Hair-like rendering / noise

   6953   10   2
User Avatar
Member
121 posts
Joined: July 2005
Offline
I'm rendering some hair-like geometry. They are just 4 point open polygon curves copied to a sphere and they have a width attribute going from 0.006 at the root to 0.001 at the tip. They also have a black to white ramp applied via point attributes which then get referenced by a constant shader. The camera moves slowly.
When I render this setup, I get noise at the tips of each “hair”, which becomes even worse when applying DOF.

I tried rendering with super sample up to 7x7, jitter down to 0 and shading rate up to 2 in various combinations, but the noise is always there.

Here is a short closeup (tga) sequence of the noise with [people.freenet.de] and without [people.freenet.de] DOF.


Any ideas how to get rid of this?

Thanks!
Achim
User Avatar
Member
196 posts
Joined: July 2005
Offline
Erm… Have you tried using the geometric surface normal instead? Ng instead of N?

H
Henster
User Avatar
Member
121 posts
Joined: July 2005
Offline
Hey Henster, where should I use Ng instead of N?

The only place where I use N is in the SEM shader, but I think changing this to Ng won't make any difference, as the noise is caused by the “hair”, which has a constant shader applied, and not the SEM shader.
User Avatar
Member
196 posts
Joined: July 2005
Offline
Sorry, I've checked the images now… it sounded a little bit like a problem I had involving 2 sided shaders….. but its not.

The noise I can see is caused by each hair only being a pixels width which jumps obviously when moving from one location to the next?

My easy fix would be render off a seperate pass and blur it!

Or render at a higher res and shrink it hoping the resampling deals with it?

This hasn't helped has it….

H
Henster
User Avatar
Member
121 posts
Joined: July 2005
Offline
Thank you for the suggestions Henster

I rendered at 2 x the solution, and this certainly reduces the noise, but renders way to slow.

I'll also try to render a seperate pass and blur it, but I guess this will result in the same problem, as DOF is basicly also a blur effect. But I'll try it anyway.
User Avatar
Member
212 posts
Joined: July 2005
Offline
If it was renderman I would say to change the Output Driver parameter for MotionFactor from 0 to 1. For mantra…. i'm not sure what the option would be called. Mind you I haven't looked at the images but from the description it sounds like the micropolygons are being sampled at a lower shading quality for the depth of field rendering. That would be the case in Renderman so I'd assume it would be similar in Mantra.

I believe Wolfwood posted a nice file for depth of field with mantra on the Odforce pages.

good luck
-k
User Avatar
Member
121 posts
Joined: July 2005
Offline
I had a look at that file, but I forgot to mention previously, that I'm using cops for the DOF effekt, not the mantra build in version.
User Avatar
Member
212 posts
Joined: July 2005
Offline
If you are doing it in COPs you are stuck with the option of rendering it larger and blurring it and then scaling it down. Your are stuck with the pixels as is and there isn't much you can do when bluring them in post.
I don't know if a 16 bit image for the ramp would help…

sorry.
-k
User Avatar
Member
121 posts
Joined: July 2005
Offline
Do you mean to apply DOF and then scale the image down, or first to scale it down and then apply the DOF to it?

I already render the z-depth map at floating point, so no room for improvement there. I tried rendering with mantra DOF, but the render doesn't even start after 10 minutes, so I guess the larger rendering/scaling down option is the only thing I can do.
User Avatar
Member
412 posts
Joined: July 2005
Offline
i don't know how much i can help, but i recently did a project involving a dandelion seed pod with all those tiny seeds coming out of it and each were being animated by sound…

so in order to get the noise down for those suckers, i had an 8x8 supersample and a micro-poly shading level of 3 (at the rop level, rather than obj)..

sounds like crazy render times, huh? well it was.. 2.5 hours/frame when copying the seedlings (lystem actually) to each point of the head pod via copy sop..got that down to 2.5 min/frame after using point instancing instead of copy sop…and practically no noise as well…

hth somewhat… good luck… like your results so far.. keep us updated…

dave
Dave Quirus
User Avatar
Member
121 posts
Joined: July 2005
Offline
Thank you deecue, I'll try it with shading quality 3 and point instancing instead of copy sop solution.
  • Quick Links