View topic - Reflective cautics bug

Reflective cautics bug

Ask here your questions about development and about raytracing theory & implementation. Here you can post your patches for review.

Reflective cautics bug

Post Sat Nov 03, 2012 6:53 pm

Given:

One ring to rule them all on
One floor (not a single pixel between the floor and the ring).
One spot light (photons only)
10m photons on the sceene
Size of photon is 0.002 , Caustic mix = 5000.

Problem: light on the floor outside of the ring. The bigger is the photon size, the wider the outer glow. I've got that strange feeling, there is some sort of error causing caustics photon reflection through the reflective surface.

Yafaray-0.1.3-caustics-bug.jpg


Sign on the ring is not glowing (for the sake of the test).
Version 0.1.3

2012-11-03_223607.png
lucidlook
 
Posts: 32
Joined: Sun Jun 24, 2012 9:43 am

Re: Reflective cautics bug

Post Sun Nov 04, 2012 6:18 am

lucidlook wrote:Given:

One ring to rule them all on
One floor (not a single pixel between the floor and the ring).
One spot light (photons only)
10m photons on the sceene
Size of photon is 0.002 , Caustic mix = 5000.

Problem: light on the floor outside of the ring. The bigger is the photon size, the wider the outer glow. I've got that strange feeling, there is some sort of error causing caustics photon reflection through the reflective surface.

Yafaray-0.1.3-caustics-bug.jpg


Sign on the ring is not glowing (for the sake of the test).
Version 0.1.3

2012-11-03_223607.png


you probably are using a simple plane as floor. photons can leak through planes sometimes. to avoid that add some thickness to the floor (extrude it or use a cube scaled down in Z instead) and use a Search Radius lower than the "(thickness of the floor)"/2

"Power tends to corrupt, and absolute power corrupts absolutely. Great men are almost always bad men."

User avatar
condar
 
Posts: 771
Joined: Sun Apr 12, 2009 2:41 am

Re: Reflective cautics bug

Post Sun Nov 04, 2012 6:25 pm

Nope, the floor is at least as thick as the ring itself. But the corner of the ring which touches the flor is really thin. If I push the ring into the floor, let say, twice as much as photon radius, then glow is off.

So, I believe you are right, the caustic photos are leaking (and not sometimes, but quite significantly) through the border of the ring where it touches the floor.

P.s. the same sceene rendered in Octane or Mitsuba shows not such artefacts at all.
lucidlook
 
Posts: 32
Joined: Sun Jun 24, 2012 9:43 am

Re: Reflective cautics bug

Post Sun Nov 04, 2012 11:51 pm

lucidlook wrote:Nope, the floor is at least as thick as the ring itself. But the corner of the ring which touches the flor is really thin. If I push the ring into the floor, let say, twice as much as photon radius, then glow is off.

So, I believe you are right, the caustic photos are leaking (and not sometimes, but quite significantly) through the border of the ring where it touches the floor.

P.s. the same sceene rendered in Octane or Mitsuba shows not such artefacts at all.



a blend file would be helpful to check the scene

"Power tends to corrupt, and absolute power corrupts absolutely. Great men are almost always bad men."

User avatar
condar
 
Posts: 771
Joined: Sun Apr 12, 2009 2:41 am

Re: Reflective cautics bug

Post Mon Nov 05, 2012 8:49 am

condar wrote:a blend file would be helpful to check the scene


Ok, here it is. Just press F12. From left to right: original ring (original placement), original ring pushed into the floor, reference "cylinder" ring.

I hope it all makes sense.
Attachments
Yafaray-caustics-bug.zip
(77.59 KiB) Downloaded 375 times
lucidlook
 
Posts: 32
Joined: Sun Jun 24, 2012 9:43 am

Re: Reflective cautics bug

Post Mon Nov 05, 2012 9:39 am

I had the same problem several times.
I wish there was a button applied to an object saying "avoid photon leeks" making it easy to have a closed mesh thinner than photon radius.
This aproach would allow us to reduce the photon number and radius and therefore render times.
Arquitecto can be contacted here:
Arquitectos
Arquitecto
 
Posts: 249
Joined: Mon Jul 27, 2009 8:42 am
Location: Portugal

Re: Reflective cautics bug

Post Tue Nov 06, 2012 4:01 pm

sorry... no problem here... I'm using the last version of Yafaray (Core and Exporter) and the last svn version of Blender.
Attachments
caustics.png
settings.png

"Power tends to corrupt, and absolute power corrupts absolutely. Great men are almost always bad men."

