Every once in awhile a movie comes along that has the potential to change how we in the film and video world work. Focus is one such movie. It’s a romantic caper film in the vein of To Catch a Thief or the Oceans franchise. It stars Will Smith and Margot Robbie as master and novice con artists who become romantically involved. Focus was written and directed by the veteran team of Glenn Ficarra and John Requa (Crazy, Stupid, Love) who decided to use some innovative, new approaches in this production.

Focus is a high-budget, studio picture shot in several cities, including New Orleans and Buenos Aires. It also happens to be the first studio feature film that was cut using Apple’s Final Cut Pro X. The production chose to shoot with ARRI Alexa cameras, but record to ProRes 4444, instead of ARRIRAW (except for some VFX shots). At its launch in 2011, Final Cut Pro X had received a negative reaction from many veteran editors, so this was a tough sell to Warner Bros execs. In order to get over that hurdle, the editorial team went through extensive testing of all of the typical processes involved in feature film post production. They had to prove that the tool was more than up to the task.

Proving the concept

df1515_focus_5_smMike Matzdorff, first assistant editor, explains, “Warner Bros wanted to make sure that there was a fallback if everything blew up. That ‘Plan B’ was to step back and cut in either [Apple] FCP 7 or [Avid] Media Composer. Getting projects from Final Cut Pro X into Media Composer was very clunky, because of all the steps – getting to FCP 7 and then Automatic Duck to Avid. Getting to FCP 7 was relatively solid, so we ran a ‘chase project’ in FCP 7 with dailies for the entire show. Every couple days we would import an XML and relink the media. Fortunately FCP X worked well and ‘Plan B’ was never needed.”

df1515_focus_8_smGlenn Ficarra adds, “The industry changes so quickly that it’s hard to follow the progress. The studio was going off of old information and once they saw that our approach would work and also save time and money, then they were completely onboard with our choice.” The editorial team also consulted with Sam Mestman of FCPworks to determine what software, other than FCP X, was required to satisfy all of the elements associated with post on a feature film.

df1515_focus_4_smThis was a new experience for editor Jan Kovac (Curb Your Enthusiasm), as Focus is his first Hollywood feature film. Kovac studied film in the Czech Republic and then editing at UCLA. He’s been in the LA post world for 20 years, which is where he met Ficarra and Requa. Kovac was ready to be part of the team and accept the challenge of using Final Cut Pro X on a studio feature. He explains, “I was eager to work with John and Glenn and prove that FCP X is a viable option. In fact, I was using FCP X for small file-based projects since the fall of 2012.”

Production and post on the go

df1515_focus_1_smFocus was shot in 61 days across two continents, during a three-month period. Kovac and three assistants (Mike Matzdorff, Andrew Wallace, Kimaree Long) worked from before principal photography started until the sound mix and final delivery of the feature – roughly from September 2013 until August 2014. The production shot 145 hours of footage, much of it multicam. Focus was shot in an anamorphic format as 2048 x 1536 ProRes 4444 files recorded directly to the Alexa’s onboard cards. On set the DIT used the Light Iron Outpost mobile system to process the files, by de-squeezing them and baking in CDL color information. The editors then received 2048 x 1152 color-corrected ProRes 4444 “dailies”, which were still encoded with a Log-C gamma profile. FCP X has the ability to internally add a Log-C LUT on-the-fly to correct the displayed image. Therefore, during the edit, the clips always looked close the final appearance. Ficarra says, “This was great, because when we went through the DI for the final grading, the look was very close to what was decided on set. You didn’t see something radically different in the edit, so you didn’t develop ‘temp love’ for a certain look”.

df1515_focus_6_smA number of third-party developers have created utilities that fill in gaps and one of these is Intelligent Assistance, which makes various workflow tools based on XML. The editors used a number of these, including Sync-N-Link X, which enabled them to sync double-system sound with common timecode in a matter of minutes instead of hours. (Only a little use of Sync-N-Link X was made on Focus, because the DIT was using the Light Iron system to sync dailies.) Script data can also be added to Final Cut Pro X clips as notes. On Focus, that had to be done manually by the assistants. This need to automate the process spurred Kevin Bailey (Kovac’s assistant on his current film) to develop Shot Notes X, an application that takes the script supervisor’s information and merges it with FCP X Events to add this metadata into the notes field.

