Every time I try to use procedural (any noise type) bump or displacement on clothing, these artifacts show up on alternating polygons. Please do not suggest any of the following as they don't fix the problem:

1. Turn on or off shadows
2. Turn on or off raytracing
3. Turn on or off smoothing
4. Turn on or off displacement
5. Change min shading rate (.01, .1, .2, .5, 1, nothing works)
6. Turn on or off Normals Forward
7. Any form of Post filtering.

This is really grinding my gears as I have avoided many interesting shader effects due to this problem.

I haven't yet tried installing SR2 as I'm trying to get a feel for P7 quirks before changing to SR2. I need to know the differences with and without SR2 so that when people have questions or problems I know what matters and what doesn't.

However, I don't want to go crazy over an issue that has been fixed in SR2. I suspect it hasn't been fixed.

I've not been able to publish any good velvet or furry textures except a few very limited cases.

Any thoughts?

Uli - are you lurking?

have you tried switching the noise from Original to Improved? not sure, but I think that fixes it. (the option is on each noise node) I had this problem in my Morgue shader nodes and that switch fixed it ok.

The Noise node does not have an "improved" option.

I wouldn't worry about supporting P7 without the sr2 patch.. I know of very few people who run P7 without the updates.

If you can supply a base pz3 scene with a cube or ball with the material applied I'm sure someone can do a quick text render to see if it's resolved in the updates

It looks to me as if the bump shading is conflicting with or overriding the Phong/smooth shading, which has probably already occurred to you. 

PJ - "conflicting" is the point of it. When you want the specular reflection of a non-smooth surface, the bump takes care of it. And it works perfectly in the P4 renderer. The same sort of thing works perfectly in every other renderer I've ever used. Only Firefly has this problem.

your right,,, my bad. I used the  granite node instead.. I find that gives a better result. (I got the 2 mixed up - hence the original / improved confusion)

Yea but better is still not good enough. It was a good suggestion, but the reality is that its not the node, its Firefly. The same nodes work perfectly in the P4 renderer.

I can also demonstrate that the problem exists even WITHOUT bump. It's just much more subtle but its there all the time. It becomes especially obvious after running a cloth simulation.

Quote - 1) Turn on or off shadows

1. Turn on or off raytracing
2. Turn on or off smoothing
3. Turn on or off displacement
4. Change min shading rate (.01, .1, .2, .5, 1, nothing works)
5. Turn on or off Normals Forward
6. Any form of Post filtering.

those are all the fixes. increasing pixel samples, displacement bounds, irradiance caching, raytrace bounces, remove backfacing polygons, shadow bias, occlusion strength, refl_lite_mult et al. will likely have no effect. as there's no texture, one can't turn on texture filtering. the post filter might help to blur it out (along with everything else). one of the hidden settings at the bottom of the render list in the pz3 file might fix it, but I dunno what they do (excepting the hidden GI settings).

What I was getting at is that it appears to me as if Phong/smooth shading may not be implemented as well in Firefly as it is in the P4 renderer and isn't very compatible with bump mapping, thus the poly lines (which is probably so obvious to you that it isn't worth mentioning).

Does this happen with every combination of lighting?  I've noticed that I sometimes get poly lines REGARDLESS of material, and it may be that it's just more obvious in your example here.

Quote - I can also demonstrate that the problem exists even WITHOUT bump. It's just much more subtle but its there all the time. It becomes especially obvious after running a cloth simulation.

^^ or that.

You normally don't notice these unless you're really looking for them. I started seeing them everywhere, and learned to just ignore it and avoid it.

I just went ahead and installed SR2. I guess I really shouldn't bother testing the original P7.

Anyway, care to guess the outcome?

Yep - it's still not working right.

stupid question. would it possible to create your own noise / granite node using matmatic? one that may bypass the renderer's limitations?

also.. just a hunch. try scaling the figure / clothes 1000%.

I can use any mathematical function I want, builtin, or made from pure math nodes. As long as it is applied to bump or displacement (and it has to be in order to get the sparkly effect at oblique angles) then it does this. There is no way to hand code the bump function. It is part of Firefly's internal machinery. I have no trouble replacing the "Diffuse" calculation with one of my own and so on, but bump has no alternative as it MUST disturb the surface normal and there is no other way to do that.

