-
HDV to SD DVD, QT vs direct to Compressor
-
I’m on a G5 dual 2.3, 3.5MB Ram, FCP6. It takes about 7 mins. to render a one min. clip using the 10 bit unc., render hi prec.yuv, best motion. The only effect on the clip is a color correction for
Same clip takes 6 mins.to conform/render 1080i hdv, render 8bit yuv, best motion.
Unless I did something wrong, it seems like this workflow would not be a viable option because of having to re-render my fcp TL for the 10b seq., making 2 very long renders before going to compressor. . . unless compressor time with the 10b unc. is significantly shorter than with an hdv sc QT movie.
Do you think it is, or am I doing something wrong? If it does not save much time, does it give significantly better image quality?
-
[Christopher Wright] “is which codec works best for the highest res encoding of HDV to standard DVD. I read all the past postings on the subject, and at times you suggest going the Uncompressed SD route, at times the ProRes route, and I even saw (other peoples’)posts suggesting using the DVCPRO50 SD route”
Hi Christopher,
The point is to go to better quality (?) codec than the one you started with.
Your HDV is 420/8b.
You can go:
– DVCPro50: 422/8b Compressed
– Proress SQ: 422/8b Compressed (Better than DVCPro50)
– 8b Unc
– Proress HQ: 422/10 Compressed
– 10b Unc
The most critical step in the process HDV> SD DVD is the downscaling of the footage. You crush your 1920×1080 pixels in 720×480 while changing the pixel aspect.
If you do it in a 10b sequence (10bUnc/Proress HQ + High precission rendering) is that will be done in 32b Floating point instead of 8b. Much, much more precision and quality yield.
Another reason why I go 10b is because I always use the Nattress “Chroma Smooth/Sharp” filter.
This really improve the look of the DV and HDV footage. You get the best results (absolutely noticeable) when applied in a 10b sequence.[Christopher Wright] ” I have seen both referenced and self-contained QT methods mentioned, and encoding from within FCP (using Quicktime conversion), or going to Compressor with your self-contained or reference QT movie.”
The problem of doing things with Compressor is that Compressor works very well when you set Frame Control ON. For example if you send a 10b Unc movie, the process will be done in 10b only if Frame Control is ON. Otherwise will be done in 8b. But Frame Control ON means time. Some times, long, long time.
This is why I try to do the most of the task in FC. Much faster than Compressor. If Compressor can yield a better quality I can not tell with my eyes.
Probably the area in which Compressor beats FC is in the Time Base engine. But in this case there is no Time Base changes going on. You start and finish with the same number of Frames per second.[Christopher Wright] ” I am really spoiled by having a component DVD recorder that takes the Kona LH signal and downconverts it in real time to a stunning letterboxed DVD.”
If you get good and fast MPG2 this way, why don’t you do it like that, then rip the MPG2 from the disk? You just need to bring them to DVDSTP where you can set chapter markers, etc.
rafael -
[chris gorman] “I’m on a G5 dual 2.3, 3.5MB Ram, FCP6. It takes about 7 mins. to render a one min. clip using the 10 bit unc., render hi prec.yuv, best motion. The only effect on the clip is a color correction for
Same clip takes 6 mins.to conform/render 1080i hdv, render 8bit yuv, best motion. “
Thats not a big price to pay for the difference in the rendering. When you are working in 8b you are dealing with 256 levels per channel. In 10b you do it with 1024.[chris gorman] “Unless I did something wrong, it seems like this workflow would not be a viable option because of having to re-render my fcp TL for the 10b seq., making 2 very long renders before going to compressor. . . unless compressor time with the 10b Unc. Is significantly shorter than with an hdv sc QT movie. “
If you give to Compressor something that only needs to be transcoded, Compressor will be very fast.
This happens when you bring a Standard PAL or NTSC movie. No downscaling, no pixels change.
When you send HDV, the 1920×180 Square pixels have to be converted in 720×480/576 NTSC/PAL Pixels. This is the step that really takes time. Once the picture has been downscaled, the conversion in MPG2 it takes less than real time in a more or less fast Mac.
What I would suggest you is capture HDV. Edit in a HDV sequence with the Rendering in Proress. When you have your sequence ready get rid of all the rendering and drop your HDV sequence in the 10b time-line.
You want to avoid double rendering. Some times you can not. And some time you shouldn’t.
Is a good policy to trash all the renders before a final export. Normally i do it. I only keeps those renders that really take ages.
rafael -
I may just do that!
I will try the Prores route, and then if the difference is worse, I’ll just rip the VOB files from the other DVD.Dual 2.5 G5, IO, Kona LH, IO, Medea Raid, UL4D, NVidia 6800, 4Gig RAM
Octocore 8 GB Ram, Radeon card, MBP, MXO
Windows XP Adobe Studio CS3, Vegas 8.0, Lightwave 9.2, Sound Forge 9, Acid Pro 6, Continuum 5, Boris Red 4, Combustion 2008, Sapphire Effects -
The batch monitor was still showing a progress bar and about 10 mins. remaining after over 40 hrs. compression time on the 10b unc QTmovie. So, I think the batch monitor was broken, or something was wrong with the file.
Finally, when I looked at the m2v file in it’s destination folder, it looked done. If I can judge by the “created” vs. “modidified” time listed for the file, which showed a 3 hr. difference, maybe the compression actually took 3 hours.
So I burned a DVD and it did not look good. The other one I burned from the hdv QT movie to compressor looks better than the 10b unc.
This 90 min. video did not have much movement in most of it, was cuts only edit, one short clip had a filter, and several had color correction. Otherwise, no complicated effects, no graphics.
Concerning “rendering”, does Compressor require that any QT sc movie from fcp be rendered in fcp before export, or can you dump or not do the render in fcp and let Compressor do it.
My major concern is final quality, but time starts to become a critical factor when it gets excessive.
-
Hi Chris,
I really fill guilty for having you suggested this way. basically I work like that, going from EX-1 to 10b Unc SD.
I really get very god results and much shorter render times.
But, Chris, you must make sure that your system is optimized. When I see that Compressor or FC gives unusual render times, or I see that they work using very little CPU, I make some system maintenance.
Normally after running DiskWarrior things really speed up.
rafael -
It’s ok, don’t feel guilty. For now I’ll use the hdv qt movie version of the sd dvd. But, I’d still like to solve this problem by doing some tests on short pieces.
mpeg-2 just seems like such a substandard format. i wish i knew if it was a limitation of hdv, compressor, my lack of expertise, or what.
For basics, I’d like to get a better understanding of such things as:
does export direct from fcp to compressor eliminate double compression on a rendered hdv seq., or is it beneficial to do same after dropping it into a pro res seq.?
besides direct from fcp, what are the other options for eliminating double compression, eg., choosing export using QT conversion and choosing some uncompressed settings vs. dropping the footage into a new seq. w. unc. settings.
or, (for minimal compression) after rough cuts only edit, drop hdv footage into prores seq., apply effects, export QT self contained, current settings to compressor.
Or something else?
Reply to this Discussion! Login or Sign Up