It's not the calculation of the min/max that's expansive - it's pulling every pixel of the source image through the pipe just to calculate the min/max.
For instance Nuke skips scanlines if your zoomed out - if it's forced to calculate min/max then it must pull the invisible scanlines too. Same for if you're zoomed into the image and the rest of the image is cropped offscreen. You get no speedup for any cropping. Not arguing against having a normalize feature - just say'in it ain't free... -jonathan Sent from my iPhone On Oct 14, 2012, at 3:38 PM, "Nathan Rusch" <[email protected]> wrote: > Seems like this feature is a perfect candidate to be implemented directly on > the GPU, since it already has the display buffer, and absolute accuracy isn’t > paramount. > > -Nathan > > > From: Jonathan Egstad > Sent: Sunday, October 14, 2012 9:50 AM > To: Nuke user discussion > Cc: [email protected] > Subject: Re: [Nuke-users] Normalize Viewer > > Hi Frank, > > Not to a devil's advocate or anything...but calculating the min/max of an > image means sampling the entire image before a single pixel can be drawn in > the Viewer. Needless to say this will destroy Nuke's update speed. > > As long as that's understood as a side-effect of this feature, then soldier > on. > > -jonathan > > Sent from my iPhone > > On Oct 13, 2012, at 6:22 PM, Frank Rueter <[email protected]> wrote: > >> None of those solutions actually produce what we're after though (some of >> your solutions seem to invert the input). >> >> We need something that can compresses the input to a 0-1 range by offsetting >> and scaling based on the image's min and max values (so the resulting range >> is 0-1). You can totally do this with a Grade or Expression node and a bit >> of tcl or python (or the CurveTool if you want to pre-compute), but that's >> not efficient. >> >> I reckon this should be a feature built into the viewer for ease-of-use and >> speed. > _______________________________________________ > 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
