Maybe there should be a clamp option in 'merge3'. I think divide has similar issues. That would allow for backward compatibility to 'merge2'.
Clamping for matte purposes is valid isn't it? Howard > On 28 Jan 2015, at 7:53 pm, Deke Kincaid <d...@thefoundry.co.uk> wrote: > > Mads: > > When you say we changed it, did we at one time have a merge2 node that did > not clamp and we changed the formula on the Merge2 node to then clamp? > > I suspect this predates The Foundry's involvement in Nuke (Merge2 node was > introduced before 2007) and anyone who did make the modification has long > since left. I looked through our bug database and no one has filed a bug > though I did find a previous post on this list asking the same question back > in 2012. The trouble is if no one emails support and files a bug then it > does not exist as a feature request/bug. > > Nuke-Users is not an official Foundry channel for filing bugs/features, it is > just a channel for like minded users to converse with each other. You should > send this into supp...@thefoundry.co.uk and get this logged. > > > -- > Deke Kincaid > Media & Entertainment OEM Development Manager > The Foundry > Skype: dekekincaid > Tel: (310) 399 4555 - Mobile: (310) 883 4313 > Web: www.thefoundry.co.uk > Email: d...@thefoundry.co.uk > >> On Wed, Jan 28, 2015 at 9:31 AM, Mads Lund <madshl...@gmail.com> wrote: >> But i guess that leads back to the original question. >> I can't find a single reason (artistically or technically) why they changed >> this to clamp between 0-1. >> >>> On Wed, Jan 28, 2015 at 6:03 PM, Ivan Busquets <ivanbusqu...@gmail.com> >>> wrote: >>> For more crude / unclamped versions of some merging operations (including >>> "overlay"), you can also use the olde "Merge" node (as opposed to the >>> default "Merge2". >>> >>> Merge { >>> inputs 2 >>> operation overlay >>> name Merge1 >>> selected true >>> xpos -254 >>> ypos 199 >>> } >>> >>> >>>> On Wed, Jan 28, 2015 at 8:34 AM, Mads Hagbarth Lund <madshl...@gmail.com> >>>> wrote: >>> >>>> It's just bad with the overlay thing, as it is a pretty useful way to add >>>> grain and do dodge/burn effects. It works out of the box in Photoshop, >>>> Fusion and AfterEffects but you have to make your own overlay gizmo to >>>> make it work in Nuke. >>>> >>>> >>>>> Den 28/01/2015 kl. 17.11 skrev Daniel Short <danisnotsh...@gmail.com>: >>>>> >>>> >>>>> I wondered how the screen would need to be calculated differently since >>>>> the invert function might get you some crazy values in float space. >>>>> It makes sense that they changed it to give us a visually accurate result >>>>> when a mathematical one would look like an error. >>>>> >>>>>> On Wed, Jan 28, 2015 at 10:49 AM, John Mangia <j...@johnmangia.com> >>>>>> wrote: >>>>>> Looks like the nuke screen algorithm has a low pass to allow for values >>>>>> over 1 to comp more naturally. You can see the difference if you expose >>>>>> down. >>>>>> >>>>>>> On Wed, Jan 28, 2015 at 10:37 AM, Mads Lund <madshl...@gmail.com> wrote: >>>>>>> Ahh ok, so the "Screen" operation in the merge node does not just >>>>>>> screen. >>>>>>> >>>>>>> Screen is the same as the Merge node, only with operation set to screen >>>>>>> by default. It layers images together using the screen compositing >>>>>>> algorithm: A+B-AB if A or B ≤1, otherwise max(A,B). In other words, If >>>>>>> A or B is less than or equal to 1, the screen algorithm is used, >>>>>>> otherwise max is chosen. Screen resembles plus. It can be useful for >>>>>>> combining mattes and adding laser beams. >>>>>>> >>>>>>> However i still don't see why they don't apply this to the overlay >>>>>>> method aswell. (why would you want clamp the image?) >>>>>>> And it also doesnt explain why it clamps at 0 (since multiply does not >>>>>>> clamp at 0) >>>>>>> >>>>>>> >>>>>>> >>>>>>>> On Wed, Jan 28, 2015 at 2:46 PM, John Mangia <j...@johnmangia.com> >>>>>>>> wrote: >>>>>>>> Screen is also a multiplicative operation, it's just inverted images >>>>>>>> multiplied together and re-inverted. >>>>>>>> >>>>>>>> >>>>>>>>> On Wed, Jan 28, 2015 at 8:35 AM, Ron Ganbar <ron...@gmail.com> wrote: >>>>>>>>> The math used in Nuke for Screen makes it inheritingly clamp (it >>>>>>>>> doesn't really clamp, but can not produce colors brighter than 1). >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> Ron Ganbar >>>>>>>>> email: ron...@gmail.com >>>>>>>>> tel: +44 (0)7968 007 309 [UK] >>>>>>>>> +972 (0)54 255 9765 [Israel] >>>>>>>>> url: http://ronganbar.wordpress.com/ >>>>>>>>> >>>>>>>>>> On Wed, Jan 28, 2015 at 2:15 PM, Mads Lund <madshl...@gmail.com> >>>>>>>>>> wrote: >>>>>>>>>> According to the documentation: >>>>>>>>>> >>>>>>>>>> • overlay - Image A brightens image B. >>>>>>>>>> Algorithm: multiply if B<.5, screen if B>.5 >>>>>>>>>> >>>>>>>>>> However Nuke also clamps the colors between 0 and 1. >>>>>>>>>> It seem that all other compositing software does not clamp between 0 >>>>>>>>>> and 1, so i am wondering if this is a bug in Nuke? >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> _______________________________________________ >>>>>>>>>> Nuke-users mailing list >>>>>>>>>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>>>>>>>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>>>>>>>> >>>>>>>>> >>>>>>>>> _______________________________________________ >>>>>>>>> Nuke-users mailing list >>>>>>>>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>>>>>>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> _______________________________________________ >>>>>>>> Nuke-users mailing list >>>>>>>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>>>>>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>>>>>> >>>>>>> >>>>>>> _______________________________________________ >>>>>>> Nuke-users mailing list >>>>>>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>>>>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> John Mangia >>>>>> >>>>>> 908.616.1796 >>>>>> j...@johnmangia.com >>>>>> >>>>>> _______________________________________________ >>>>>> Nuke-users mailing list >>>>>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>>>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>>>> >>>>> >>>>> >>>>> -- >>>>> Daniel Short >>>>> www.danisnotshort.com >>>>> 215.859.3220 >>>>> _______________________________________________ >>>>> Nuke-users mailing list >>>>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>>> >>>> _______________________________________________ >>>> Nuke-users mailing list >>>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>> >>> >>> _______________________________________________ >>> Nuke-users mailing list >>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> >> >> _______________________________________________ >> Nuke-users mailing list >> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > _______________________________________________ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
_______________________________________________ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users