Using Compressor with multiple cores

Here's a quick tip that a lot of people still don't know about. So you've just bought that shiny new Mac Pro, you finish your Final Cut Project, export to Compressor and wait.... and wait... and wait. Hang on a second, that brand-new 8-core machine is only utilizing one core!

Welcome to the world of Final Cut Studio, which is completely unaware of multiple cores in your system. Hopefully this will change in future versions but in the meantime, here is a trick to using those cores in Compressor.

1. Go to System Preferences and open the Apple Qmaster section.



2. In the Setup tab, click Stop Sharing if Qmaster is already activated.



2. Select QuickCluster if it is not already selected.

3. Under Services, select the service marked Compressor and select the checkboxes for Share and Managed.

3. Then click Options for Selected Service.

4. Now you need to select the number of instances to use for the cluster. This is the number of copies Qmaster will spawn to compress the job, and is a question that's open to debate.



Apple's official advice is that the number of instances should be half the number of cores. So if you have a quad-core machine, you should select two cores and four if you have an 8-core machine.

However, newer Mac Pros (early 2009 onwards) and MacBook Pros (early 2010 onwards) support a feature called HyperThreading which allows each core to act as two separate "virtual cores", allowing two tasks to run simultaneously on a single core. So for these models you should match the instances to the number of cores - i.e. four instances for a quad-core and eight for an 8-core.

Some people have suggested setting the instances to the total number of cores (or virtual cores on the newer machines) minus one. I find that this depends on the format you are compressing. If you are compressing a "light" format like DV, you could probably get away with it. However, more processor-intensive formats like H.264 will need a lot more headroom. You would also need at least 1 GB of RAM per instance.

Another point to bear in mind is that Compressor needs to stitch the separate pieces of the file together once processing is complete. More instances mean more pieces to assemble, which adds to the time taken. Consequently, the time saved by increasing the number of cores may be offset by the increase in the time taken to assemble the final file. I would therefore recommend sticking with Apple's advice of half the number of cores for early Mac Pros and matching the number of cores for later models.

5. If you are using Shake. Maya or have any other command-line renderer set up with Qmaster, select the Rendering item in the list and click Options for Selected Service.

6. Click the + button and add as many "Local machines" as necessary (see step 4 for details), then click Ok.



7. Under the QuickCluster section, type a name for your new virtual cluster.

7. Click Start Sharing and close System Preferences.

8. In Final Cut Pro, export your sequence as a QuickTime movie. This is important. Do not choose File > Export > Compressor, as this will fail.

9. In Compressor, import the movie, set up your batch and click Submit.

10. Instead of "This machine", select the name of your new cluster in the Cluster list and click Submit. The movie clip will be processed by multiple cores in your machine.



This is called Virtual Clustering, and it is achieved by launching a new version of Compressor for every instance. This is much easier from the perspective of Compressor's programmers but it takes up more memory and is less efficient than if they were to implement true multi-core capabilities into the application.

So be aware that using this method will require considerably more memory than rendering on a single instance. Apple recommends 1 GB RAM per instance but again, this depends on the media you are transcoding to. If you set too many instances and don't have enough memory, your machine would slow down (or worse, crash) thus undoing the whole point of setting up the virtual clusters in the first place. For best results, follow the recommendations in Step 4.

See the article Speeding up Compressor for more tips on improving Compressor performance.

Update 2010-12-06: This forum thread has an interesting comparison between This Computer and a Quick Cluster with various instances. Results vary depending on the chosen codec but the fastest times seem to occur with roughly half the number of instances compared to cores. The data would also suggest that a Quick Cluster is more beneficial on long files than short ones.

How to reinstall Compressor and Qmaster

We recently released Compressor Repair, which aims to fix some of the most common problems with Compressor - but it can't help you if you've got missing or corrupted files. Sometimes a reinstall of Compressor and Qmaster is needed (although these steps apply to any software in the Suite). Here's how to do it properly.

1. Download FCS Remover and run it.

2. Select "Compressor and Qmaster Only" as the preset.



3. Click Remove and enter your admin password.



4. Restart your machine, and when it boots up again, empty your Trash.

5. Insert your Final Cut Pro or Logic DVD (I'm using Logic here), ctrl-click on "Install Logic Studio" or "Install Final Cut Studio" and select Show Original.



