Compressor

Dec 16, 2008

Compressor Droplets for Vimeo

Phil Powell , , , Editing, Video 7 Comments

Update 2: Hooray! Thanks to a very thoughtful and generous commenter (much appreciated Shahin), the Droplets are back. The download links below are working again.

Update: Looks like I’ve clumsily lost these when updating the site to a new server – oops! The links will no longer work. Here’s a useful link to a really useful video explaining how to create custom settings and droplets though.

Since I recently posted on a great method for encoding for Vimeo using Compressor resets, I’m currently adopting that workflow for all of the work I’ve been doing this past month or so. It may well be that I’m losing a certain amount of quality by using this technique, but I’m happy that it’s good enough for online distribution.

What I’ve done, is created a set of Compressor Droplets, which live on my desktop and allow me to just drag-and-drop files for encoding. The one-click dialog box which appears feels a lot cleaner and more efficient than having to launch compressor and fiddle with settings and sources etc.

I’m providing the droplet files I’ve created here for download. They’re provided as-is, and with absolutely no warranty – they work for me, and they should work for you. There are three droplets: standard definition, anamorphic standard definition and high definition.

Vimeo SD Droplet

Vimeo SD – Anamorphic Droplet

Vimeo HD Droplet

All you need to do to use them, is drag-and-drop the file you want to encode onto a droplet, and a simple dialog will appear asking for confirmation. Hit “Submit”, and Batch Monitor will launch and process the job. The output files are set to be created in the same directory as your source file.

Jul 23, 2008

Working with Qmaster

Phil Powell , Editing, Video 0 Comment

I’ve never used Qmaster before. I’d never gotten around to poking around to see how it worked, and what benefits it might bring. And after the last installation I had on a G5 refused to function under Leopard, I just left well alone.

Qmaster is, quite simply, a way to create a cluster for video rendering. Rendering is just a whole load of number crunching; it’s processor intensive – just squillions of calculations going on to process and render frames of a moving image. If you’ve got multiple processors in your machine, they get used, but on the basis of whenever they’re needed By allocating a cluster, Compressor can allocate these calculations more efficiently, treating each processor as a separate entity.

I like to think of it a bit like counting votes in an election. If you were to put all of the ballot papers in one big pile, and allow the counters to grab whatever they wanted, then go off an count, it wouldn’t be a very efficient use of resources – time would be wasted to-ing and fro-ing, some people might be counting more ballots than others, and there’d probably just be pandemonium and confusion. By allocating everyone ballots to count, and topping them up when they’re done, you save a lot of time and resources are allocated efficiently.

You can also make clusters available to other machines, so that you can allocate rendering tasks across a network, spreading the load and using processors on other machines which might just be sitting idle (I’ve yet to investigate whether you can include multiple machines in a cluster – haven’t delved that far yet).

Creating a cluster turns out to be relatively simple, and is done through System Preferences. Here’s a step-by-step example on how to set your machine up with a cluster for your own use (note I’m using OS X Leopard with Final Cut Studio 2 installed):

1. Open System Preferences.

2. Under “Other”, click on “Apple Qmaster”.

3. The first thing you’ll see is a series of options for “Share this computer as…”. You want to make sure you select “QuickCluster with services”.

4. Next, you’ll see a list of “Services”. Make sure that “Compressor” is checked, and highlighted.

5. Below the list of Services, hit the button labelled “Options for selected service…”. A little sheet will pop out at the top of the window with an option to select “Number of instances”. This is the number of processors you want to assign to the cluster (the language Apple use here isn’t very clear). Ideally, you want to select as many as you can – I have two processors in my machine, so I select “2″. The more processors you make available, the more efficient your cluster will be.

6. Click “OK”.

7. Under “Quickcluster”, you can assign it a name, so that you can easily identify the cluster when it comes to the point of using it.

8. That’s pretty much it. If you click on “Start Sharing” your cluster will spring into life! (sounds more dramatic than it is – it won’t “spring” at all, it’ll just meander in the background, out of sight).

So, now that you have a cluster set up, how do you go about using it? Really simple – assuming you’re familiar with Compressor 3, you just need to do the following:

When you come to submit a job to the queue, by clicking “Submit…”, you get presented with a popup, containing three options: “Name”, “Cluster” and “Priority”. Just select the cluster you’ve just set up and hit “Submit”.

A few things to note:

Using a cluster works best when you have render which will benefit from multi-processing – by that I mean jobs which it will be possible to split to assign to each process of the cluster. For example, rendering a multi-pass DVD file will split the job into several parts and then merge again at the end. Quicktime movies which are reference movies also work well, as there are lots of referenced files which can be assigned to different processes. If all that sounds a little confusing, it’s probably worth doing a little research and reading up on what happens behind the scenes when you render video files.

Also, I’ve yet to toy around with the “unmanaged services on other computers” options – so that might have to be the subject of a future post.

Apr 8, 2008

HD to SD conversion: the Holy Grail

Phil Powell , , , , , Editing, Video 55 Comments

I remember the first time I was able to experience the magic of Final Cut and real-time HDV editing.  It was like a revelation; an epiphany in my editing life.  I didn’t have to think about working in a different way from DV – it just worked the same: same disk usage, same real-time effects and transitions; same capture process.

Editing is a breeze, and since we set up our new company, we decided that in order to future-proof all of our work, and to attain the highest quality video footage, we’d put an HDV workflow in place.  That means that we shoot in HDV, and we edit HDV.  That way we can distribute HD resolution video online, and be prepared for when we’re (finally) able to produce HD resolution DVDs.