User avatar
condar
 
Posts: 771
Joined: Sun Apr 12, 2009 2:41 am

Re: Reflective cautics bug

Post Tue Nov 06, 2012 11:40 pm

condar wrote:sorry... no problem here... I'm using the last version of Yafaray (Core and Exporter) and the last svn version of Blender.


No needs to be sorry. Just:

1. Turn off the ambient occlusion
2. Make the surface's color lighter (x2)

;-)
lucidlook
 
Posts: 32
Joined: Sun Jun 24, 2012 9:43 am

Re: Reflective cautics bug

Post Wed Nov 07, 2012 12:58 am

lucidlook wrote:
condar wrote:sorry... no problem here... I'm using the last version of Yafaray (Core and Exporter) and the last svn version of Blender.


No needs to be sorry. Just:

1. Turn off the ambient occlusion
2. Make the surface's color lighter (x2)

;-)


still nothing... :?
Attachments
caustic2.png

"Power tends to corrupt, and absolute power corrupts absolutely. Great men are almost always bad men."

User avatar
condar
 
Posts: 771
Joined: Sun Apr 12, 2009 2:41 am

Re: Reflective cautics bug

Post Wed Nov 07, 2012 8:43 am

condar wrote:still nothing... :?


Are you kidding?..

2012-11-07_094031.png
lucidlook
 
Posts: 32
Joined: Sun Jun 24, 2012 9:43 am

Re: Reflective cautics bug

Post Fri Nov 09, 2012 5:30 pm

OK. I got it :)

rendering with SPPM the 'Leak" is more clear. I think it is not a bug but an expected behaviour for the algorithm. in the "contact" area of the two meshes (ring and floor) the thickness of the Closed mesh is in theory 0, which of course is lower than Caustic Search Radius. :)

making the bottom part of the ring thicker shows no problem. So the issue here is how a bias algorithm (Photon Mapping, irradiance Cache, etc) interpolate the solution.

I think it is not a bug but that's only my opinion ;). Even so you should fill a bug report; maybe devs can find a better approach to this issue.
Attachments
caustics.png

"Power tends to corrupt, and absolute power corrupts absolutely. Great men are almost always bad men."

User avatar
condar
 
Posts: 771
Joined: Sun Apr 12, 2009 2:41 am

Re: Reflective cautics bug

Post Sat Nov 10, 2012 8:58 pm

Agreed 100%. The only question which remains for me is -- how I can fill the bug report if the forum doesn't allow me to add anything in here http://www.yafaray.org/development/bugtracker/yafaray ?
lucidlook
 
Posts: 32
Joined: Sun Jun 24, 2012 9:43 am

Re: Reflective cautics bug

Post Sat Nov 10, 2012 11:02 pm

lucidlook wrote:Agreed 100%. The only question which remains for me is -- how I can fill the bug report if the forum doesn't allow me to add anything in here http://www.yafaray.org/development/bugtracker/yafaray ?


the link "Create New Issue"
Attachments
bug_report.png

"Power tends to corrupt, and absolute power corrupts absolutely. Great men are almost always bad men."

User avatar
condar
 
Posts: 771
Joined: Sun Apr 12, 2009 2:41 am

Re: Reflective cautics bug

Post Sun Nov 11, 2012 1:37 pm

condar wrote:the link "Create New Issue"


Thanks!

http://www.yafaray.org/node/550

/Blind me.
lucidlook
 
Posts: 32
Joined: Sun Jun 24, 2012 9:43 am

Re: Reflective cautics bug

Post Sun Apr 26, 2015 8:07 am

Hello,

Please see my answer to this in the bug tracker entry:

http://www.yafaray.org/node/550#comment-1003

I hope it makes sense to you.
User avatar
David Bluecame
 
Posts: 418
Joined: Mon Jan 21, 2013 12:42 pm
Location: Spain


Return to Developers' Corner



Who is online

Users browsing this forum: No registered users and 2 guests

cron