Activity › Forums › Creative Community Conversations › NTSC ProRes project wrong
-
Rafael Amador
October 1, 2012 at 4:24 pmDuring the brief timeIi used FCPX, I made few tests on downscaling HD to SD for DVDs, and for small web videos.
I found a noticeable difference in quality when using “Fit” than when downscaling manually. When setting the downscaling manually the picture looked much better, kept much sharper. With “Fit” the picture always looked softer.
I’m a “downscaling maniac”; I always use SHAKE, and i found FCPX (no Fit) of similar quality with SHAKE. Better than compressor with best settings.
rafael -
Jeremy Garchow
October 1, 2012 at 4:46 pm[Jon Chappell] “I suspect that they may be issues within AV Foundation rather than FCPX.”
While I agree that it might be AV Foundation, doesn’t Adobe have any culpability in this?
If you batch process all the clips through Qt Edit, everything work as normal and shows up with the proper color space.
Is Adobe not injecting the right metadata or is AVFoundation not flexible enough to notice what’s going on when there’s missing metadata?
-
Oliver Peters
October 1, 2012 at 5:11 pm[Joseph Owens] “I have to wonder if this is rounding error in a non-hexadecimal-compatible picture aspect ratio. 480 is divisible by 16 but 486 obviously is not.”
In the DV-into-486 case… When you use “fit” it looks soft, as it’s a slight blow-up. If you use “none”, then FCP X centers the image vertically, so a pad of 3 top, 3 bottom. This means field order is wrong. If you use transform to shift it up or down, there’s no correlation to lines or pixels and a Y-value of 1 is too much. I think the last time I did this .2 looked correct. Much easier to deal with this in 7.
– Oliver
Oliver Peters Post Production Services, LLC
Orlando, FL
http://www.oliverpeters.com -
Rafael Amador
October 2, 2012 at 5:02 am[Oliver Peters] “In the DV-into-486 case… When you use “fit” it looks soft, as it’s a slight blow-up. If you use “none”, then FCP X centers the image vertically, so a pad of 3 top, 3 bottom. This means field order is wrong. If you use transform to shift it up or down, there’s no correlation to lines or pixels and a Y-value of 1 is too much. I think the last time I did this .2 looked correct. Much easier to deal with this in 7.”
in FC.7 was 4 on top and 2 at the bottom or the other way around.
rafael
Reply to this Discussion! Login or Sign Up