Just a short mention, in case anyones else runs into this: as it appears, some of the problem P6 had with rendering reflections and refractions have been solved, but bump maps still aren't handled correctly. I tried a recent P5 scene in P6 and got very noisy reflections and refractions, but only for those parts with bump maps applied. Disabling the bump maps apparently solved the problem, and it didn't occur with displacement maps, either. Oh, and increasing the sample rate didn't seem to help. I haven't yet tried playing with the shading rate though, which was already at 0.2, or the shadow bias. At the moment, changing bumps into displacements seems to be fair enough as a workaround for me.

-- I'm not mad at you, just Westphalian.

I am still having problems with the refractions. This is in Poser 6 with the SR1 installed.

This is the exact same file rendered in Poser 5. (image was cropped, forgot to do it to the first image) Notice the odd black pattern at the top of the lens in the Poser 6 render.

Message edited on: 05/27/2005 01:28

Still Poser 6 is better than it was before, this is how it used to render refractions. At least all the odd interferance patters are gone.

Thanks for the examples. Looks pretty nasty there. I've noticed that the sky shines through the hair in the P6 SR1 example, so you must have turned the backside suppression feature, or whatever it's called, on. Could that be causing the refraction problem?

-- I'm not mad at you, just Westphalian.

You mean "Remove backfacing polys"? Okay, I unclicked that, wasn't the problem. I tried moving the lens and it seems to be affecting the eyebrow. It looks like Poser 6 is having problems dealing with transparent ojects.

Firefly always had a problem with scale in poser. There's a thread by ronstuff explaining how to parent all your scene to a box and scaling the box 1000%. This fixed a LOT of issues for raytrace artifacting. Scaling the box will not scale bumpmaps so, you'll have to do it manually (0.01 becomes 0.10) at 1000%. I'm of course, guessing here. P6 could have new probs.

I don't have Poser 6, but does it have a Raytrace Bounces option in the render settings? Try increasing it.

Your refraction render looks like a problem I used to have in Cinema 4D when rendering Poser hair if I had the Ray Depth set too low. It would have black areas because the ray was passing through more hair layers than the Ray Depth setting.

Since V3 has a transparent Eyebrow section that would add 1 or 2 extra layers the ray has to pass through.

I hope that makes sense.

Message edited on: 05/27/2005 10:04

The problem happens when not using bump maps at all. Here, a small solid sphere is just poking out of a big checkered one behind a refracting lens object (left). Changing the small sphere to be totally transparent we see artifacts in the P6 render (center) but not in the P5 render (right). The artifacts occur in the region where the transparent object intersects another object. Notice the checker pattern appears in the artifact. (Perhaps OT, I remember Bryce having difficulty rendering transparent object intersections). P5 seemed to have this licked. So, what new code in P6 having to do with objects near each other, angles of objects near each other, intersecting objects, that might be causing this? Ambient occlusion, crease thresholding, something else entirely?

Message edited on: 05/27/2005 10:06

That's a very instructive example. Thanks, Jeff01! Incidentally, does anyone know how refraction is computed in Firefly? I mean, is it physically correct in the sense that both the front and back face change the direction of the light? If so, what if an object only had a front face? I just had that crazy thought that maybe the transparent white sphere might be taken as an object with a different refractive index. I don't think, however, that the problem with transparent objects is necessarily related to the bump map problem. I will try the scaling approach richardson proposed to see if it solves that issue and post my findings here.

-- I'm not mad at you, just Westphalian.

Attached Link: http://www.renderosity.com/messages.ez?Form.ShowMessage=1870874

Scaling is a must for closeups and object proximity problems using RayTrace. It fixes a lot of things. In P5 anyway. There's a glitch when you parent your primary camera to the box. Just correct the "shift" and parent again. Save the 1000%pz3 so you won't have to bother again. And use hierarchy editor to drag and drop onto the box for speed.

Now this is a riot! I reran my test above, this time parenting everything to an independent sphere (not visible) and scaling the sphere to 1000%. I moved the camera back to reframe the shot and rendered. The artifacts near the intersection of the small transparent sphere and checkered sphere are still there but not as prevalent due to the change in scale. They seem limited to the proximity of the intersection. But look what else showed up. The same artifacts that I had before SR1. See those whitish discs around the top and down the left edge of the lens?

J, Remember to scale with a box as it scales differently than a sphere. The ball probably threw off all your cams. Some node settings get left behind. Also distant start/end. Infinites do not parent easily so you may have a new lightset here. The new spheres are ...??? I dunno. Looks like a "glossy" node came to life. These are ronstuff questions.

Incidentally, here is what I get when refracting a bump mapped surface in Poser 6 with SR1 applied. The scaling trick, unfortunately, doesn't help. Despite of what I wrote before, changing from bump to displacement maps doesn't seem to work either. There's a new 'ray bias' option in the refraction node as of SR1, which I've played around with, as with the refraction quality, all to no effect.

-- I'm not mad at you, just Westphalian.

LOL @odf!!! Hollywood is calling

Yeah, it's called the "instant monster" effect and it's of course a feature, not a bug.

-- I'm not mad at you, just Westphalian.

Richardson, I scaled up with a cube and got the same results (not shown). I did use the default infinite lights which Poser doesn't allow to parent to the cube.

I rendered the scaled up (1000%) pz3 in P5 and got similar refraction disc artifacts, but spotted (a clue?). However the transparent ball intersection with the checkered ball had no artifacts, scaled up in P5 (not shown here).

odf, Curious about the bump map thing, I applied a Turbulence node to the bump node of the checkered ball in P6, unscaled. The refracted bump looked OK but the intersection artifacts were still there (not shown). I added James HiRes with bump map in front of the checkered ball and behind the lens (see the turbulence bump on the ball). I disconnected the bump map for the Head (so I could examine the eyes).

The refraction bump map artifacts are there, just like odf's renders, for the body. Are bitmap bump maps a culprit but not procedural bump maps? Also look at the eyes. The right eye is all of a sudden looking way left and penetrates the lower lid. Is that the pupil still looking straight? The left eye looks straighter, but also penetrates the lower lid,and there's some horizontal artifact cutting across the lower left iris.

This gets curiouser and curiouser.

Curiouser Labs. Catchy. Bitmaps sound like a good starting point. Any chance to put a V3 behind the evil sphere? I'm too lazy to set up a scene...lol

Hmm, this all reminds the of the "Buffy" episode I just watched: the invisible girl (=lens) setting out to destroy the face of the beauty queen (=James). Better not risk putting V3 in danger here, or something dreadful might happen. Alas, there was no mentioning of bitchmaps.

-- I'm not mad at you, just Westphalian.

Reminds me, of course.

-- I'm not mad at you, just Westphalian.

In my James render above, I think the horizontal artifact cutting across the lower left iris is actually the top edge of the lower lid.

I think James eye weirdness behind the refraction lens is related to the artifacts we see on transparent objects behind the lens when they're in close proximity to other objects, as when intersecting. I think the James eyeball object is actually two spheres, an outer and an inner, very close to each other. The entire outer sphere is assigned the EyeTrans material and is totally transparent, used for specular highlights. While not intersecting per se, the entire outer eyeball sphere is close enough to the inner eyeball sphere to suffer this artifact. Similarly, the lower eyelash is transparent, close to the lower lid, and also exhibits the artifact, making the eyeball appear to penetrate the lower lid.

Message edited on: 05/28/2005 10:21