Render / Exporting curiosity

Like Ryan described, i dont think it "recompresses" that DV file again.
Its already DV, if you export "current settings" your not adding any compression.

Good points on the "mastering/distributing" end.
Usually i deliver a few different formats, so i use that route.
(making sure all my files are coming from the master)

Or sometimes i have to give the master to someone who doesn't have FCP.

shawn
 
Spartacus said:
Please elaborate on this! How is this done?
Click to expand...

Check out the "Distributed Processing Setup" PDF from the help menu within Compressor 3.

Basically what you are doing is creating a render farm. You can do this within a single multi-processor computer or to a group of computers if you are lucky enough to have access to them. For example, a Mac Pro 8-Core can be configured for up to 8 "instances". Basically one per core.

When you tell Compressor or FCP to render anything (especially h.264) you are letting much of the horsepower of the 8-Core machine go to waste. Software is not yet fully optimized to take advantage off all 8 processor cores efficiently. So Apple came up with a workaround for Compressor using Qmaster called Virtual Clusters.

Virtual Clusters divide the task into multiple sections instead of rendering one frame at a time in sequence. You then in essence could be rendering 8 frames of video in the same moment. Obviously if you can render 8 frames at once, you will finish rendering the whole file a heck of a lot faster then just one frame at a time.

So a Virtual Cluster could divide up your video into maybe 16 sections. The more processors you have as part of your farm, the more sections compressor will divide the clips into. As one section finished rendering, the next section will be assigned to the next available "instance" Once all the sections are complete, Compressor will recombine all the sections in order to complete your full and final compressed video.

Even setting up just 2 "instances" can gain quite a lot of time in rendering. Check out the test chart BareFeats created from their Compressor 3 Tests.

http://www.barefeats.com/octopro5.html


You will need goobs of RAM to make the most out of your Virtual Cluster. It's estimated at least 1.5GB of RAM for every instance. Then you may still want some 2GB of buffer left over for regular OSX needs. So here is where a minimum of 8GB RAM for and 8-Core computer is needed. Anything less and you will be getting into Hard Drive -page file territory and it will effect your render times. The machine I use has 16GB. And i wish i could up it to 32GB. But not yet...
 
Last edited:
Yea We've got a FiberCable XServe Video Server linked up to 4 8Core Mac Pros. Quite often I'll borrow 2 cores from two of the towers to help out in my rendering. Yet the other editors using those machines don't notice a thing!!! Not even a slight hitch. :~) Since they are all linked to the same high speed server, they can all access the same file for rendering at the same time. Compressor & Qmaster manages the rest.

My tower and a third tower I set up for 6 Instances each. So i have 16 "instances" total crunching away at my projects. I leave out two instances on these two machines so I can work on other programs/software and in case someone jumps on that third Editor, they won't notice much of anything.

It's a beautifull thing.
 
I am gonna throw in a small monkey wrench here. One thing I'm currently experimenting with is going from interlaced to progressive. Much of our source material is interlaced of course. But when we send anything to our streaming server, we want it all progressive. Compressor 3 can de-interlace footage and it does a pretty decent job of it. It does take a bit longer time however. So I'm toying with the idea of de-interlacing the video before it goes into Compressor. Since I create 5 different sizes for different streaming bandwidths, it might actually save me time in the long run to have FCP do the de-interlacing work beforehand where it would only do it once instead of 5+ different times in compressor. So I'm currently testing out Progressive Masters. When going to DVD it will re-interlace things again. But the image quality difference might not be noticeable enough to matter.
 
OK - I was just reading about this the other night (book is at home and memory fails), but something about, if you export from FCP via Compressor, Compressor uses the raw material from the Timeline rather than what FCP did to the QT file (or someting like that. if I can find it, I'll update).
 
Great Info. Thanks. For some reason I always thought "current settings" meant that it was the previous settings, not the settings that the video files already had.
 