If you are running Final Cut Studio 3 (FCP 7), perform Step 6. If you are running Final Cut Studio 2 (FCP 6), skip to Step 7.



7. Run Compressor.pkg (if using Final Cut Studio 3) or Compressor.mpkg (if using Final Cut Studio 2 - note the extra "m").



8. When that finishes, run Qmaster.pkg (FCS 3) or Qmaster.mpkg (FCS 2 - again note the "m").



That's it! At this point, it is a good idea to update Final Cut Studio. Several users have reported that this method did not work until they had updated Compressor and Qmaster to the latest versions after reinstalling.

(Note: I've seen other people recommend Pacifist as a way of installing the files again once they've been removed. Pacifist is a great piece of software but if you make a mistake, it's possible to seriously screw things up unless you know what you're doing. I think this method is a lot safer and less susceptible to user error. However, if the above steps don't work for you, try Scott Simmons' Pacifist method here.)

11-5-08: New AppleCare articles

In the past week or so, Apple has posted several new ProApp-related articles to the Support section of its web site.

Final Cut Studio: Switch to higher performance graphics with MacBook Pro (Late 2008) - Learn how to configure the MacBook Pro (Late 2008) for better graphics performance with Final Cut Studio.

Motion: QuickTime proxies from RED sometimes import with no audio - Sometimes there may seem to be no audio when you import QuickTime proxies generated by RED cameras or by REDAlert into Motion. This may happen if you import the movie by using the Import button in the File Browser, drag the movie from the File Browser to the Canvas, or drag the movie from the Finder to the Canvas.

Final Cut Pro, Final Cut Express: Unexpected quit during AVCHD ingest - Final Cut Pro 6 or Final Cut Express 4 may sometimes quit unexpectedly when you attempt to transfer AVCHD media, if the third-party Perian QuickTime component is installed.

Final Cut Pro: "AE Effects Error: Unable to find plug-in" warning when opening some projects - You may see the warning dialog "AE Effects Error: Unable to find plug-in..." when you open certain projects in Final Cut Pro.

Motion: Higher performance graphics mode may remain active if Compressor is used to transcode Motion projects on MacBook Pro (Late 2008) - When you transcode a Motion project in Compressor on a MacBook Pro (Late 2008), your computer may stay in "Higher performance" graphics mode after the project has been transcoded. This can happen even after you change the graphics setting in the Energy Saver pane in System Preferences to "Better battery life," log out, and then log back in. The Energy Saver pane in System Preferences may report the graphics mode is "Better battery life," but it will still be in "Higher performance" mode.

DVD Studio Pro: Authoring a DVD that plays automatically without a menu - Most DVDs are created with at least one menu from which the user can select tracks, other menus, special features, and so forth. In some cases, you may wish to create a DVD that just plays its content without the need for user interaction.

Logic Pro: Using DSP powered plug-ins with systems that include nodes - Logic Pro 8 allows you to utilize the processing power of additional computers (nodes) on a Gigabit ethernet network. Most of the plug-ins native to Logic, and many third party Audio Unit plug-ins, can be processed by node computers.
Learn more about using DSP powered plug-ins (plug-ins that run from a card or externally connected device, rather than directly from the CPU) and nodes.

Pro Application Replacement Serial Numbers - A new serial number can be generated in the event that yours is missing or invalid. Apple requires a valid proof of purchase to request a replacement serial number submitted via fax or email. The following guidelines will help you to determine what is best for you.

Final Cut Pro Audio Filter Guide Part 2

This is the second part of the Final Cut Pro Audio Filter Guide. Part 1 covered the Audio Units that ship with Mac OS X and is available here. This part will cover the audio filters that ship as part of Final Cut Pro.

This guide is intended to inform Final Cut Pro users of the options available to them for fixing bad sound. It is worth mentioning at this point that this guide is intended for users of Final Cut Pro 6.

Now, one thing you may have noticed is that there are often duplicate filters such as AUBandpass (an Audio Unit) and the built-in Band Pass Filter. So which should you use? Well, for a start, bear in mind that some of those filters have the same function but different parameters. I would generally advise you to pick the built-in filters over the Audio Units whenever possible, as they tend to use up fewer system resources.

