Ivan, you are of course correct. I shouldn't do math right before bed and allow myself to have a mind lapse like that. However, all values would be fit into 0-1 as long as the near value is more than 1m. Come to think of it, values of 1-20m becomes 1-0,05, a very viewable range. And pretty common distance range (from camera). I'm excluding huge crane shots and heli shots here. The infinity "feature" sure is handy as well!
Cheers, Elias Ericsson Rydberg 5 dec 2013 kl. 02:31 skrev Ivan Busquets <[email protected]>: > Elias, > 1/z does not normalize depth to 0-1. > Any values with depth < 1 will have values > 1 when inverted. > > IMO, the main benefit of treating depth as 1/z is that you don't have to deal > with "infinity" values for empty areas. > > > > On Wed, Dec 4, 2013 at 5:02 PM, Elias Ericsson Rydberg > <[email protected]> wrote: >> 1/z would make sense to me as it would fit an arbitrariily deep map in a 0-1 >> space. Any point in the pass would be easily visualised with nukes viewer >> tools, exposure mainly ofc. Although I'm a fan of having absolute values, >> 1/z has it's benefits. >> >> Cheers, >> Elias >> >> 5 dec 2013 kl. 00:49 skrev Deke Kincaid <[email protected]>: >> >>> yup, your right, long day, brain not working. I converted my normalized >>> pass wrong, so it was giving me the inverse which appeared right. :) >>> >>> -- >>> Deke Kincaid >>> Creative Specialist >>> The Foundry >>> Skype: dekekincaid >>> Tel: (310) 399 4555 - Mobile: (310) 883 4313 >>> Web: www.thefoundry.co.uk >>> Email: [email protected] >>> >>> >>> On Wed, Dec 4, 2013 at 6:31 PM, Ivan Busquets <[email protected]> >>> wrote: >>>> Not sure if I follow, or what combination you used in your test, but the >>>> standard depth output of ScanlineRender (1/z) is what DepthToPoints wants >>>> as an input by default. >>>> >>>> >>>> set cut_paste_input [stack 0] >>>> version 7.0 v8 >>>> push $cut_paste_input >>>> Camera2 { >>>> name Camera1 >>>> selected true >>>> xpos 1009 >>>> ypos -63 >>>> } >>>> set N73f6770 [stack 0] >>>> push $N73f6770 >>>> CheckerBoard2 { >>>> inputs 0 >>>> name CheckerBoard1 >>>> selected true >>>> xpos 832 >>>> ypos -236 >>>> } >>>> Sphere { >>>> translate {0 0 -6.449999809} >>>> name Sphere1 >>>> selected true >>>> xpos 832 >>>> ypos -131 >>>> } >>>> push 0 >>>> ScanlineRender { >>>> inputs 3 >>>> shutteroffset centred >>>> motion_vectors_type distance >>>> name ScanlineRender1 >>>> selected true >>>> xpos 830 >>>> ypos -43 >>>> } >>>> DepthToPoints { >>>> inputs 2 >>>> name DepthToPoints1 >>>> selected true >>>> xpos 830 >>>> ypos 91 >>>> depth depth.Z >>>> N_channel none >>>> } >>>> >>>> >>>> >>>> >>>> >>>> On Wed, Dec 4, 2013 at 3:11 PM, Deke Kincaid <[email protected]> wrote: >>>>> Actually I think we are both wrong. I was just playing with a camera >>>>> from the 3d scene with depth and it needs to be distance to match. 1/z >>>>> gives you the reversed coming out of a little window look. >>>>> >>>>> -deke >>>>> >>>>> >>>>> On Wednesday, December 4, 2013, Ivan Busquets wrote: >>>>>> I don't think that's right, Deke. >>>>>> >>>>>> DepthToPoints expects 1/z by default, not a normalized input. >>>>>> Same as the output from ScanlineRender. >>>>>> >>>>>> The tootip of the "invert depth" knob states that as well: >>>>>> >>>>>> "Invert the depth before processing. Useful if the depth is z instead of >>>>>> the expected 1/z" >>>>>> >>>>>> >>>>>> >>>>>> On Wed, Dec 4, 2013 at 2:45 PM, Deke Kincaid <[email protected]> >>>>>> wrote: >>>>>> 1 is near though there is an invert depth option in depth to points. >>>>>> >>>>>> >>Am I wrong? >>>>>> >>>>>> Nuke is 32 bit floating point so it shouldn't matter that much as long >>>>>> as the original image was a float. Precision would only matter if you >>>>>> were working in a 8/16 bit int box. >>>>>> >>>>>> -deke >>>>>> >>>>>> On Wednesday, December 4, 2013, Ron Ganbar wrote: >>>>>> 0 is near? >>>>>> >>>>>> Normalised values aren't precise, though. They're very subjective to >>>>>> what was decided in the render. It won't create a very precise point >>>>>> cloud. >>>>>> Am I wrong? >>>>>> >>>>>> >>>>>> >>>>>> Ron Ganbar >>>>>> email: [email protected] >>>>>> tel: +44 (0)7968 007 309 [UK] >>>>>> +972 (0)54 255 9765 [Israel] >>>>>> url: http://ronganbar.wordpress.com/ >>>>>> >>>>>> >>>>>> On Wed, Dec 4, 2013 at 11:51 PM, Deke Kincaid <[email protected]> >>>>>> wrote: >>>>>> It's just looking for 0-1. You can do it with an expression node.... or >>>>>> Jack has a handy J_Maths node in J_Ops which converts depth maps between >>>>>> types really easily. >>>>>> >>>>>> -deke >>>>>> >>>>>> >>>>>> On Wednesday, December 4, 2013, Ron Ganbar wrote: >>>>>> Hi all, >>>>>> for DepthToPoints to work, what kind of depth do I need to feed into it? >>>>>> 1/distance? distance? normalised? >>>>>> And how do I convert what comes out of Maya's built in Mental Ray so it >>>>>> will work? >>>>>> >>>>>> Thanks! >>>>>> Ron Ganbar >>>>>> email: [email protected] >>>>>> tel: +44 (0)7968 007 309 [UK] >>>>>> +972 (0)54 255 9765 [Israel] >>>>>> url: http://ronganbar.wordpress.com/ >>>>>> >>>>>> >>>>>> -- >>>>>> -- >>>>>> Deke Kincaid >>>>>> Creative Specialist >>>>>> The Foundry >>>>>> Skype: dekekincaid >>>>>> Tel: (310) 399 4555 - Mobile: (310) 883 4313 >>>>>> Web: www.thefoundry.co.uk >>>>>> Email: [email protected] >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> Nuke-users mailing list >>>>>> [email protected], http://forums.thefoundry.co.uk/ >>>>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> -- >>>>>> Deke Kincaid >>>>>> Creative Specialist >>>>>> The Foundry >>>>>> Skype: dekekincaid >>>>>> Tel: >>>>> >>>>> >>>>> >>>>> -- >>>>> -- >>>>> Deke Kincaid >>>>> Creative Specialist >>>>> The Foundry >>>>> Skype: dekekincaid >>>>> Tel: (310) 399 4555 - Mobile: (310) 883 4313 >>>>> Web: www.thefoundry.co.uk >>>>> Email: [email protected] >>>>> >>>>> >>>>> _______________________________________________ >>>>> 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