My immediate response would be turning off texture filtering or changing the settings but I am not sure how well that would work since they are proceedural. Wish I could help more. In 2D printing they look like they are a moire pattern. There are some things that can counteract the effect. Maybe something might work for this as well. It could be like the stair stepped displacement that needs the other node to work smooth. Best of luck.

Have any image maps been applied to the top?  This includes specular,  displacement,  or any other.

Is this in P7 only?
I'm curious because I had encountered some similar things in p6 firefly as well when experimenting with procedural shaders.  as if they're off by a phase or the renderer takes the math too literally, or lacks wsome sort of smoothing across polygons.
sometimes I get a seam where it should smooth through - like the seam down apollos apron - well within smoothing angle, but it still shows a seam on procedurals. For example, on an image mapped texture, that seam doesn't show up.

Now I'm going to have to load it in a few different programs and see if they all do the same with raytraced procedurals...

Sorry, I'm not being much help... I'd like to see solution to this.
I don't think its in camera or lighting.

This is a wild guess, but I think perhaps the renderer is setting a temporary UV map to render the noise onto, and it's doing it per poly face in Firefly, vs. for the whole mesh for the P4 renderer.

Hi pjz99

That was my guess too.  I've seen the same effect when a uvmap template,  used as a background in PS,  was merged with the texture layers instead of deleted.

Cheers
Patorak 

Has anyone tried rendering these models in a different renderer to see if this is Poser specific?

Quote - My immediate response would be turning off texture filtering or changing the settings but I am not sure how well that would work since they are proceedural. Wish I could help more. In 2D printing they look like they are a moire pattern. There are some things that can counteract the effect. Maybe something might work for this as well. It could be like the stair stepped displacement that needs the other node to work smooth. Best of luck.

Thanks. As you answered yourself, there is no texture file involved here. Filtering does nothing.

Quote - Have any image maps been applied to the top?  This includes specular,  displacement,  or any other.

No - this is 100% procedural - no images used.

Quote - Is this in P7 only?
I'm curious because I had encountered some similar things in p6 firefly as well when experimenting with procedural shaders..

It is in all versions of Poser 5 6 or 7 all SRs. As I said in the OP, "since the dawn of time" this has been happening.

Quote - This is a wild guess, but I think perhaps the renderer is setting a temporary UV map to render the noise onto, and it's doing it per poly face in Firefly, vs. for the whole mesh for the P4 renderer.

Hmmm - interesting idea. But I don't think that's how it works. The noise is calculated on the fly during rendering on a per-pixel basis - at least I think it is.

In any case - I'm not asking why its happening. I was mostly making sure that nobody is aware of a configuration setting that fixes it. I didn't want to file a bug report only to learn that it was operator error on my part.

Quote - Has anyone tried rendering these models in a different renderer to see if this is Poser specific?

Yes - this clothing prop works fine in Modo with regard to a noise bump. That's one data point. However, Modo shows artifacts with this mesh on reflections. The reflection artifacts look just like those in Poser when you have the bias too low.

Anybody else? You LW, Vue, 3ds Max etc. users out there - does this prop produce bad results when you apply noise to the bump channel?

It is Sydney's "G2_Tank" top. in P7 Female Clothes/Poser 7 Casual

hmm. it looks facet based.. like each polygon is showing up.. I wonder.. this is your lame shader right? I just wanna try something out...

Quote - It is in all versions of Poser 5 6 or 7 all SRs. As I said in the OP, "since the dawn of time" this has been happening.

oops, I missed that part :(    Thanks for clarifying :)

Quote - Darn! Now I'm noticing it on my renders! This is a procedural displacement map in P6, though I can't remember how I rendered it and the file isn't available right now. The effect is subtle but still there now I know to look. I see you released the lamé shaders though, implying you fixed it? Quoting from the readme: "Best results obtained with raytracing and at least 2 bounces. The effect is strongly dependent on specular light, so don't use ONLy IBL. You need at least one directional light." Which of those, if any, is the fix - or is it all of them?