3 Band Equalizer

This allows you to take three separate bands (low, med and high) and adjust the gain up and down individually. I personally prefer AUFilter (in the Apple folder) because it gives you a bit more control and allows you to adjust 5 bands.

This is useful when you need to boost the bass of a voice or improve a flat-sounding voice.

Band Pass Filter

This will allow a range of frequencies on either side of the center frequency to pass through and reduce (attenuate) frequencies outside this range.

This is identical to the AUBandpass filter (in the Apple folder) except that instead of the bandwidth parameter, it has a mysterious one called Q. Q stands for Quality Factor and is a different way of representing the bandwidth. There are numerous articles about it on the internet that get quite technical but all you need to know is that it's the relationship between the center frequency and the bandwidth (f/b) so Q is inversely proportional to the bandwidth (i.e. when the bandwidth goes up, Q goes down by a proportional amount). It is worth noting that Q is not the bandwidth itself but it is related to it. If you want to find out the bandwidth, just divide the center frequency by Q.

If you don't understand the above explanation just play it by ear or use the AUBandpass filter.

Compressor/Limiter

The Compressor / Limiter reduces the volume of sounds above the threshold amplitude. This is a useful way of minimizing the difference between two subjects talking at different volumes or making sure that the audio fits within the limits of the playback device.

Obviously this reduces the volume of the overall audio so Preserve Volume rectifies this (although I find it is often then too loud). Attack time refers to the time it takes for the filter to decrease the volume once it has detected a frequency with an amplitude above its threshold. Release time refers to the time taken for the filter to increase the volume again once the high amplitude frequency has finished playing. Higher values allow for a smoother and less noticeable response but set them too high and the compressor won't respond quickly enough. This is something inherited from live audio mixing where you don't know what's coming next. Setting the threshold to just under your preferred limit allows time for the compressor to lower the volume in anticipation for a louder sound once the threshold is reached.

Ratio tells the compressor by how much it should reduce the volume when a sound exceeds the threshold. If the ratio is set to 2 (2:1), then a 10 dB increase in volume above the threshold will be halved to a 5 dB increase. It is worth noting that the compressor lowers the volume of sounds above the threshold but does not necessarily reduce them to a value at or below the threshold. Be aware that very loud sounds could still theoretically peak.

Finally, it is also worth mentioning that the compressor reduces the volume difference between the subject and any background noise, so background noise will be more noticeable upon boosting the audio after applying the compressor.

DC Notch

Sometimes you may experience a DC current leakage through the mic, causing noise in the recorded audio. A DC Notch filter will remove the DC offset component which you probably need a degree in audio engineering to fully understand. It has no parameters.

This is one of the lesser-used filters FCP provides.

Echo

This one obviously adds an echo. Effect Mix allows you to mix the echoing audio with the original in order to better blend it in. Effect Level controls the volume level of the echoes (but not the original audio). Brightness controls the degree that the echoes will overlap. Feedback controls how long each echo will last and delay controls the spacing between each repeat.

It is worth noting that the echoes will be abruptly cut off unless you lengthen the audio clip or fade it out.

Expander / Noise Gate

The opposite of a compressor. A compressor takes high amplitude sounds and lowers them whereas an expander takes low amplitude sounds (sounds below the threshold) and lowers them. I know it probably seems strange to lower sounds that are already quieter than the rest of the audio but this is designed to increase the dynamic range of the clip.

Threshold is the amplitude below which the expander will kick in. Ratio controls the proportion of expansion - for example with a ratio of 2 (2:1), a 3 dB fall below the threshold will be adjusted to a 6 dB fall. Attack time is the time for the volume change to be applied once it falls below the threshold. Release time is the time for the volume to return to normal once the signal goes above the threshold again.

A noise gate is a more extreme expander that will completely eliminate frequencies below the amplitude threshold. This can be achieved with high ratios (e.g. 10).

Gain

This is a simple filter that raises or lowers the volume of the audio. It is added automatically to clips when you use the Modify > Audio > Apply Normalization Gain command.

High Pass Filter

This will attenuate (reduce) low frequencies below the threshold. This is the same as a Low Shelf Filter. Use the Q slider to modify the bandwidth (width of the frequency range).