But it introduces a sticky element to our workflow, which ought to be easy to negotiate, but which can really clog up the whole process.  Having all this HD resolution footage around is great, but we need to create DVDs to distribute our films, and to prepare stuff for broadcast.  And Final Cut makes that really, really difficult.

In the past, I’ve used Compressor to render MPEG files ready for use in a DVD project.  But there’s three snags with that:

  1. Compressors rendering quality is pants.
  2. There’s some voodoo you have to carry out in order to deinterlace the source video in Compressor.
  3. It’s really slow.

And I’ve been forced to completely abandon the Final Cut > Compressor > DVD Studio workflow entirely since upgrading to Mac OS X Leopard, since Compressor no longer works (thanks Apple – as far as I know, this is something which Apple are aware of, but are only fixing in newer, Intel versions of Final Cut Studio).

That could have been a really nasty brick wall to have hit, and for a while I was a little lost with Compressor.  I trialled a few other programs which were meant to do a similar job, but they just didn’t cut the mustard.  I was ready to break stuff, until I realised that DVD Studio Pro has it’s own renderer built-in – and from what I can tell it’s much faster, and better than the one found in Compressor.

But, there’s a fly in the ointment.  If you take a HDV file and throw it at Compressor, it’ll happily import it and allow you to work with it just like any other asset.  But when it comes to render, there’s a serious problem with deinterlacing, where you get strange shimmers appearing in some shots, and some really ugly jitters in others.  It just doesn’t work as a method for downconverting.

That’s when I stumbled across the Bonsai method.  This technique is absolute genius, and gets around all the problems, with a pretty fast workflow.  The theory goes:

  1. Create a standard definition timeline, using the DV50 codec.
  2. Drop your HDV footage into that DV50 timeline – Final Cut will deal with all of the rescaling and de-interlacing.
  3. Export the timeline.
  4. Import it into DVD Studio

Et voila!  It sounds simple, and in essence it is.  There’s just a few settings which might trip you up along the way, and I’ve had to experiment with using a few different settings to get good results.

I’ve converted the workflow so that I generate PAL footage, and here’s a brief step-by-step guide to what to do to get HDV footage onto a SD DVD.

First you need to create your sequence for the downconversion:

  1. You should have your finished edit in a HDV or Apple Intermediate Codec sequence.
  2. Create a new timeline, and call it something like “Final Edit – DV50″.
  3. Make sure you have this sequence selected (sometimes Final Cut can be picky over editing settings on something you haven’t selected) – just click on the sequence timeline to make sure it’s at the forefront.
  4. Go to “Sequence settings”.
  5. At the bottom of the settings window, click the “Easy setup” button.
  6. Select “DV50 PAL 48 kHz Anamorphic”.
  7. Back in the sequence setting window, make sure that “Field dominance” is set to “Upper (Odd)”.  This will save you a world of pain when it comes to importing into DVD Studio.
  8. Click the “Video processing” tab.
  9. Towards the bottom, you’ll see a “Motion filtering quality” setting.  Set this to “Fastest (linear)”.  This makes sure that your deinterlaced image doesn’t look like it was done on an Etch-a-Sketch!
  10. Drag your master sequence (your HDV one), and drop it into the DV50 sequence.  Final Cut will scale your footage anamorphic DV resolution.
  11. View the master sequence in the viewer (select it in your timeline and hit return).just check the Filters tab for the sequence, as Final Cut can be prone to add a Field Shift filter if you haven’t set the field dominance correctly – you don’t want that, as it’ll cause headaches later on, and add time to the rendering process.

That’s all the tricky stuff out of the way, you can now go ahead and export a master copy:

  1. Select “Export > Quicktime Movie”.
  2. Make sure “Make movie self-contained” is selected (else you’ll end up with reference movies, and that’ll open a world of pain for you).
  3. Don’t change any other settings – you want to be exporting with the default codecs for the sequence.
  4. Go make some coffee.

Once your file’s rendered, just give it a check.  It’ll play in Quicktime et al in 4:3, so will look all squashed – that’s just because it’s an anamorphic file – we’ll be stretching it out again in DVD Studio.

Final step is to import this footage into DVD Studio Pro:

  1. With your DVD Studio project open, select “Import” and locate your file.
  2. In your asset viewer, right-click on the video file, and choose “Encoder settings…”
  3. It should be set by default, but just make sure that “Field Order” is set to “Top”.  This makes sure that the interlacing of the MPEG encoded file will be the same as your source file.
  4. Set “Aspect ratio” to “16:9″.  This will stretch your footage back to it’s original size.

And that, should be that!  Work with your assets just like you would for any other DVD project, and when you come to build and format your discs, they ought to play back with crystal-clear clarity.

Some things to note:

  • It really is worth always checking your DVD builds on a proper cathode TV.  A lot of the time, the scanning of LCD monitors will hide nasty interlacing problems, and you won’t be able to tell if things have gone astray.
  • I’m working with Final Cut Pro 5 – I’ve not had a chance to upgrade to version 6, so couldn’t say whether this process works in the latest version – or indeed, whether this whole process is no longer needed.
  • There are other elements of the Bonsai Method, such as applying a channel blur to reduce flicker.  I’ve not felt the need to use these, but it’s worth reading about them, as they could be useful to you.

So now, I’m happy because I have a full HD to SD workflow back in place, and it works much faster and with better results than the Compressor workflow I’ve used in the past.  It may not be the best way to do this process, but it’s the best I’ve found so far.  It’s interesting that this is a workflow which Apple shys away from discussing or documenting on their website, or in any of their documents.

I’d be interested to here any input from anyone who has suggestions on improving this process – or indeed from anyone with a quicker, easier alternative.