Dick Campbell said:
OK - I was just reading about this the other night (book is at home and memory fails), but something about, if you export from FCP via Compressor, Compressor uses the raw material from the Timeline rather than what FCP did to the QT file (or someting like that. if I can find it, I'll update).
Click to expand...


It certainly will. Which is why you find you can't use FC. Compressor is tying it up. Exporting into one Master file frees up FCP. You are just copying the RAW material to a new Master file.
 
ryan, just curious...

what kind of material do you edit that needs that much rendering power?
or is it just to really help cut down your render times?

shawn
 
I work for a public school district where much of our stuff consists of meetings and presentations. We produce training videos for staff and teachers as well as school spotlight videos for the public. A majority of the meetings and presentations are quite long. We've had meetings up to 6 hours long.

Some of the footage we get is not shot professionally. But it gets used for staff and teacher training. So we need to make them look and sound as acceptable as possible to train teachers and staff without annoying them with dark or unbalanced video or air conditioner humming for hours straight.

Almost everything that we produce or that gets sent to us becomes streamed via a quicktime streaming server. We also provide DVD copies when we request. And that's one part of what our department does.

With over 20,000 employees in the district and 180 schools to support, we stay very busy with our 9 person staff. We are basically a production company, a post house, a tech support service, and a distribution house all in one department within the school district. All of us wear a bunch of different "hats". Being as efficient as possible without loosing quality is a key requirement for us to get everything done. So I'm constantly working on finding the best workflows. It's a never ending battle, especially in this current economy where huge budget cuts and staff cuts actually means tons more work for us.
 
Last edited:
RyanBellaCine said:
It certainly will. Which is why you find you can't use FC. Compressor is tying it up. Exporting into one Master file frees up FCP. You are just copying the RAW material to a new Master file.
Click to expand...
Yes, I understand that. So , is "Exporting QuickTime Movie" the same as "Exporting into one Master file" (or how do you mean?) thanks.
 
I guess I am still confused about this. I had always thought that the multipass encoding of Compressor gave you a better quality of file than exporting as a QT file for DVD authoring. If that is true, then I would think that going to QT first would compromise that somewhat, since no encoding can really improve a file once the information is gone.

If it isn't true, than why use compressor for full res output for DVD authoring at all?
 
"I had always thought that the multipass encoding of Compressor gave you a better quality of file than exporting as a QT file for DVD authoring."

if your encode settings are exactly the same,
i dont think you will see any quality difference.
your essentially working with the same media.

"I would think that going to QT first would compromise that somewhat"

not if your working with a master file.
*see ryans explanation above
as long as you don't compress/encode upon export.

not all encoders interface with editing software.
where i work we feed high end encoders full resolution QT files.

i think ryan also made a good point in splitting up the "mastering/distribution" aspect

hope that helps
shawn
 
It does. But then why use compressor at all for DVD authoring at full res? You import that QT file right into DVD SP4.
 
dvds use a specific kind of compression and file architect to function, if u have any other file like an uncompressed mov file, somewhere before the dvd is burned it will be transcoded to mpeg2, it would be the default settings to the compression and transcoding used by the software and if automated it will definitly favor time over quality settings, so it will do one pass, it will again favor lower bit rate, and other settings for various reasons, and also dvd or mpeg 2 encoders differs allot in quality from each others and there is a broad selection to choose from.
that's why u would go to compressor, to choose the best possible settings to get the best results.
 
found the reference I was looking for: from the Apple Pro Taining Series "Compressor Quick-Reference Guide" page 19"

"When FCP exports every individual frame from elements such as titles, Motion project files, complex composites and animations, Compressor receives the raw data regardless of the current sequence settings." To paraphrase, exporting as a QT will be rendered first with the sequence codec. And on page 24, "Compressor (will) ultimately render the source media using the output codec, instead of having FCP render with the sequence codec first."

but as has been stated, the down side is, it ties up FCP. hope this helps.
 
You must log in or register to reply here.
Back
Top