Final Cut Pro X does a number of wonderful things. It can analyse your footage in all sorts of fascinating ways to help you fix problematic audio, to colour balance your shots and stabilise that wobbly footage. With a couple of third party stabilisers now on the market promising to provider better and faster results than FCP X’s built-in stabilisation, we thought we’d ask Chris Roberts to provide a steadying hand…
We’ve all had that shot. The great shot filmed by the cameraman hanging out the back of a moving vehicle (health and safety assessments all filed correctly, of course). Or the shot of the VIP arriving from above the heads of other onlookers. Or the awesome drone footage of the cyclists coming down the mountain or the GoPro POV shot. Such a shame they’re a bit wobbly. And the director wants to know what you can do about it.
There are now three main options for stabilising your footage within Final Cut Pro X: the built-in FCP X stabilisation options, Coremelt’s Lock & Load plugin and - the new kid on the block - CrumplePop’s BetterStabilizer. The latter two options are available for $99 each - Lock & Load from the Coremelt website and BetterStabilizer through FX Factory.
For full disclosure, I am a fan of Coremelt’s FCP X plugins and bought Lock & Load some time ago. I have used the FCP X stabiliser and Lock & Load on numerous occasions. For the purposes of this review I was provided with an NFR licence of BetterStabilizer.
(And just for the record, yes, I am fully aware of other solutions such as Premiere Pro’s Warp Stabilizer and DaVinci Resolve’s excellent tracking features. But my brief was stabilisation in FCP, thank you very much. I’m also a fan of, where possible, working within the edit application rather than needlessly round-tripping.)
Both the third party offerings promise better results much quicker than the built-in options in FCP X. But how much better are they (if at all?), and do you really need to purchase something more than what FCP X can offer?
But, before we delve into these deep questions though, it’s always useful to take a moment and consider why we would need to stabilise a shot in the first place.
So, what is stabilisation and what are we trying to achieve with it?
In my opinion, stabilisation should be used when you need to include a particular shot in your edit that might have been filmed under less than optimal conditions. It might be that there just wasn't the time or opportunity to set up a tripod. Or maybe it’s filmed from an angle where it just simply isn’t possible or practicable to use a tripod, especially with the proliferation of drone or other GoPro-type footage.
Or maybe the shot was captured on a long lens, or even macro, magnifying even the tiniest movement of the camera. Plus, the popularity of DSLRs and other cameras like as the iPhone mean that we’re seeing shots that jitter and shake more and more often, especially if they don’t have some kind of optical stabiliser built-in.
When should we consider using stabilisation in post? Well, that’s always a subjective thing of course. Your director might be happy with a wobbly shot that you just feel distracts from the story, or vice-versa. I’ve also used stabilisation quite successfully to lend my edit the impression of a higher-level of production that was actually available on our budget. Whatever your reasons for using stabilisation, you should do so wisely and for the good of the show. “We’ll fix that in post” should never be the solution on location…
When I considered how I was going to be able to compare these options I thought it best to look at different factors in turn. The first test was going to be the speed of the analysis. Unsurprisingly, none of the three contenders seemed to detail just how that analysis works - I guess that would be like revealing the secret sauce. Nevertheless, speed is one of the major selling points in the marketing of both Lock & Load and BetterStabilizer, so this would be an out-and-out showdown.
The next consideration is results. You may have the fastest analysis going, but if the results aren’t good enough, then what’s the point? This falls in to two categories - firstly, how good are the results off the bat, based on the analysis? Secondly, how can we tweak these results to give us better overall results?
So, my tests are (in this order): Speed; Default Results; Advanced Settings.
Ready to go?
First up, a quick comparison of the different tools. What are the general advantages of each?
FCP X’s built-in stabiliser has the huge advantage of being available with no extra cost. If you’re using FCP X, you have a stabilisation option. This is also important when considering the use of any third-party effect, title or generator: if you’re sharing your library with anyone else, does that other person have access to those same third-party plugins? If not, then expect them to be on the phone asking why FCP X is showing them a red warning sign.
Also, never neglect the FCP X online help. In this case, the manual’s not hugely informative about how the stabiliser in FCP X works, but it certainly has some interesting points about what each of the controls affect when adjusted (twirl down the Advanced Stabilization Settings on the above link).
Putting aside for the moment the boast of being faster and, well, better than FCP X’s offering, BetterStabilizer also has the advantage of being an applied effect. Why is this important? Well it means you can reorder the stabiliser with other effects if needed. I’m struggling to give a practical example of that here though one main advantage is that BetterStabilizer can be applied directly to both compound clips and multicam clips without having to open either and apply the stabilisation to the contents. Indeed, the whole workaround to stabilising part of a longer multicam angle without BetterStabilizer is to open the multicam angle editor, find the bit you want to stabilise, blade the clip in the angle to isolate the wobbly part, then stabilise just that section. Phew! Still with me?
Long story short: if I had a multicam clip I need to stabilise, I can apply BetterStabilizer directly to the clip on the timeline just like any other effect, which isn’t something you can do with FCP X’s stabiliser. It also works just like any other effect on a multicam, so if you switch the angle then the effect disappears from the clip but reappears if you switch back. It would also appear that the clips doesn’t need to be re-analysed. Seriously, this would have saved me so much time on a job recently!