Useful for cutting out low frequency noise such as the rumbling of traffic or very low notes in a deep voice.

High Shelf Filter

This will attenuate (reduce) high frequencies above the threshold. This is the same as a Low Pass Filter. Gain allows you to adjust the volume of audio that passes through the filter.

Useful for filtering out high frequency noise such as buzzing on the soundtrack.

Hum Remover

This is similar to the Shelf / Pass filters but it has several extra parameters. Frequency, Q and Gain have been explained many times above. In order to explain what the harmonic check boxes mean, we need to delve into a little audio theory.

The fundamental frequency is the lowest frequency in a harmonic series (the Frequency value in this case). Harmonics are integer multiples of the fundamental frequency - e.g. if f=60, 2f=120, 3f=180, etc. These play at the same time as the fundamental frequency and contribute to the tone of a sound. The Hum Remover is more powerful than a Shelf / Pass filter because it allows you to remove these specific frequencies without removing any frequencies in between these. For example, if f=60 and you wanted all harmonics up to 5f removed, a Low Shelf or High Pass filter would remove ALL frequencies up to 300 Hz, potentially affecting the quality of your sound. The Hum Remover would not do this.

Despite being called Hum Remover, the use of harmonics makes it useful for other purposes such as enhancing or reducing a musical instrument on a soundtrack.

Low Pass Filter

The Low Pass Filter has the same effect as the High Shelf Filter - it will attenuate (reduce) frequencies higher than the specified frequency range, keeping lower ones intact.

Frequency refers to the center frequency and Q is a way of representing the bandwidth (the width of the frequency range).

Useful for reducing high frequency noise such as buzzing.

Low Shelf Filter

The Low Shelf Filter has the same effect as the High Pass Filter - it will attenuate (reduce) frequencies lower than the specified frequency range, keeping higher ones intact.

Useful for reducing low frequency noise such as air conditioner hums.

Notch Filter

The opposite of a Band Pass filter. Instead of only allowing frequencies within a certain range, this cuts out frequencies within a certain range.

Frequency refers to the center frequency and Q is a way of representing the bandwidth (the width of the frequency range).

Useful if you have noise of a constant frequency on your soundtrack (such as a buzzing sound).

Parametric Equalizer

This is a combination of Band Pass, Notch and Shelf filters combined into in a single filter. Frequency is the center frequency, Q is related to the bandwidth (see the explanation above) and gain allows you to boost or cut the frequencies passing through the filter.

Reverberation

This is similar to the Echo filter but considerably more sophisticated. Rather than simply repeating sounds with a delay, it allows you to mimic the characteristics of echoes within various locations. This is incredibly useful when performing ADR (Automated Dialogue Replacement) because it is highly likely that the sound booth you record the ADR in will sound nothing like the original location. This allows you to mimic the effect of sound waves bouncing off walls, with some canceling each other out and some increasing in intensity. It can also be used sparingly to improve a flat-sounding voice.

Subtlety is often the key with this filter and Effect Mix allows you to mix the reverb with the original sound to help blend it in. Effect Level controls the intensity, Brightness controls the degree that the echoes overlap and Type allows you to specify various preset locations.

It is worth mentioning that the reverberation will end abruptly unless you extend your audio clip or fade it out at the end.

Vocal DeEsser

This helps to reduce the intensity of "s" sounds, most noticeable if the actor has a lisp. The controls are similar to a compressor but it is optimized for reducing sibilant ("s") sounds. Ratio controls the amount of reduction - e.g. if the "s" sound is 6 dB and the ratio is 2, it will be reduced to 2 dB.

Emphasis controls the sensitivity of the filter and Broad Band Mode widens the bandwidth so that more frequencies around the center frequency are affected.

Vocal DePopper

Sometimes if a microphone is directly in front of an actor's mouth they will accidentally breathe into it while speaking, causing a wind-type noise to be generated. This filter aims to minimize these.

The parameters are similar to a compressor - ratio controls the ratio of reduction proportional to the intensity of the sound. Broad Band Mode widens the bandwidth so that more frequencies around the center frequency are affected.

One Final Note

These filters do a decent job of repairing troublesome audio but they are not miracle cures. Sometimes (and this is never popular with producers) it just isn't cheaper to fix it in post. Even big Hollywood movies with access to multi-million dollar sound studios re-record a lot of their audio. It is important to always think realistically.

