Hi Paolo, It looks like this conversation was continued, but split into a different thread. Here's a copy of Jonathan's reply in case you missed it:
On July 15, 2011, at 18:56, Jonathan Egstad wrote: - Hide quoted text - > Well, the warning is pretty self explanatory - though you need to look at > MatrixArray.h to figure out what the indices mean. > > More than likely the Scene containing the lights hasn't had its > evaluate_lights() method called to get all the matrices up to date - this is > normally done automatically by the Render class. It can be expensive to call > so it's not done automatically in validate() - it's only done if the Render > subclass is actually going to start rendering. > > Your calculations are correct because the individual matrices in the array > are initialized to the identity, but their valid flag is false which means > they've never been explicitly set (look at the MatrixArray::MatrixEntry > structure.). > > So from the indices listed it appears neither the CAMERA_MATRIX, > PROJECTION_MATRIX, or the FORMAT_MATRIX have been set, which means only the > OBJECT_MATRIX has been. > > So if you did anything like transform a point into the light's UV space using > the MatrixArray's transform methods the result would be completely incorrect > unless the object matrix contained the complete transform - thus the warning. > > -jonathan > > On Jul 15, 2011, at 9:13 AM, Thomas Obermaier wrote: > >> Sorry, >> >> i was writing on the forum and it didn't bring the original message along in >> the mailing list (From Feb 15, 2011 by Moritz Moeller). i am refering to >> this thread here, which remained open: >> >> http://forums.thefoundry.co.uk/phpBB2/viewtopic.php?t=5283 >> >> When working with lights and transforms, get errors like these: >> >> MatrixArray::matrix(13) - Warning, matrix is unassigned >> MatrixArray::matrix(19) - Warning, matrix is unassigned >> MatrixArray::matrix(7) - Warning, matrix is unassigned >> MatrixArray::matrix(13) - Warning, matrix is unassigned >> MatrixArray::matrix(19) - Warning, matrix is unassigned >> MatrixArray::matrix(7) - Warning, matrix is unassigned >> >> Though, everything i'm calculating seems to be correct.... >> >> Thomas On Sat, Jul 23, 2011 at 11:04 PM, Paolo Berto <[email protected]> wrote: > Anyone has some tips for this one? > > > On Wed, Feb 16, 2011 at 2:01 AM, Moritz Moeller > <[email protected]> wrote: >> So I had some code that used a LightContext* to get a resp. light's >> matrix. This code always worked until and including Nuke 6.1 but now >> occasionally crashes in 6.2. >> So I replaced it with a direct call to the light itself. >> >> Looks like this >> >> _for_each ( const LightContext* l, lights ) { >> LightOp* light( l->light() ); >> >> // old code, worked until 6.1, now sometimes crashes in 6.2 >> ... l->transforms()->object_matrix().array() >> >> // new method, seems to work but causes below messages >> ... light->matrix().array(); >> >> This triggers lots of those: >> >> MatrixArray::matrix(13) - Warning, matrix is unassigned >> MatrixArray::matrix(19) - Warning, matrix is unassigned >> MatrixArray::matrix(7) - Warning, matrix is unassigned >> MatrixArray::matrix(13) - Warning, matrix is unassigned >> MatrixArray::matrix(19) - Warning, matrix is unassigned >> MatrixArray::matrix(7) - Warning, matrix is unassigned >> >> >> What is happening here? >> The light is fetched from a Scene that I create from a previously >> validate()d GeoOp. >> >> More interesting although: what is the suggested & working[tm] way of >> getting the xform of a LightOp, for rendering, in Nuke 6.2? >> >> And how is it possible that a minor Nuke version update makes working >> code instable and there is zero info about this in any NDK/Nuke release >> notes? >> >> >> >> .mm >> _______________________________________________ >> Nuke-dev mailing list >> [email protected] >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-dev >> > > > > -- > paolo berto > the /*jupiter jazz*/ group — visual research > mercenaries of jupiter jazz ltd. > www.jupiter-jazz.com > _______________________________________________ > Nuke-dev mailing list > [email protected], http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-dev > _______________________________________________ Nuke-dev mailing list [email protected], http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-dev