During the months of post, Apple released several updates to Final Cut Pro X and the team was not shy about upgrading mid-project. Matzdorff explains, “The transition to 10.1 integrated Events and Projects into Libraries. To make sure there weren’t any hiccups, I maintained an additional FCP X ‘chase project’.  I ran an alternate world between 10.0.9 and 10.1. We had 52 days of dailies in one Library and I would bring cuts across to see how they linked up and what happened. The transition was a rough one, but we learned a lot, which really helped down the line.”

Managing the media

df1515_focus_2_smFinal Cut Pro X has the unique ability to internally transcode quarter-sized editorial proxy files in the ProRes Proxy format. The editor can easily toggle between original footage and editorial proxies and FCP X takes care of the math to make sure color, effects and sizing information tracks correctly between modes. Throughout the editing period, Kovac, Ficarro, and the assistants used both proxies and the de-sequeezed camera files as their source. According to Kovac, “In Buenos Aires I was working from a MacBook Pro laptop using the proxies. For security reasons, I would lock up the footage in a safe. By using proxies, which take up less drive space, a much smaller hard drive was required and that easily fit into the safe.”

df1515_focus_3_smBack at their home base in LA, four rooms were set up connected to XSAN shared storage. These systems included iMacs and a Mac Pro (“tube” version). All camera media and common source clips. like sound effects libraries. lived on the XSAN, while each workstation had a small SSD RAID for proxies and local FCP X Libraries. The XSAN included a single transfer Library so that edits could be moved among the rooms. Kovac and Ficarra shared roles as co-editors at this stage, collaborating on each other’s scenes. Kovac says, “This was very fluid going back and forth between Glenn and me. The process was a lot like sharing sequences with FCP 7. It’s always good to keep perspective, so each of us would review the other’s edited scenes and offer notes.” The other two systems was used by the assistants. Kovac continues, “The Libraries were broken down by reel and all iterations of sharing were used, including the XSAN or sneaker net.”

Setting up a film edit in FCP X

df1515_focus_9_smAs with any film, the key is organization and translating the script into a final product. Kovac explains his process with FCP X, “The assistants would group the multicam clips and ‘reject’ the clip ranges before ‘action’ and after ‘cut’. This hides any extraneous material so you only have to sort through useable clips. We used a separate Event for each scene. With Sam and Mike, we worked out a process to review clips based on line readings. The dialogue lines in the script were numbered and the assistants would place a marker and a range for every three lines of dialogue. These were assigned keywords, so that each triplet of dialogue lines would end up in a Keyword Collection. Within a scene Event, I would have Keyword Collections for L1-3, L4-6, and so on. I would also create Smart Collections for certain criteria – for instance, a certain type of shot or anything else I might designate.”

Everyone involved felt that FCP X made the edit go faster, but it still takes time to be creative. Ficarra continues, “The first assembly of the film according to the script was about three hours long. I call this the ‘kitchen sink’ cut. The first screening cut was about two-and-a-half hours. We had removed some scenes and lengthened others and showed it to a ‘friends and family’ audience. It actually didn’t play as well as we’d hoped. Then we added these scenes back in and shortened everything, which went over much better. We had intentionally shot alternate versions of scenes just to play around with them in the edit. FCP X is a great tool for that, because you can easily edit a number of iterations.”