However, I hope this two-part guide has been useful in showing you just what can be improved. I noticed a significant improvement in the sound quality of my projects once I understood more about the audio filters available to me, and I hope you'll be able to say the same.

Final Cut Pro Audio Filter Guide Part 1

In my conversations with other editors it became clear that for a lot of them (myself included), audio was a major area of weakness. They could handle editing, color correction, graphics, etc, but only knew a few ways of improving bad audio. This inspired me to create this guide to show people exactly what Final Cut Pro offers out of the box.

Filters in FCP are split into two categories - Apple Audio Units (AU) which ship with the operating system and are available to any application, and FCP-specific ones. Today I'll be giving details on the Audio Units with the FCP-specific ones coming in Part 2.

AUBandpass

The bandpass filter allows a certain range of frequencies to pass through and rejects any frequencies that are outside of that range. So if you have a person talking with rumbling traffic and high-pitched TV static in the background, you could use the bandpass filter to isolate the center frequencies of the person's voice.

AUDynamicsProcessor

This filter allows you to adjust the dynamic range of the audio (the difference between the loudest and softest sounds). Compression lowers the volume when it exceeds a certain threshold (making loud sounds quieter and therefore decreasing dynamic range). Expansion lowers the volume when it is lower than the threshold (thus making quiet sounds quieter and therefore increasing dynamic range).

Using a compressor and expander at the same time allows you to avoid overly-loud peaks while still maintaining dynamic range. The compressor will reduce the louder portions of the audio in proportion to their distance from the threshold (i.e. louder sounds will be reduced by a much larger amount), helping to even out amplitude variations. The expander will then take the quieter sounds and reduce their volume (again, proportionally), thus evening out the amplitude variation of the quieter sounds and increasing the dynamic range.

Imagine some music where you have loud crashing drums, a relatively quiet triangle and a trumpet somewhere in between. The drums are far too loud so you apply a compressor to bring them down. The drums are now a lot quieter but you've noticed that the subtlety of the triangle has been lost - it is now far too prominent. So you apply an expander to lower the volume of the triangle to an acceptable level. The AUDynamicsProcessor allows you to do this with one filter instead of two, which is a much more efficient use of system resources.

Attack time refers to the amount of time it takes for the compressor to implement a change in volume and release time is the time taken to reduce the compressor back to its original level afterwards.

There's a bit of debate over what "headroom" actually means as it's not a standard term. To the best of my knowledge, it refers to the number of dB the signal is permitted to exceed the threshold of the compressor/limiter.

The Dynamics Processor is a great way of compressing and/or expanding your audio to make sure that it fits within the audible range of your listening equipment.

AUDelay

This repeats parts of your audio. Dry/wet mix controls whether or not the repeated sounds overlap each other. 100% dry means no overlapping and 100% wet means full overlapping. Delay time is the delay before the repeat starts. Feedback controls the number of repetitions. Lowpass cutoff frequency stops certain low sounds from passing through the filter.

This filter is normally used for atmospheric effects. For more environmentally realistic results (at the cost of performance), I recommend the Reverberation filter that comes with Final Cut Pro.

AUDistortion (Leopard only)

This is a very comprehensive distortion filter offering a lot of control. I can't give much advice on the various parameters as I don't actually know what they mean and there's no documentation available. I would suggest adjusting by ear (it is probably worthwhile lowering the render quality while adjusting, and then raising it again once you're happy with the results).

AUFilter

I use this one a lot. It allows you to take 5 bands (5 ranges of frequencies) and adjust the gain up or down for each one. If a person's voice is muffled, increasing the gain on the higher frequencies can often improve this. I also sometimes use it to add bass to make a person's voice sound richer.

The bandwidth slider allows you to adjust the size of each frequency range, with the frequency sliders referring to the center of the range.

AUGraphicEQ

(image truncated for space reasons)

This is similar to the AUFilter above but the size of the frequency ranges cannot be adjusted and you can have up to 31 bands instead of just 5. This site provides a guide to common frequency ranges.

I've used this a couple of times to boost bass in a person's voice but I generally prefer AUFilter for this. It's good for removing or reducing a specific frequency range.

