Hi Foundry,

Since starting this thread back in Nov. I personally haven't ran into any
major issues with the add_layer problem mentioned, until recently.

The Foundry says that this is a "third party" issue but under similar
conditions (creating new layers that make their way into scripts, gizmos,
comps), this could possibly be recreated at any facility.

With the attached files, I have been able to reproduce this bug but haven't
narrowed it down to the catalyst.  It could be the redguard1.glow or the
rgba.water channels or a combination of these channels when mixed together
(e.g. channels existing in compA and importing/cutting&pasting another comp
with similar channels).

Just CC'ing the thread contributors so ya'll know what's up.

BTW - I recently heard from a buddy at Imageworks that this issue has
brought some renders down!

Best,

Dan




On Fri, Nov 25, 2011 at 7:33 AM, Diogo Girondi <[email protected]>wrote:

> I have my doubts about messing with this stuff. From what I see is more of
> a matter of people paying more attention to what they author/share and
> treat as project/pipeline independent than a actual problem.
>
> Implementing ways to control/restrict this stuff via user interaction
> could in fact end up severely crippling certain workflows. Like not
> allowing gizmos to create custom channels/layers on a more invisible way to
> artists without prompting them for action.
>
> However I do agree that Nuke should deal with channels/layers conflicts
> more elegantly, and could benefit from having a few ways to track down
> channel/layer creation/usage. A Layer Info window that lists all the layers
> present in a script with their usage and creation points would suffice in
> my opinion.
>
> A window with something like:
>
> Layer name, channels, used by x number of nodes, created at (node
> name/init)
>
>
>
> As for allowing the addition of custom channels to the RGBA layer I agree
> that this is one layer that shouldn't allow that. Same would be true for
> the RGB pseudo-layer.
>
> Eight channels per layer is another thing that I find very questionable
> since its something you can never access at once in Nuke and can make your
> life trickier layer wise.
>
>
>
> Cheers,
> Diogo
>
>
> On 25/11/2011, at 10:27, chris <[email protected]> wrote:
>
> > hmm, i can see that is kinda hard to solve this issue without crippling
> functionality... however, corrupted scripts are no fun either.
> >
> > it seems to me that it adding unnecessary channels could greatly reduced
> by some confirmation/warnings boxes...like:
> >
> > if you paste some code into a script that will add new channels, pop up
> a dialog panel showing which channels will be added, with a "yes" "cancel"
> "strip channels" button
> >
> > and similarly, if you export a gizmo with custom channels, show a dialog
> with all the channels listed.
> >
> > btw, is there a good reason why rbga.customLayers should be even
> allowed? (as those seem to often cause troubles)
> >
> > i don't often work with a lot of channels, so maybe this options are not
> really practical for other typical workloads. one could always add an
> option in the preference to turn those warnings off i guess.
> >
> > just some thoughts
> > ++ chris
> >
> > ps: i also like the idea of feature 15561
> > - Don't generate channel adds at save when the channels are not
> referenced in the script.
> >
> >
> >
> >
> > On 11/24/11 at 10:32 PM, [email protected] (Dennis
> Steinschulte) wrote:
> >
> >> I got an answer from the foundry support team. It's not a
> >> bug and they can't see a solver for this issue because it
> >> seems to be a third party plugin thing. They have read
> >> this mailing list about this topic and will think about
> >> the creation of channels, though. I don't have the mail at
> >> hand now, but I'll post post the whole answer (incl bugID)
> >> in about 12 hours. Cheers Dennis
> >>
> >>
> >> On 24.11.2011, at 21:03, "[email protected]" <
> [email protected]> wrote:
> >>
> >>> I'm still not sure that this breaks anything, but it
> >> might... My main
> >>> thing though is that we should have the ability to
> >> remove/rename
> >>> channels from the script through python, for example (not
> >> just add
> >>> them, like it works now).
> >
> > _______________________________________________
> > 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
>
::::::::::::::::::  TESTING AREA FOR ADD_LAYER BUG  ::::::::::::::::::


1.) NUKE -> FILE -> OPEN -> (before_import_v001.nk)
2.) OPEN ANY OF THE RED SHUFFLE NODES
3.) NOTICE THAT THE SHUFFLE NODES (in 2) IS SET TO (rgba)
4.) FILE -> IMPORT SCRIPT... -> (unpremult_node_BAD.nk)
5.) FILE -> SAVE AS... -> (after_import_v001.nk)
6.) FILE -> CLOSE

7.) FILE -> OPEN... -> (after_import_v001.nk)
8.) OPEN ANY OF THE RED SHUFFLE NODES
9.) NOTICE THAT THE SHUFFLE NODES (in 2) IS SET TO (rgb) ALONG WITH OTHER 
        SHUFFLE NODES IN THE COMP.

===========================================================

THE CULPRIT OF THIS PROBLEM IS THE ADD_LAYERS DEFINED IN THE IMPORTED SCRIPT 
(unpremult_node_BAD.nk).

IF THIS NODE WAS CUT AND PASTED OR IMPORTED INTO THIS COMP THE ADD_LAYERS 
DEFINITIONS 
BECOME PART THE COMP SCRIPT AS WELL AND BREAKS THE COMP UPON SAVING.

NOW TRY COMMENTING OUT THE (add_layer) in the (unpremult_node_BAD.nk) FILE
AND RUN THROUGH THE STEPS AGAIN. YOU'LL NOTICE THAT THE ISSUE NO LONGER EXISTS.

add_layer {rgb rgb.UV}
add_layer {rgba redguard1.glow rgba.water}

===========================================================

Attachment: before_import_v001.nk
Description: Binary data

Attachment: unpremult_node_BAD.nk
Description: Binary data

_______________________________________________
Nuke-users mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users

Reply via email to