Possible Reflect/Refract Node Bug ? Anybody Else See It ?



  • Can other people try this on other versions of Poser please.

    Very simple setup - just the Poser GROUND, a Poser One-Sided Square, and an infinite light. All material/render settings and my PP2014 render are shown in attached image.

    The problem is the green line that appears at the intersection of the two planes. It shouldn't be there.

    (I've posted this same observation at several other sites as "PP2014 Refract Node causing white line where surfaces intersect" , but haven't yet received any feedback as to whether anybody else sees it on other versions. )

    Image from Renderosity forum



  • If I add a second 'wall' of a different colour and rotate it slightly the problem becomes clearer - you can see exactly what's happening

    Two walls test



  • It's worth zooming in on the render in an image editor to be sure what we see (if you zoom in with the Poser camera the line seems to remain about single pixel wide)...

    alt text



  • The same effect seems to occur with the Reflect node too.


  • Poser Ambassadors

    What happens when you set the ray bias closer to 0?



  • Changing Refract node RayBias (default 0.1) up or down makes no difference at all (details here http://www.cgbytes.com/community/forums.aspx?g=posts&m=125617#post125617 )


  • Poser Ambassadors

    Why did you say default .1? In your screen shot it's .85



  • At the point where I originally checked RayBias (the linked post) I was still using a more complex shader (one of your bb_ocean ones) which had RayBias set to 0.1 - the statement was cut-and-pasted from that post.
    But I did double check setting RayBias to 0.0 with the ground/one-sided plane/infinite before making that last post here.


  • Poser Ambassadors

    @3dcheapskate said in Possible Reflect/Refract Node Bug ? Anybody Else See It ?:

    At the point where I originally checked RayBias (the linked post) I was still using a more complex shader (one of your bb_ocean ones) which had RayBias set to 0.1 - the statement was cut-and-pasted from that post.
    But I did double check setting RayBias to 0.0 with the ground/one-sided plane/infinite before making that last post here.

    You can reduce it by increasing pixel samples, but it does not make it go away
    i think it has to do with the refract node bending light - but i do not know enough to say that it should provide this result



  • @wimvdb - I'm fairly certain that the result is incorrect. Look at the colour of the line - it changes depending on what's behind the vertical surface that the horizontal refracting plane intersects. The vertical surfaces should be totally opaque. No amount of refraction/reflection should affect that.

    @bagginsbill - I've just retried the test with the additional blue wall in Poser 6, 9 and PP2014. I used automatic render settings at max quality. Setting Refract node RayBias to zero doesn't change the result (although a couple of times when I did that in Poser 6 the horizontal plane rendered totally black, and there was no line)

    I've also tried plugging the Refract node into the Refraction_Colour instead of Alternate_Diffuse. Same results.



  • This also happens with Poser 11 Firefly. I only tested with two single-sided squares, for the wall and the refractive surface.

    My workaround was to add a full scene background mesh (e.g. EnvSphere) with shading similar to the wall - this made the line (still there) almost unnoticeable, unless you were really looking for it. But this wouldn't work so well if the wall has highly contrasting shades/colours.



  • @chriswwd Thanks - I'll report this as a bug then



  • Bug reported - case ID 1304127



  • Why is it so difficult to get a straight answer ? The responses I've had from customer support are confusing me...

    The initial response (1 Jun) was: "The Engineers think they have fixed this for a future update. It was unrelated to the older issue you referenced. I am adding your Project File as a Test for them to use to confirm the fix."

    I ignored the fact that I'd never referenced any 'old issue' and just asked for some clarification - "Just to be sure I understand correctly - are you saying that the engineers think the symptoms I reported will be fixed as a sort of knock-on effect from another change they already have planned?

    On 2 Jun I got a similar response to the first: "They have listed the report as fixed. The issue was not the same as the previous issue that was fixed in the SR3 Update. This issue was different and has been addressed for a future update."

    I again asked for clarification - "Sorry, but that answer has just confused me even more :oS ! How can both (1) "the report be listed as fixed" and (2) "the issue be addressed for a future update"? Those seem to be contradictory statements.
    I'm also confused by your reference to "the previous issue that was fixed in the SR3 Update" and the statement in your previous response that "It was unrelated to the older issue you referenced", since I have not made any reference to any older issue

    On 3 Jun a similar response again: "The issue is fixed. It had no relation to the old issue that you had referenced originally. I wanted to confirm that your issue you reported this time is also fixed, since I previously had stated it was related to the old issue."

    So I've just asked again - "1) What 'old issue' are you talking about ? (Please provide a case ID or URL as I have absolutely no idea what you are referring to) 2) If 'the issue is fixed' - which versions of Poser incorporate the fix ?

    (P.S. The text formatting on this forum seems to be as bad as the Renderosity forums... and that's bad!)



  • I've had a better answer now - it wasn't me who referenced an older issue, but SM support/techies. The bug I reported was originally thought to be related to another older bug, but turned out not to be. The bug I reported has now been fixed and will be in the next update.

    Nearly there...

    So I've asked again whether that just means fixed in Poser 11, or whether PP2014 (and even Poser 9 might also get the fix). I'm only on PP2014.



  • The bug's apparently only going to be fixed in Poser 11 and Poser 11 Pro.

    Disappointed, but no real surprise.



  • The problem reported in the 'PP2014 Refraction Problem With Planes In Close Proximity' is almost certainly related to this one.