AUHighShelfFilter

This cuts off frequencies above the cutoff frequency, allowing lower sounds to pass through. The gain slider controls the amount that the frequencies passing through will be boosted.

Good for reducing high-frequency noise.

AUHipass

This reduces lower frequencies and allows higher ones to pass through. Resonance controls the intensification of the higher frequencies that pass through.

Good for removing low rumbling sounds such as traffic or for reducing very deep voices (I have had to do this at times).

AULowpass

Reduces (attenuates) higher frequencies and lets lower ones through, controlled by the cutoff frequency. Resonance controls the intensification of the frequencies that pass through.

Good for removing high-pitched noise.

AULowShelfFilter

Reduces frequencies lower than the cutoff frequency and lets higher ones pass through. This is similar to the AUHighPass filter above but this offers the ability to adjust the gain instead of resonance.

Good for removing low hums.

AUMultibandCompressor

This allows you to compress multiple bands (frequency ranges) individually for more control than a traditional compressor.

Pre-gain boosts the signal before it is processed which is useful if the signal is too low to be processed adequately by the compressor, and post-gain reduces the gain back to a normal level afterwards.

Crossovers 1, 2 and 3 define the point at which the previous band ends and the next one begins. Threshold refers to the amplitude level at which the compressor kicks in. Headroom, as stated above, probably gives you extra leeway above the threshold. Eq allows you to boost or lower each band.

Possible uses for the Multi Band Compressor include lowering the dynamic range of bass sounds without interfering with higher frequency sounds. This is different from the High Shelf Filter which indiscriminately cuts all bass below a certain frequency.

AUNetSend

AUNetSend is one of the most interesting filters in that it's not actually a filter. It does absolutely nothing to affect the way your audio sounds but what it does do is allow you to send audio across a network.

You need to have an application that implements audio generators. Audio editing applications such as Logic implement these. Alternatively, there is a tool called AU Lab in /Developer/Applications/Audio if you have the developer tools installed. Soundtrack Pro does not support this. Add an AUNetReceive generator to the track in your audio application, add an AUNetSend filter to your FCP audio clip and hit play in FCP. You should see "AUNetSend" pop up in the AUNetReceive configuration dialog. Select it and your track should be receiving the audio from Final Cut Pro. Note that it only appears while the timeline is playing in FCP.

In my testing, the Status parameter appeared to do absolutely nothing.

AUParametricEQ

This allows you to boost or lower the amplitude of a signal within a certain range. This is useful if for example you have a high-pitched buzzing noise in the background and you only want to eliminate that particular frequency and keep your higher sounds intact.

AUPeakLimiter

The Peak Limiter differs from a compressor in that a compressor reduces the volume of an entire track when a frequency reaches a certain level whereas the Peak Limiter reduces just that frequency. This is particularly useful if there is background noise on the track that would produce a noticeable fluctuation if the entire track were to receive a volume adjustment.

Attack and release time, as mentioned above, control the amount of time it takes for the filter to implement a change in amplitude, with longer times allowing a smoother transition. Pre-gain allows you to boost the volume before it reaches the filter in order to change the number of frequencies being affected. The limiting amount allows you to limit the amount that the filter will reduce the amplitude.

AUPitch

Adjusts the pitch of your audio, obviously. There are a LOT of controls though, and I have to admit that I don't have a clue what a Glb Trigger Thresh or a Loud Aggr K is. Effect blend blends the pitch-shifted audio with the original and is sometimes necessary to make voices sound natural. It is worth mentioning that I've gotten perfectly acceptable results by adjusting the first two parameters and not bothering with any of the others. I'd imagine that most people wouldn't have to adjust more than these.

I tried a Google search on some of these parameters but they only turned up forum threads where people were asking what on Earth they meant and no-one was able to solve their problem. If anyone does know, I'd be interested to hear from them.

This can be useful for making a male actor's voice more masculine (yes, I have had to do this!).

AURogerBeep (Leopard only)

This emulates the "roger beep" sound when someone lets go of the button on a walkie talkie. It will automatically play the sound when the audio level drops below a certain threshold for a certain amount of time (as if the person stopped speaking).

