dpx's are a fair amount faster in Nuke than EXR, in my tests. I suppose we'll hit a point where the camera sensors can capture more than 16 bit and then EXR might get the call, but until then, viva la dpx for live action plates. EXR for cg renders, though, obviously.
I need to update my little test with EXR 2.0 one of these days, give it a go quicktimes have no place in a Nuke script. I mean, they can work and I haven't seem them wreck stability every time, but best to get them ran out and swapped in with image seqs before you take anything to a renderfarm, that's for sure! Nice for editing and possibly for lightweight review, but that's about the extent of Quicktime's usefulness. Brenden Bolles (proEXR plugin guy) has said publically he's working on a container format that would use OpenEXR as a base - essentially an openEXR quicktime. Finally, lossless and float! Although it will likely be .ogg or something thereabouts if I recall. This would be nice! I'm not a fan of ProRes. You can say "perceptibly lossless" all you want, but it's still lossy and where I come from that ain't good. I hate that editors seem to toss it around like it's lossless. On Thu, Aug 8, 2013 at 4:06 PM, John Coldrick <[email protected]>wrote: > No, it's early days yet. I would agree that it's likely network usage > would go up but we have yet to test it. Just interactively comparing, the > speed was about the same, but I get that doesn't translate directly to > heavy usage in the middle of the day. Quite frankly I think it's not a > good idea, I was looking for someone doing violent wave-offs to save me the > trouble. :) > > But yeah, I'll check that, thanks! > > Cheers, > > J.C. > > > On Thu, Aug 8, 2013 at 7:00 PM, Randy Little <[email protected]>wrote: > >> Did you check network and render farm memory usage with QT vs DPX? I >> would think DPX put a lot less load on the nework. Also why dpx. exr with >> zips is smaller and at least as fast which in turn would lessen your >> network load during render or working even more. No? yes? am I still >> asleep trying to remember how plus works after 20 year of doing this? >> >> Randy S. Little >> http://www.rslittle.com/ >> http://www.imdb.com/name/nm2325729/ >> >> >> >> >> On Thu, Aug 8, 2013 at 3:56 PM, John Coldrick <[email protected]>wrote: >> >>> Hey all - we typically pull our plates from the above files and output >>> to dpx files for compositing. Someone here has been pushing for just using >>> the original quicktimes directly in comp(we've gotten a fix from the latest >>> release notes that addresses a subtle colour shift between nuke and >>> compressor). Apart from the arguments about speed(we found in the end it's >>> actually pretty similar) and workflow(head in and out and the rest we can >>> probably handle), it struck me that stability is a potential problem. >>> We're running windows here(win7 64 bit), and I was able to make some >>> quicktime crashes pretty trivially with Nuke 6.3v4 through 7.0v8(same >>> triggers, same crash, which suggests the issue is with quicktime). >>> >>> I'm arguing no for stability reasons, but I can see the benefits if it >>> works - just wondering if anyone here has done this with any success or >>> wildly wave their hands saying 'nooooooo!'. >>> >>> Thanks in advance >>> >>> J.C. >>> >>> _______________________________________________ >>> 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
