Hi Henning, those black areas in my screenshot are actually black, with a very small alpha. Yes I'm using the DeepColorCorrect. But I think I might have an idea for you, at least an idea what happens... Try checking your fog for overlapping samples with the DeepSample node. Those seem to cause artefacts for me now and then. The next thing is that with overlapping samples, the DeepColorCorrect splits those overlapping samples up to produce non overlapping samples. The Problem now is, that if the alpha of the front part is opaque then the back part gets discarded. In the test case I have over here I have two samples:
1.) deep.front=80.8, deep.back=84.84 alpha=1 2.) deep.front=81.8, deep.back=85.84 alpha=1 after the DeepColorCorrect I end up with a single sample: deep.front=80.8 deep.back=81.8 However this only happens with an alpha of 1. With an alpha <1 the samples get split into non overlapping samples according to Beer's law(thats what I think, haven't calculated it manually though;)) as proposed in the Technical Introduction to EXR 2.0. So this is just a vague idea and as I suppose that you don't have opacities of 1. But I would try and check for overlapping samples. Hope that helps at least a little bit. And keep me updated with your results, that's interesting me... cheers Patrick ----- Original Message ----- From: [email protected] To: [email protected] Date: 19.02.2013 13:16:57 Subject: [Nuke-users] RE: Nuke-users Digest, Vol 60, Issue 19 > Hi Patrick, > > what are you using? DeepColorCorrect? And are the black areas in your image > supposed to be black? Because that's what I meant by artefacts, that bits not > included in the depth range just get cut off or turn out black... > So in the attached screen shot I'm trying to darken down the foreground with > a fall off up to a certain point and the areas that shouldn't be affected by > the correction at all turn to black. Any ideas? > > Cheers > Henning > > > > Message: 9 > Date: Mon, 18 Feb 2013 19:31:21 +0100 (CET) > From: "Patrick Heinen" <[email protected]> > Subject: Re: [Nuke-users] Creating fall off with deep files > To: [email protected] > Message-ID: <[email protected]> > Content-Type: text/plain; charset="utf-8" > > Hi Henning, > > well I'm having no troubles over here with a smoke that I colorcorrect only > on a slice with fall off, if that's what you mean. See the screenshot > attached. > > cheers Patrick > > ----- Original Message ----- > From: [email protected] > To: [email protected] > Date: 18.02.2013 18:00:40 > Subject: [Nuke-users] Creating fall off with deep files > > >> Is there a way to create a matte with soft fall off from deep files? I'm >> trying to color correct a smoke trail that goes from dark to light, away >> from the camera. I've tried using the zmap feature in the DeepColorCorrect >> but I'm getting weird artefacts from that, like the renders being cropped if >> the deep values of the image exceed the deep values specified in the zmap >> tab. The other option I guess is DeepCrop but that doesn't give me a soft >> fall off, either. I thought about faking it by cropping it multiple times >> and then layering it back together but surely there must be a more elegant >> method... So how do you do it? >> >> Thanks in advance! >> Henning_______________________________________________ >> Nuke-users mailing list >> [email protected], http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > -------------- next part -------------- > A non-text attachment was scrubbed... > Name: Capture8.JPG > Type: image/jpeg > Size: 162972 bytes > Desc: not available > Url : > http://support.thefoundry.co.uk/cgi-bin/mailman/private/nuke-users/attachments/20130218/fd21fdc0/Capture8-0001.jpeg > > _______________________________________________ > 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