In gate threshold and in gate time refer to the time that the threshold amplitude must be maintained before the sound kicks in. Out gate threshold and out gate time refer to the amount of time that the threshold must be maintained before it ends the roger beep (adjust this if there is background noise). Roger level is the volume of the roger beep, and I think sensitivity and roger type are self-explanatory.

This is an effects filter with quite a narrow purpose so it's not something that gets used very often.

AUSampleDelay

Similar to AUDelay except that the delay time is set as a number of audio samples instead of a number of seconds.

Important:

Two of the above filters are Leopard only. If you use those filters and then transfer your project to a Tiger machine, you will receive an error message and will be unable to use those filters within the project.

Part 2 covers the filters built into FCP.

Pro Applications Update 2008-03

Don't know how I missed this. On Thursday, Apple released Pro Applications Update 2008-03 that patches both Compressor and Qmaster to 3.0.4.

Only one fix is included in the release notes but it's a big one:



I'm sure you know the drill by now - don't update in the middle of a project, only update if it's entirely necessary and make sure everything is backed up before you do.

This update is concerning to me for two reasons. Firstly, it requires OS X 10.5.5, completely locking Tiger users out. Presumably 2008-04 will require 2008-03 so will all patches require 10.5.5 from now on? Was 2008-02 the last Tiger-compatible patch for Final Cut Studio? There has been no official word from Apple on this.

Secondly, when you go to the download page it only lists 2008-03 as an option, with 2008-02 nowhere to be seen. How are you supposed to download Final Cut Pro 6.0.4? There's always Software Update but that's not really practical if you have a lot of machines. Apple seriously needs to rethink the way it offer patches to FCS users.

QuickTime 7.5.5 released

Apple today released QuickTime 7.5.5 to coincide with iTunes 8, FrontRow and various other patches released today. The update is required for the latest iTunes update and also fixes vulnerabilities from maliciously-crafted movies and images. If you just use your machine for editing, this patch is not necessary. If you are likely to be using iTunes 8 or doing a lot of web surfing on your FCP machine (it's ideally best to avoid using your FCP machine for other purposes) I would advise waiting around for at least a week for any issues to surface.

As always, only update if it is absolutely necessary, NEVER update in the middle of a project unless there is a serious problem preventing you from completing the project, and always clone your disk before updating so that you have an exact copy of your original configuration to revert back to.

QuickTime 7.5.5 for Leopard
QuickTime 7.5.5 for Tiger

Speed editing

There is a great article currently up on the COW about ways to speed up your Final Cut Pro editing. This inspired me to add a few of my own. Of course, they're specific to the kind of work I'm doing (currently TV) and the way I work but other people might find them useful.

Playing things back at a faster rate
I think the purists will hate this one. For TV shows, I often get pre-cut footage from the studio (from the on-set mixer) and it's my job to cut out the parts where they screwed up. I normally get a wide cam as well, as something to cut between for safety. I've found that I can cut it very quickly if I play it at a faster-than-normal rate and use keyboard shortcuts.

If you press L once, it will play at a faster rate (1.5x?) BUT, crucially, you can still understand every word that is said. Go to Tools > Keyboard Layouts > Multiclip and you can use the Cmd-Numpad 1 and Cmd-Numpad 2 (and so forth) buttons for cutting between cameras. Then I press Ctrl + V when someone screws up so that a cut point is generated, which I later come back and delete. So I can get a rough cut done in less than the time it would take to watch it normally. This works very well for the kind of shows I'm currently working on, but this method obviously won't work for every situation such as drama and music videos.

Use Multiclips whenever you can
If you have two cameras shooting in sync, there is no reason to not use Multiclips. I'm a big fan of them because you sync them once and once only. Then the hard work is over and when the Producer demands that you change a shot to the other angle, you can do it in a matter of seconds.

