sounds very similar to a bug (#15587) I reported about 10 months ago I never had problems with imagedata switching eyes, in my case it was the evaluation of expressions that got mixed up between eyes and also only on the first frame of a render bucket
my resolution was also to use a dedicated write node for each eye. the oneview is not necessary as long as you do not have any write node in the script that is set to more than one eye, at least in my experience Best regards, Michael ------------------------------------------ DI (FH) Michael Habenicht compositing - vfx :: motiongraphics :: dvd http://www.tinitron.de [email protected] ** Digital Compositor & TD TRIXTER Film Munich http://www.trixter.de ------------------------------------------ ----- Original Message ----- From: [email protected] To: [email protected] Date: 27.09.2011 17:37:47 Subject: Re: [Nuke-users] Re: I got a weird one here guys.... > OK! > > So, it is definitely a bug and it has to do with STEREO. > > This only happens on the RIGHT eye and again, it only happens when viewing a > Write or the output of a write that is an EXR. > > Heres the bug: > > When my Write was set to "left,right", nuke was SOMETIMES grabbing the LEFT > eye layers for some of its calculations on the RIGHT eye up the pipe. For > instance, the neg values I was seeing were coming from subtracting out the > layer "specular", grading, and adding back to the "rgb", but nuke would grab > the LEFT eye "specular" and minus it from the RIGHT eye "rgb" resulting in > the neg values. Nuke would also sometimes grab the LEFT "depth.Z" to apply a > Zblur to the RIGHT eye. > > Also of note, when I would send this job to the farm in blocks, only the > FIRST FRAME of the block would have this error. If I submitted it to the > farm on 1's, EVERY FRAME had this error. > > So here's all the Write node scenarios I tried that didn't work: > 1) One Write node set to "left,right", stereo EXR output (both eyes in > single file) > 2) One Write node set to "left,right", but write to 2 files using the > filename_%v convention > 3) Two Write nodes, one set to "left" and one to "right" but NOT having a > OneView node before the Writes. > > So, the only work-around that worked was to OneView out each eye and after > that have 2 Writes. > > [image: image.png] > > -ak > > > > > On Mon, Sep 26, 2011 at 11:47 AM, Farhad Mohasseb wrote: > > > This sounds like a cache issue more than anything else....have you tried > > emptying the cache and the buffer? Do you still get the same thing? If you > > change the viewer lut do you still get it or does it go away until you go > > back to the last used lut? > > On Sep 26, 2011 11:28 AM, "Anthony Kramer" > > wrote: > > > Its not that the neg values are being retained by the write node, its > > that > > > the negative values APPEAR when viewing the write node. > > > > > > This morning I open up the same comp and the issues were gone for about > > an > > > hour, and then they came back. Deleted the viewer node, and it went away > > > again. I'm pretty sure this is a bug of some sort but unfortunately I > > > cant consistently reproduce it at this point. I'll keep you all posted. > > > > > > -ak > > > > > > > > > On Mon, Sep 26, 2011 at 10:47 AM, Mason wrote: > > > > > >> 32 bit Floating Point EXR permits negative values. Sounds like you are > > >> getting these values from your comp and EXR is not clipping those > > values. > > >> > > >> -Mason > > >> > > >> > > >> On Sep 26, 2011, at 4:24 PM, Anthony Kramer > > >> wrote: > > >> > > >> Sorry, yes, I should have mentioned that. When I write out the exr and > > read > > >> it back, it still has the errors I see when just viewing the write node. > > >> > > >> -ak > > >> > > >> On Sat, Sep 24, 2011 at 4:22 PM, Sean Looper < > > >> [email protected]> wrote: > > >> > > >>> The Write node passthrough does make certain assumptions when your > > output > > >>> is exr. Try writing the file out and compare the written frames to your > > last > > >>> node. > > >>> > > >>> -sean > > >>> > > >>> Sent from my cellular device > > >>> > > >>> On Sep 24, 2011, at 2:16 PM, "Anthony Kramer" > > >>> [email protected]> wrote: > > >>> > > >>> NukeX 6.3v2 > > >>> Windows 7 > > >>> > > >>> On Sat, Sep 24, 2011 at 2:11 PM, Anthony Kramer > > > > >>> [email protected]> wrote: > > >>> > > >>>> It gets worse.. > > >>>> > > >>>> Seems like it's also calculating a few other key operations in my comp > > >>>> differently when selecting EXR as the output format in a write. > > Besides the > > >>>> neg values below which seem to be coming from a grade, it also makes > > my > > >>>> Zblur and LensDistortion look different. > > >>>> > > >>>> If I disable those 3 nodes, I get no difference between the last node > > and > > >>>> the write. If I enable ANY ONE of those nodes I get a slight > > difference in > > >>>> their calculation when viewing the write node only. > > >>>> > > >>>> Bizarre. Serves me right for working on a Saturday. > > >>>> > > >>>> -ak > > >>>> > > >>>> > > >>>> On Sat, Sep 24, 2011 at 1:06 PM, Anthony Kramer > > > > >>>> [email protected]> wrote: > > >>>> > > >>>>> So I have a fairly simple comp here and at the end when it goes to > > the > > >>>>> write node, all of a sudden I get negative values. Take a look at the > > screen > > >>>>> caps below and let me know if you seen anything like this. > > >>>>> > > >>>>> If I switch the write node to dpx/Cineon the problem goes away. If I > > >>>>> switch to exr/Cineon the problem stays so it seems to be something > > with the > > >>>>> filetype being exr, not the colorspace. > > >>>>> > > >>>>> -ak > > >>>>> > > >>>>> > > >>>>> > > >>>> > > >>>> > > >>> _______________________________________________ > > >>> Nuke-users mailing list > > >>> > > [email protected], > > >>> > > > > >>> http://forums.thefoundry.co.uk/ > > >>> > > >>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > >>> > > >>> > > >>> _______________________________________________ > > >>> Nuke-users mailing list > > >>> > > [email protected], > > >>> http://forums.thefoundry.co.uk/ > > > > >>> > > >>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > >>> > > >> > > >> _______________________________________________ > > >> Nuke-users mailing list > > >> [email protected], > > >> http://forums.thefoundry.co.uk/ > > >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > >> > > >> > > >> _______________________________________________ > > >> Nuke-users mailing list > > >> [email protected], http://forums.thefoundry.co.uk/ > > >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > >> > > > > _______________________________________________ > > Nuke-users mailing list > > [email protected], http://forums.thefoundry.co.uk/ > > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > > _______________________________________________ > Nuke-users mailing list > [email protected], http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users _______________________________________________ Nuke-users mailing list [email protected], http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
