I was playing with nodes (well, my gf is away for a few days) and after many renders that were fine and dandy, suddenly a vertical line appeared in the reflections of both of these cylinders (arrowed). Not the cylinders themselves, just the reflections. I reloaded the floor texture, deleted and reloaded one of the cylinders, saved the scene and restarted Poser... etc. etc. No change.

Then I tried rotating one of the cylinders, and it disappeared, as shown in the inset render.

Does anyone have a clue what causes this? This is the standard Poser cylinder, scaled up by 300%. I use them a lot, but have never seen this before.

(edit) rotating by only one degree gets rid of the line, but it returns at zero rotation.

I believe there was a discussion on the the forum here somewhere about the bad UV mapping of Poser's primitive cylinder. This could be causing the appearance of your thin little line.

Oh, and btw, could you tell us your light set up and render settings? That is a very, very nice sharp and clean render you have there.

Thanks, luv! :O)

That's probably it. Very strange that it should suddenly appear in the middle of several test renders.

Lights and render settings? Well, nothing special. I'll just reload the test scene and take a screenshot or two. Back in a few...

Lights:

The most important point to note for the lighting setup is that I am using gamma correction. If you don't have Poser Pro, you will need to do this in your material shaders.

Normally for testing I would use only one IBL and one infinite. However, I left a third light switched on by mistake.

Diffuse IBL for general ambient light - white, intensity 40%, which seems very high but I use a fairly dark image "Attic.jpg" that comes with Poser. I like this IBL image and use it nearly all the time because it has good directional contrast without very strong colours. IBL contrast is set to 1.0 (in the Material Room). The default is 3.0, and this is too high - it creates too fast a transition from light to shadow. I have AO switched on in this light, strength 0.5, bias 0.03 (these surfaces are nice and flat, so a low value is possible without artifacts from bump etc), max distance 36inches, samples 8. With higher intensity on the IBL, I found I need to turn down the AO strength, or decrease the max distance, otherwise it becomes too strong. These days I'm tending to use material-based AO in my proper renders, but IBL AO is convenient and good enough for test scenes.

Infinite light: white diffuse+specular at 20% intensity, raytrace shadows with 2.5 blur radius, positioned behind, above and left of the camera.

(Normally I would have the IBL at lower intensity (10 to 20%) and the infinite higher, but I wanted brighter all round lighting in this case.)

The third light that I meant to switch off is a point light positioned between and just above the spheres. It has bagginsbill's inverse square falloff shader in it, which combines an intensity value with a distance from the light at which this intensity occurs. Mine is set to 200% at 72 inches. The intensity at the actual light source itself is in theory infinitely high. Its full effect is not seen here as the other lighting drowns it out rather.

The floor and wall are both procedural shaders by bagginsbill. The funky one on the wall is actually one of his Orb shaders!

Well... I hope these are useful details. I'm going to bed now, but I'll answer anything else you want to know tomorrow. Nite nite!

Oh, as a matter of interest, the five small spheres are a light meter. The white highlight on the larger central sphere indicates that there is too much light in the scene at that point. That's because the middle sphere is positioned just underneath the point light.

Thanks! I've copied your description out for study. :)

any reason for setting IC=0?  this maximises certain bucket-edge-related calculational errors in FFRender.  I dunno if IC is covered in great detail in manual.

Miss Nancy... these are just my settings for the fast test renders I was doing. I wasn't aiming for top quality. That's the only reason I was using IC = 0.   :O)

There's a lot of stuff that isn't covered in any useful detail in the manual. I was trying to find out about math node bias and gain functions last night, in order to understand what some of bb's shaders were doing. This is what the manual tells me:

Bias: adjusts the bias of Value_1 by Value_2.
Gain: adjusts the gain of Value_1 by Value_2.

Wow. That's really good to know!

Bias This calculates Ken Perlin's Bias function. Given an input value between 0 and 1, it calculates an output value which is also between 0 and 1 according to: y(x) = x^(log(B)/log(0.5)) where the input value x and bias B correspond to Value_1 and Value_2 on the Math node. If B=0.5, then y(x)=x. Values of B less than 0.5 push the output toward smaller values, while values of B greater than 0.5 push the output toward larger values. An interesting artifact of this function's definition is that Bias(A, B) is the same as Bias(B, A).

Gain This calculates Ken Perlin's Gain function. Given an input value between 0 and 1, it calculates an output value which is also between 0 and 1 according to: y(x) = Bias(2x, 1-G)/2 if x<0.5 1-Bias(2-2x, 1-G)/2 if x>0.5 where the input value x and gain G correspond to Value_1 and Value_2 on the node, and Bias(x, B) is the Bias function described above. If G=0.5, then y(x)=x. Values of G less than 0.5 smooth the input by pushing the output toward 0.5, while values of G greater than 0.5 sharpen the input by pushing the output toward 0 or 1, away from .5.

Thank you, bb. Where do you get details like this?
(Please, please don't tell me "My head".)

Google