Use keyboard shortcuts instead of tools where possible
Keyboard shortcuts for certain functions can save time. But what really saves a lot of time is using a keyboard shortcut to replace a tool. Here are a few of them -

• Ctrl-V - Razor blade (cuts at the playhead)
• R, then V, then < or > - R selects the Roll tool, V selects the nearest cut point and < or > rolls the clip backwards or forwards.
• RR, then V, then < or > - RR selects the Ripple tool, V selects the nearest cut point and < or > ripples the clip backwards or forwards.
• S, then < or > - S selects the Slip tool and < or > slips the selected clip backwards or forwards.
• SS, then < or > - SS selects the Slide tool and < or > slips the selected clip backwards or forwards.
• [ and ] - Press V to select the nearest edit point and press [ or ] to trim it left or right.
• Type a number anywhere on the timeline (e.g. 5 or -10) - Moves the selected clips the specified number of frames

Keep your timeline rendered as much as possible
If you go on a break or something, press Alt + R to render your timeline. That way, the majority of the timeline should stay rendered, with only the parts you change requiring a re-render. When your client asks you for a preview copy, you can very quickly render out a reference movie, plug it into Compressor and then continue with your work. The next time you do it, it will be even quicker. And when (if) you come to put it on tape, it will speed up the rendering process there too.

Use Motion projects instead of rendered movies
If you use Motion for effects, it is far quicker to insert the actual project into your timeline than it is to render out a movie clip. And if you want to make a change just Alt + Tab to Motion, make your change, save, Alt + Tab back to FCP and there it is. No exporting necessary. This method depends on you having a decent graphics card.

Use Master Templates where possible
Taking it one step further, if, say you have a standard animated namestrap format for a show. You could take it into Motion, change the name then save it as a new copy but this takes time and you end up with hundreds of files. Instead, create the template version and then in Motion go to File > Save as Template. Create a new folder and name your new template. Now go back to Final Cut Pro and in the Generators pop-up or the Effects window, click on Master Templates and select your new template. Go to the Controls tab and you can enter any text you like. Much quicker.


One final note - there is a fine line between a fast editor and a sloppy one. There is very much an art and a science to editing and I believe that you can speed up the science but rushing the art will make it suffer. Producers like to give me a lot of footage. I like this a lot because it gives me more to work with but for some reason, the more footage they give me, the faster they expect the end product. The more footage I have, the more decisions I can potentially make and so it takes longer to decide on the "right way" to cut it (there is no right way but you know what I mean). This can't really be sped up.

I'll probably think of some more later so keep your eyes peeled. And if you've got any speed tips of your own, I'd love to hear them.

Update: Thought of another one. When I'm cutting montages, I like to go through the footage finding short clips I like and putting them up the other end of the timeline so that they are easily within reach. I used to mark the ins and outs, manually drag the clip to the timeline, then click on the Viewer to select it again. You can do this much quicker with keyboard shortcuts. Mark the ins and outs (I and O) then press F9 to insert edit the clip or F10 to overwrite edit it (note: Tiger and Leopard have F9 and F10 bound to Expose by default). Then press Cmd-1 to re-select the Viewer. I can go through footage very quickly with this method.

FCP Quick Tip: Viewing extra render information

As you probably know, the render bar above the Final Cut Pro timeline changes color to indicate whether a clip is unrendered, fully rendered or rendering on-the-fly. This color coding is useful but it all depends on remembering what each color refers to.

If you hover the mouse over the render bar for a few seconds it will tell you not only the status of video and audio but, if it is unrendered, exactly why FCP is unable to play it in real-time. Very useful.

Comparing footage with difference mattes

It's a common question - how much quality do you lose when converting from Format X to Format Y or when you are forced to recompress footage? You can compare data rates and squint at it all day but that's not really telling you much.

Here's an easy way to tell exactly what has been lost.

Final Cut Pro
1. Take your Final Cut Pro sequence and export to the second format (the one that you plan to convert to). Or, if your footage didn't originate in FCP, import both versions (before and after) and place them into a new sequence.

2. Place the "before" version onto the bottom track and the "after" above it.



3. Ctrl-click on the "after" version and select Composite Mode > Difference.



Shake
1. Import the "before" and "after" versions into Shake using FileIn nodes.



2. Click on the Layer tab and click on MultiLayer.

3. Connect the noodles from the FileIns to the MultiLayer.



4. In the Parameters tab, change the composite mode of the second clip (the one listed at the top) to Difference.



Both applications will produce an image similar to the one below. This image tells you what has been lost from one version to the other. The clearer this image is, the more data has been lost. If you can clearly make out the edges of objects from the original image, this data loss is probably unacceptable. I personally would consider the image below unacceptable but how much data loss you can accept depends, of course, on the final destination of the clip. For example, web users will be much more forgiving than cinema-goers.