Perhaps you should get some glasses :biggrin: (joking)

I released the shaders as smooth surface shaders. I wanted to include the sparkly bumpy version I'm showing here, but it looked like crap so I didn't. I released the ones that worked fine (no bump) and then posted this thread to see if there was a way to fix it.

Exactly right, Khai, the anomalies line up perfectly with the underlying mesh polygons. Its not the original lame - it is the lame with the addition of some noise in the bump channel.

Screen shot shows the addition. The two nodes at the bottom do the sparkly thing.

The Noise node has been set up to produce values between negative 3 and positive 1. Then the Clamp node zeros out the negative values.

Lest anybody thinks that is the problem, it isn't. All the values are between 0 and 1 after the Clamp. I'm just doing the negative 3 so that the noise is sparse.

ok comfirmed what I thought I saw.. each square artifact is a polygon. looks like the render engine is working out the displacement per polygon and it's not matching up to it's neighbours.. not one we're gonna fix.. thats in the firefly code.

This may be waaay off and a longshot, but you know what this looks like to me... just mathematically speaking...  
Like a graphical representation of a rounding error - being that it's so rhytmic in appearance, and considering that poser works with rather small units to start with, one idea that pops into my head is that maybe it runs out of significant figures to calculate a smooth transition.
I don't hve clue how to test that theory though... but maybe it gives you some inew ideas...

this sort of gave me the idea:  http://www.softimage.com/community/xsi/discuss/archives/xsi.archive.0607/msg00749.htm
Maybe it gives you more clues then it did to me....

few more links that maybe contain clues:
http://local.wasp.uwa.edu.au/~pbourke/modelling_rendering/aliasing/

heck, here's my whole yahoo search:  http://search.yahoo.com/search;_ylt=A0oGkmh21ZNGHmEAdcFXNyoA?p=raytracing+artifacts+rounding+error+procedural&y=Search&fr=yfp-t-501

Have you tried running a diagnostic test on the mesh?  scratched polys,  self- intersecting polys,  1 or 2 point polys,  poly alignment,  etc.

ALWAYS FILE A BUG REPORT! Worst case, they know abotu it and have a fix they'll tell you about. Best case, is you uncover something that might be related to another issues, and help them resolve both issues.

Funky behaviour at polygon edges? More here.

It must be a bug. It does look like a moire pattern to me as well. I remember similar results in Bryce if the scale of the procedural texture wasn't right.

A work around is to use the Gradient Bump Channel. You'll have to fine tune the min. and max. noise values a little bit, but the checkerboard pattern is gone.

**Fiontar, you are a bona fide out-of-the-box creative thinker!
**Plugging the noise into the gradient bump channel works! I'm still going to file a bug report because the bump and displacement channels should work, but now I have a workaround.

Just as you said, I had to tweak the min and max.

The gradient bump is mathematically different and I should be using a 3-value normalized vector instead but the effect is good enough. (Plugging gray-scale noise in causes x, y, and z to always be equal, producing a bump that is always aligned with the diagonal in world space instead of the surface normal, but who cares!)

Look, look! Click to see detail.

nice work!

Looks great!

Glad I could help.

Definitely a work around that resulted from intuition unhindered by indepth knowledge. I had no reason to think it wouldn't work so it was the first thing I tried. :)

I tried to duplicate your full node settings and failed, but I can see from trying how powerful node use can be for someone who does know what they are doing. (Not in the pic I posted, that was just a basic color with the bump/gradient bump).

o.k., thx to fiontar for the gradient bump tip.

Cool, nice workaround :)

Let's hope the eventual "fix" isn't to get the two channels to be consistent by copying the Bump channel code to Gradient Bump!

The workaround isn't perfect either.

I just logged in to this thread from home on my laptop which has a very very good lcd display.

I can see the squares even on my last render. They're very faint but they are there. It's easier to see them if you stand farther back from your monitor.

Maybe that's something more exaggerated on certain monitors - taking a really close look at it here at work (HP LCD monitor, not very good at fine contrast) and it actually looks very clean.

nope, can't see them on a Samsung Synchmaster 731b... (and I tried it on all 3 of them I use)