#299 Emissive image masked texture + scattering = freeze
I have a procedural texture that uses an image to mask parts of the surface as transparent. The opaque areas of the surface emit light. I put this texture on a plane and put the plane in a fog cube. When I try to render with photon mapped scattering AoI sits at 100% CPU utilization forever (it seems). I didn't actually wait till infinity was up, but I decreased the photon count to 1 and noticed that it still sat at 100% for longer than I was willing to wait. I'm pretty sure it was frozen in an infinite loop of some sort. Just for kicks I tried swapping out the image module for a noise module and it worked fine. This might relate to the image module bug I recently posted (uses the same test scene, linked below). I put the scene I'm using at:
http://icrashedtheinternet.com/scatteringTestScene3.aoi
-John
Logged In: YES
user_id=74743
Originator: NO
I can't reproduce this. Does it still happen in 2.5? (If so, it's also possible it was fixed by one of the changes I've made for 2.5.1, so test it again when the beta of that is released.)
Logged In: YES
user_id=1078421
Originator: YES
The bug has mutated in 2.5, but it's still there. Might be fixed as a result of the other bug fix you eluded to that will be fixed in 2.51.
To show this bug, remember that you first have to follow the workaround from a different bug, posted at: https://sourceforge.net/tracker/?func=detail&atid=115063&aid=1673372&group_id=15063
This will get the image mapped emitter to show up in the scene, which is a requirement for this bug. In 2.5, instead of freezing with 100% cpu it freezes with 0% and throws this exception:
Exception in thread "Raytracer main thread" java.lang.NullPointerException
at artofillusion.raytracer.Raytracer$6.execute(Raytracer.java:1100)
at artofillusion.util.ThreadManager.run(ThreadManager.java:148)
at artofillusion.raytracer.Raytracer.run(Raytracer.java:1113)
at java.lang.Thread.run(Unknown Source)
Let me know if there is anything else you need.
Thanks a ton for working on these bugs. :)
-John
Logged In: YES
user_id=1078421
Originator: YES
This bug is still alive in 2.5.1 beta. But it's much easier to reproduce. Just put an emissive sphere inside a fog cube with no surface (so you can see in) and render with photon mapped scattering. It'll freeze with 0% cpu and throw this:
Exception in thread "Raytracer main thread" java.lang.NullPointerException
at artofillusion.raytracer.Raytracer$6.execute(Raytracer.java:1102)
at artofillusion.util.ThreadManager.run(ThreadManager.java:148)
at artofillusion.raytracer.Raytracer.run(Raytracer.java:1115)
at java.lang.Thread.run(Unknown Source)
File Added: fogScatteringTest.aoi
Logged In: YES
user_id=1078421
Originator: YES
This bug is still alive in 2.5.1 beta. But it's much easier to reproduce. Just put an emissive sphere inside a fog cube with no surface (so you can see in) and render with photon mapped scattering. It'll freeze with 0% cpu and throw this:
Exception in thread "Raytracer main thread" java.lang.NullPointerException
at artofillusion.raytracer.Raytracer$6.execute(Raytracer.java:1102)
at artofillusion.util.ThreadManager.run(ThreadManager.java:148)
at artofillusion.raytracer.Raytracer.run(Raytracer.java:1115)
at java.lang.Thread.run(Unknown Source)
File Added: fogScatteringTest.aoi
Logged In: YES
user_id=74743
Originator: NO
This scene renders without any problem for me in 2.5.1 beta (all default render settings except that scattering is set to "Photon Mapping"). Also, the stack trace doesn't look like it has anything to do with photon mapping. It looks more like a threading issue, where a render is getting canceled unexpectedly. How exactly are you rendering it? Are you using the standard raytracer or AR?
Logged In: YES
user_id=1078421
Originator: YES
You are using "fogScatteringTest.aoi" right? That's the one I used most recently. I'm using the standard renderer (not the advanced one). All defaults except for photon mapped scattering is turned on. And it doesn't render a single pixel ever. It just stops at the exception.
The exact same exception was thrown on a different bug that anderswc posted, maybe it's the same bug? Anderswc seems to think it might be a cross platform bug. The other bug is at https://sourceforge.net/tracker/index.php?func=detail&aid=1808367&group_id=15063&atid=115063
If you need the help from a Windows user to debug this, let me know if I can do anything to help. It's a total show stopper for scattering on my machine. :(
Digging through old bugs again: Does this scene still give you issues? As of today, Seems to work properly in [r417] (3.0.1 + a few commits)
Related
Commit: [r417]
I can't repro on the latest release, 3.0.1, as far as I can interpret my old report here. :)
In that case, I'm closing. No repro.