Forums Technical Discussion Artifacts with Specular Angle Controlled by Texture Map

Artifacts with Specular Angle Controlled by Texture Map

   3516   3   0
User Avatar
briankulig
Member
34 posts
Joined: Nov. 2007
Offline
 
Hey all,

I am seeing an artifact when trying to use H11/PBR Render and the mantra surface shader. I want to control the specular angle with a texture map, but when I do, I get a ragged edge of black around areas transitioning between values.

I used the mantra “uber” shader. I went in and added a texture map to the specular angle using the plug controls on the right of the parameter in the vop network. I tried different render options, including the micro poly renderer. The pattern of the artifacts changes with the sampling but never goes away.

I provide a simple hip(nc) and image that illustrate the problem.

Any ideas? Am I doing something wrong or is something broke?

-bk

Attachments:
sample.jpg (69.7 KB)
SpecAng.tif (3.3 MB)
testingMaterials.hipnc (865.4 KB)

 Report
User Avatar
briankulig
Member
34 posts
Joined: Nov. 2007
Offline
 
I was able to fix the issue by never letting the map drive the specular angle below 0.1. I did this by adding a “Fit Range” node to the vop network after the texture.

Attachments:
testingMaterialsFixed.hipnc (863.5 KB)
sample2.jpg (79.8 KB)

 Report
User Avatar
malexander
Staff
5212 posts
Joined: July 2005
Offline
 
Just a guess - is the specular angular map an 8bit texture? I've seen similar artifacts when using COPs with 8bit images driving lookups. (I can't tell because the tif didn't appear to attach).

Try making it a FP32 map, or increase the resolution of the texture, and see if that makes a difference.
 Report
User Avatar
briankulig
Member
34 posts
Joined: Nov. 2007
Offline
 
Thanks for looking at it! The tif is the supplied circle image in the original post. If you right click and save the image you will get it correctly. It is a 16bit tiff. But as I am testing, it seems unrelated to the tif image.

After reading your response, I tried removing the tif from the equation and instead used the “stone wall” procedural instead. I assume this is 32bit float. With the clamp the image looks good, without it I get a similar artifact as before.

This issue makes me wonder if there is some threshold on the specular angle in the code for the shader that switches to a cheaper calculation when its under 0.1 and it is just clamping there to 0… thus getting the artifact.

I supplied the new hipnc file if it helps.

Attachments:
sample3_fitRangeOFF.jpg (116.7 KB)
sample3_fitRangeON.jpg (68.3 KB)
testingMaterialsWithStone.hipnc (970.8 KB)

 Report
  • Quick Links