Engineered for speed

df1515_focus_10_smWhile many veteran editors experienced in other systems might scoff at the claims that FCP X is a faster editor, Mike Matzdorff was willing to put a finer point on that for me. He says, “I find that because of the magnetic timeline, trimming is a lot faster. If you label roles extensively, it’s easier to sort out temporary from final elements or organize sound sources when you hand off audio for sound post. With multi-channel audio in an Avid, for example, you generally sync the clips using only the composite mix. That way you aren’t tying up a lot of tracks on the timeline for all of the source channels. If you have to replace a line with a clean isolated mic, you have to dig it out and make the edit. With FCP X, all of the audio channels are there and neatly tucked away until you need them. It’s a simple matter of expanding a clip and picking a different channel. That alone is a major improvement.”

Ficarra and Kovac are in complete agreement. Ficarra points out, “As an editor, I’m twice as fast on FCP X as on Avid. There’s less clicking. This is the only NLE that’s not trying to emulate some other model, like cutting on a flatbed. You are able to move faster on your impulses.” Kovac adds, “It keeps you in the zone.”

The final DI was handled by Light Iron, who conformed and graded Focus. The handoff was made using an EDL and an FCPXML, along with a QuickTime picture reference. Light Iron relinked to the original anamorphic camera masters and graded using a Quantel Rio unit.

Filling in the workflow gaps

A number of developers contributed to the success of FCP X on Focus. Having a tight relationship with the editing team let them tailor their solutions to the needs of the production. One of these developers, Philip Hodgetts (President, Intelligent Assistance) says, “One of the nice things about being a small software developer is that we can react to customer needs very quickly. During the production of Focus we received feature requests for all the tools we were providing – Sync-N-Link X, Change List X and Producer’s Best Friend. For example, Sync-N-Link X gained the ability to create multicam clips, in addition to synchronizing audio and video, as a result of a feature request from first assistant Mike Matzdorff.” This extended to Apple’s ProApps team, who also kept a close and helpful watch on the progress of Focus.

df1515_focus_11_smFor every film that challenges convention, a lot of curiosity is raised about the process. Industry insiders refer to the “Cold Mountain moment” – alluding to the use of FCP 3 by editor Walter Murch on the film, Cold Mountain. That milestone added high-end legitimacy for the earlier Final Cut among professional users. Gone Girl did that for Adobe Premiere Pro and now Focus has done that for a new Final Cut. But times are different and it’s hard to say what the true impact will be. Nevertheless, Focus provided the confidence for the team to continue on their next film in the same manner, tapping Final Cut Pro X once again. Change can be both scary and exciting, but as Glenn Ficarra says, “We like to shake things up. It’s fun to see the bemused comments wondering how we could ever pull it off with something like FCP X!”

For those that want to know more about the nuts and bolts of the post production workflow, Mike Matzdorff released “Final Cut Pro X: Pro Workflow”, an e-book that’s a step-by-step advanced guide based on the lessons learned on Focus. It’s available through iTunes and Kindle.

For some additional reading on the post production workflow of Focus, check out this Apple “in action” story, as well as Part 1 and Part 2 of’s very in-depth coverage of how the team got it done. For a very in-depth understanding, make sure you watch the videos at PostPerspective.

Originally written for Digital Video magazine / CreativePlanetNetwork.

©2015 Oliver Peters

Understanding SpeedGrade

df1615_sg_1How you handle color correction depends on your temperament and level of expertise. Some editors want to stay within the NLE, so that editorial adjustments are easily made after grading. Others prefer the roundtrip to a powerful external application. When Adobe added the Direct Link conduit between Premiere Pro CC and SpeedGrade CC, they gave Premiere Pro editors the best of both worlds.


df1615_sg_4SpeedGrade is a standalone grading application that was initially designed around an SDI feed from the GPU to a second monitor for your external video. After the Adobe acquisition, Mercury Transmit was eventually added, so you can run SpeedGrade with one display, two computer displays, or a computer display plus a broadcast monitor. With a single display, the video viewer is integrated into the interface. At home, I use two computer displays, so by enabling a dual display layout, I get the SpeedGrade interface on one screen and the full-screen video viewer on the other. To do this you have to correctly offset the pixel dimensions and position for the secondary display in order to see it. Otherwise the image is hidden behind the interface.

