output Quicktime First then Compressor vs output thru FCP into Compressor

just curious if there a difference in quality here, or just something wrong with my FCP:

Process #1. I've been outputting my timeline in FCP for DVDs, using export using 'compressor', then go the 60min high quality widescreen encode option. This process takes about 12+ hours to export a 10min video with moderate effects.

Process #2. I just outputted a "quicktime" of the same timeline, then I took the quicktime into 'compressor', and export with the same settings as above. This process takes only 4+ hours.

Is there reason for the export time differential? is one quality better one way than the other? or is there something wrong with my FCP program itself?

Any thoughts on this? Thanks.
r.

What rig are you running? A ten minute encode shouldn't take anywhere near that long unless you are upressing or something else hokey - not just endcoding for DVD.

Oh - wait - is it HDV?

In any case, it's been mentioned before that outputting to compressor from within FCP is not the most efficient method. Going out to a self contained movie first is a lot faster and the quality is the same.

As a guide, I find straight encoding for DVD (inside DVDSP) takes about a third of the time of the media length, so 10 minutes on my octocore would take around three minutes.

this is how i understand it:

exporting via compressor means that compressor reaches right into fcp and then all the way back to the original media.
it ignores render files, and re-renders everything AS it's encoding.

i guess that's what takes time.

generally a simple export as QT Movie which you then use in compressor is just as good.
and if your not HDV, you can export a reference movie very quickly.
(i think HDV will take longer... not sure??)

My 02.
Exporting from within FCP to Compressor has pro-cons.

Con:
Locks up FCP

Pro:
Automatically places compression markers at every cut/dissolve.

About the re-render thing.... I didn't see that in the docs.

>Pro:
>Automatically places compression markers at every cut/dissolve.

Export a self contained movie with compression markers (i suppose dvd studio pro markers work too). Import that into Compressor. Click on the "show compression markers" in the preview window. That does the trick. Similar to a direct export off the timeline in this sense.

> About the re-render thing.... I didn't see that in the docs.

While exporting, check the log in Compressor. Some parts of it should say "rendering 4:4:4" or something like that. What's happening here is that Compressor is rendering assets straight from your original referenced media in the timeline. You save a generation exporting files this way. Rendering in final cut causes files to be decompressed to uncompressed 601 for calculation, filter by filter (what Compressor does when encoding off the timeline) before compressing it back to the codec you're working with, effectively losing a generation (which is one benefit of working in uncompressed). Text and graphics will be rendered straight off the timeline too, preserving as much sharpness, especially if you're working off a lossy codec like DV.

Cons: 1. Locks up Compressor. 2. Some Boris text (and i heard Motion files) seem to have problems encoding in this manner.


For a 10 min video, 12 hours for encoding is way too long, even off a 2 pass. It sounds like HDV rendering times. For other formats, I'd only be expecting that kind of rendering time if Compressor is running motion compensated deinterlacing.

If you want the best quality, I always export from FCP using Compressor first to a high quality format like ProRes HQ before encoding to web/dvd/etc. It pulls the original source material and exports at the highest quality. All text generators and filters will be processed at 4:4:4 and if you used Motion for title, it will go to what whatever your project colorspace was, up to 32bit.

While it does take ages, the quality is worth it. I always export at lunch or over night. Then i have a HQ SCQT source i can make multiple distribution formats from and then you can use virtual clusters and cut your render time down significantly and do multiple encodes at once.

"I always export at lunch or over night."

another approach might be to drop your finished sequence into a PRoRes Sequence,
and render.

the advantage of his is that you can stop / start the renders when ever you want.
an export has to be done in one go, and if it stops for any reason, you have to start all over.

you could then export as a reference move, which wouldn't take much time at all, and use that in Compressor.



nick

Quote

Then i have a HQ SCQT source i can make multiple distribution formats from

Dude... you said SCQT!

_______________________________________
SCQT! Self-contained QuickTime ? pass it on!

Thanks everyone for your thoughts on this. I'm curious on what way would be the best to export a SD DVD

My timeline was a 720p60 23.98 project with various filters and color corrections on all clips. I'm exporting out on a powerbook G4 1.67ghz, 2gb ram, i know this is definitely not an optimal machine to be working on but that's what i've had for the last few years and it's been ok.

I guess if the quality is the same as when you export a Quicktime first, then take into Compressor or export out of FCP to Compressor, i'll probally do the first way as that is way faster.

Should i be working in an uncompressed timeline for projects in the future?

Scott Erickson Wrote:
-------------------------------------------------------
> If you want the best quality, I always export from
> FCP using Compressor first to a high quality
> format like ProRes HQ before encoding to
> web/dvd/etc. It pulls the original source material
> and exports at the highest quality. All text
> generators and filters will be processed at 4:4:4
> and if you used Motion for title, it will go to
> what whatever your project colorspace was, up to
> 32bit.
>
> While it does take ages, the quality is worth it.
> I always export at lunch or over night. Then i
> have a HQ SCQT source i can make multiple
> distribution formats from and then you can use
> virtual clusters and cut your render time down
> significantly and do multiple encodes at once.

Nice, thank you.
C

thanks again!