I remember seeing a difference if I sent the geometry to render with the subd on the fly IF it had normals and if it didn't. Just a thought.. -k Quote Link to comment Share on other sites More sharing options...
Jason Posted May 26, 2011 Share Posted May 26, 2011 I find that rendertime subdivision often returns pixels with invalid normals. Hm, if this is so, maybe make sure your normals are normalized before displacement? Quote Link to comment Share on other sites More sharing options...
Serg Posted May 27, 2011 Share Posted May 27, 2011 Hm, if this is so, maybe make sure your normals are normalized before displacement? I tried everything. Its literally the odd pixel here and there, visible in a straight N output. Next time I see it I'll setup a simple scene to send to Support, again. I say "again" because I have previously sent them a scene which had pixel errors related to dicing pole topology, which they fixed for 11 during beta period. This seems something else... random distribution frame to frame, more concentrated on concave areas iirc. Quote Link to comment Share on other sites More sharing options...
thekenny Posted May 27, 2011 Share Posted May 27, 2011 hmm, do you have unique point{s} that result in unshared edge{s}? I just went through a problem (not houdini/mantra) but the theory is sound based on various approaches to subdivide mesh that you might encounter it. If your mesh is suspect you could try a divideSOP and use the unshared edges option to see if you have something evil lurking, or you could try a subdivideSOP with a value of 1. If there is something in your mesh that is not making the subdivision on the fly happy, that SOP should prepare the mesh well for it to be acceptable. General rule of thumb.. if it looks right in a subdivideSOP then it should be fine to be subdivided on the fly. Just a thought. -k I tried everything. Its literally the odd pixel here and there, visible in a straight N output. Next time I see it I'll setup a simple scene to send to Support, again. I say "again" because I have previously sent them a scene which had pixel errors related to dicing pole topology, which they fixed for 11 during beta period. This seems something else... random distribution frame to frame, more concentrated on concave areas iirc. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.