Using Mercury Transmit, the viewer image is sent to an external monitor, but you’ll need an appropriate capture/monitoring card or device. AJA products seem to work fine. Some Blackmagic devices work and others don’t. When this works, you will lose the viewer from the interface, so it’s best to have the external display close – as in next to your interface monitor.


df1615_sg_3When you use Direct Link, you are actually sending the Premiere Pro timeline to SpeedGrade. This means that edits and timeline video layers are determined by Premiere Pro and those editing functions are disabled in SpeedGrade. It IS the Premiere Pro timeline. This means certain formats that might not be natively supported by a standalone SpeedGrade project will be supported via the Direct Link path – as long as Premiere Pro natively supports them.

There is a symbiotic relationship between Premiere Pro and SpeedGrade. For example, I worked on a music video that was edited natively using RED camera media. The editor had done a lot of reframing from the native 4K media in the 1080 timeline. All of this geometry was correctly interpreted by SpeedGrade. When I compared the same sequence in Resolve (using an XML roundtrip), the geometry was all wrong. SpeedGrade doesn’t give you access to the camera raw settings for the .r3d media, but Premiere Pro does. So in this case, I adjusted the camera raw values by using the source settings control in Premiere Pro, which then carried those adjustments over to SpeedGrade.

df1615_sg_2Since the Premiere Pro timeline is the SpeedGrade timeline when you use Direct Link, you can add elements into the sequence from Premiere, in order to make them available in SpeedGrade. Let’s say you want to add a common edge vignette across all the clips of your sequence. Simply add an adjustment layer to a top track while in Premiere. This appears in your SpeedGrade timeline, enabling you to add a mask and correction within the adjustment layer clip. In addition, any video effects filters that you’ve applied in Premiere will show up in SpeedGrade. You don’t have access to the controls, but you will see the results interactively as you make color correction adjustments.

df1615_sg_17All SpeedGrade color correction values are applied to the clip as a single Lumetri effect when you send the timeline back to Premiere Pro. All grading layers are collapsed into a single composite effect per clip, which appears in the clip’s effect stack (in Premiere Pro) along with all other filters. In this way you can easily trim edit points without regard to the color correction. Traditional roundtrips render new media with baked-in color correction values. There, you can only work within the boundaries of the handles that you’ve added to the file upon rendering. df1615_sg_16Not so with Direct Link, since color correction is like any other effect applied to the original media. Any editorial changes you’ve made in Premiere Pro are reflected in SpeedGrade should you go back for tweaks, as long as you continue to use Direct Link.

12-way and more

df1615_sg_5Most editors are familiar with 3-way color correctors that have level and balance controls for shadows, midrange and highlights. Many refer to SpeedGrade’s color correction model as a 12-way color corrector. The grading interface features a 3-way (lift/gamma/gain) control for four ranges of correction: overall, shadows, midrange, and highlights. Each tab also adds control of contrast, pivot, color temperature, magenta (tint), and saturation. Since shadow, midrange, and highlight ranges overlap, you also have sliders that adjust the overlap thresholds between shadow and midrange and between the midrange and highlight areas.

df1615_sg_7Color correction is layer based – similar to Photoshop or After Effects. SpeedGrade features primary (“P”) , secondary (“S”) and filter layers (the “+” symbol). When you add layers, they are stacked from bottom to top and each layer includes an opacity control. As such, layers work much the same as rooms in Apple Color or nodes in DaVinci Resolve. You can create a multi-layered adjustment by using a series of stacked primary layers. Shape masks, like that for a vignette, should be applied to a primary layer. df1615_sg_10The mask may be normal or inverted so that the correction is applied either to the inside or the outside of the mask. Secondaries should be reserved for HSL keys. For instance, highlighting the skin tones of a face to adjust its color separately from the rest of the image. The filter layer (“+”) is where you’ll find a number of useful tools, including Photoshop-style creative effect filters, LUTs, and curves.

Working with grades

df1615_sg_13The application of color correction can be applied to a clip as either a master clip correction or just a clip correction (or both). When you grade using the default clip tab, then that color correction is only being applied to that single clip. If you grade in the master clip tab, then any color correction that you apply to that clip will also be applied to every other instance of that same media file elsewhere on the timeline. Theoretically, in a multicam edit – made up of four cameras with a single media file per camera – you could grade the entire timeline by simply color correcting the first clip for each of the four cameras as a master clip correction. All other clips would automatically inherit the same settings. Of course, that almost never works out quite as perfectly, therefore, you can grade a clip using both the master clip and the regular clip tabs. Use the master for a general setting and still use the regular clip tab to tweak each shot as needed.

df1615_sg_9Grades can be saved and recalled as Lumetri Looks, but typically these aren’t as useful in actual grading as standard copy-and-paste functions – a recent addition to SpeedGrade CC. Simply highlight one or more layers of a graded clip and press copy (cmd+c on a Mac). Then paste (cmd+v on a Mac) those to the target clip. These will be pasted in a stack on top of the default, blank primary correction that’s there on every clip. You can choose to use, ignore, or delete this extra primary layer.

SpeedGrade features a cool trick to facilitate shot matching. The timeline playhead can be broken out into multiple playheads, which will enable you to compare two or more shots in real-time on the viewer. This quick comparison lets you make adjustments to each to get a closer match in context with the surrounding shots.

A grading workflow

df1615_sg_14Everyone has their own approach to grading and these days there’s a lot of focus on camera and creative LUTs. My suggestions for prepping a Premiere Pro CC sequence for SpeedGrade CC go something like this.

df1615_sg_6Once, you are largely done with the editing, collapse all multicam clips and flatten the timeline as much as possible down to the bottom video layer. Add one or two video tracks with adjustment layers, depending on what you want to do in the grade. These should be above the last video layer. All graphics – like lower thirds – should be on tracks above the adjustment layer tracks. This is assuming that you don’t want to include these in the color correction. Now duplicate the sequence and delete the tracks with the graphics from the dupe. Send the dupe to SpeedGrade CC via Direct Link.

In SpeedGrade, ignore the first primary layer and add a filter layer (“+”) above it. Select a camera patch LUT. For example, an ARRI Log-C-to-Rec-709 LUT for Log-C gamma-encoded Alexa footage. Repeat this for every clip from the same camera type. If you intend to use a creative LUT, like one of the SpeedLooks from LookLabs, you’ll need one of their camera patches. This shifts the camera video into a unified gamma profile optimized for their creative LUTs. If all of the footage used in the timeline came from the same camera and used the same gamma profile, then in the case of SpeedLooks, you could apply the creative LUT to one the adjustment layer clips. This will apply that LUT to everything in the sequence.

df1615_sg_8Once you’ve applied input and output LUTs you can grade each clip as you’d like, using primary and secondary layers. Use filter layers for curves. Any order and any number of layers per clip is fine. Using this methodology all grading is happening between the camera patch LUT and the creative LUT added to the adjustment layer track. Finally, if you want a soft edge vignette on all clips, apply an edge mask to the default primary layer of the topmost adjustment layer clip. Adjust the size, shape, and softness of the mask. Darken the outside of the mask area. Done.df1615_sg_11

(Note that not every camera uses logarithmic gamma encoding, nor do you want to use LUTs on every project. These are the “icing on the cake”, NOT the “meat and potatoes” of grading. If your sequence is a standard correction without any stylized creative looks, then ignore the LUT procedures I described above.)

df1615_sg_15Now simply send your timeline back to Premiere Pro (the “Pr” button). Back in Premiere Pro CC, duplicate that sequence. Copy-and-paste the graphics tracks from the original sequence to the available blank tracks of the copy. When done, you’ll have three sequences: 1) non-color corrected with graphics, 2) color corrected without graphics, and 3) final with color correction and graphics. The beauty of the Direct Link path between Premiere Pro CC and SpeedGrade CC is that you can easily go back and forth for changes without ever being locked in at any point in the process.

©2015 Oliver Peters