Lock & Load is also an applied effect though, despite allowing you to apply it to a compound clip, it won’t allow you to use it on multicam clips. However, where Lock and Load has an advantage is in its more advanced settings, including allowing you to focus on a Region Of Interest.
So, on with the first test.
Again, sorry for the interruption, but just so you know that all these tests were carried out on the following system: iMac (Late 2012) running OS X El Capitan 10.11.6 with 32GB RAM and NVIDIA GeForce GTX 680MX GPU with 2GB VRAM. All results were reviewed by rendering and playing back in realtime with FCP X set to Better Quality and outputting to an external monitor at 1080p using Blackmagic Intensity Pro 4K. Unless otherwise stated, the footage is 1080p25 in projects set to the same resolution and framerate of the source media. None of the footage was supplied to me, it was all real-world footage shot for different projects.
TEST 1 - Speed of Analysis
For this test, I just wanted to see how quick each of these stabilisers was at analysing some footage - could both BetterStabilizer and Lock & Load live up to their boasts of being faster than FCP X?
I decided the best way to demonstrate this was to screen-capture this process as I analysed the same piece of 38 sec and 3 frame footage using each of the stabilisers.
As you can see from the above video, the results speak for themselves:
FCP X: 11:09
Lock and Load: 12:20
So Round 1 to FCP X it seems (just). But BetterStabilizer also has another distinct disadvantage here - it’s not a background task. Both FCP X and L&L can analyse the footage whilst I’m doing other things but, when BetterStabilizer is analysing, everything stops, which is a huge shame and could become a real drag if you have to re-analyse footage when changing settings.
Remember, this is just simply which is faster, not the quality of the results.
(Editor's Note: The results for BetterStabilizer seemed excessive, so we got in touch with CrumplePop who said they are aware of some issues and there is an update coming out soon with a significant speed increase)
TEST 2 - Default Quality
Ok, for this next test I used a piece of footage I often show during my training. It’s a 3sec 4frame shot of a duck. I wanted to see what each made of this piece of footage because I know it and know the results I’ve had before, so it’s a known quantity for me.
Again, the results are posted in the video below:
As you can see, FCP X’s stabiliser does a great job of taking much of the shake out of the shot making it more than useable. When FCP X has finished analysing the footage though it applies one of two settings - either Inertiacam or Smoothing (FCP 7 users will recognise the same settings for Smoothing as were available in the old Smoothcam filter). The “Automatic” setting is simply FCP X’s best choice of the two. In the Interiacam setting though, if FCP X deems it possible, it may give you the choice of enabling Tripod Mode. This isn’t always the case, but if you do see this option, it’s worth taking - and as you can see, it has a dramatic effect on our duck shot.
You’ll see that both BetterStabilizer and Lock & Load both do a similarly reasonable job on this shot too, though I think L&L probably edges it slightly. Both of these have an “autoscale” feature which minimises the amount of scaling taking place by adjusting the zooming to remove the black borders dynamically. You can see this slightly toward the end of the L&L example when it seems the camera “zooms out” a little.
BetterStabilizer doesn’t have a “tripod” option but L&L does - it’s called Lock Down from the Stabilisation Mode pop-up. Again, you can see the results of this aren’t quite the same as FCP’s as there’s more movement due to the Smart Scaling mode. Changing this to Fixed Scaling produces a result similar to FCP, as seem from the 4-Up display at the end.
(Right click for larger images)
I think in this test, L&L and FCP both come out on top, with L&L nudging in front slightly. However, FCP X’s tick box for Tripod mode simply feels more intuitive than L&L’s settings.
One more thing to mention regarding this test, you’ll notice a slight “smearing” of the image over two or three frames near the beginning of the shot - this is where the camera is moving too quickly and the pixels have “blurred”. You’ll notice that each of the stabilisers produces the same results here and is a good example of why I’m trying to use real-world footage rather than supplied demo clips. I understand that this issue could be minimised by using a higher shutter speed during filming. Worth noting.
TEST 3 - MEERKATS
Let’s give these something a bit more complicated. For this next test, I’m choosing to use a shot of a meerkat (aaaahhhh). This was to simulate an interview clip where the subject is moving independently from the background. This piece of footage is a 720p50 clip.
Again, the video below gives full-screen and split screen comparisons:
Right FCP does an ok job at this shot, but do you see that “pulsing” background on the shot? I find this more distracting than the original to be honest!
As for BetterStabilizer, well I have to say, I think the results here are spectacular. I often use this shot to highlight the kinds of problems associated with stabilising in post, but this is the best result I have ever seen on this example.
As for L&L, it seems that this is suffering with the same problem FCP had - the movement of the background is causing all manner of problems that detract from the image.
Big points for BetterStabilizer on this test.
It’s all very well looking at these results, but every shot really is different - much like colour correction, when stabilising shots each has it’s own unique movement both in terms of the subject and the camera. Indeed, the results I’ll get here probably won’t necessarily be reflected in your footage. The best way of achieving the desired results is often to manually adjust the settings yourself.
The below video is the same meerkat clip but adjusted manually:
You can see that compared to the previous example a little tweaking has made the FCP and L&L stabilised footage much more acceptable.
So what changes can we make?
A good technique I’ve found for this is to mark the clip you are trying to stabilise in your timeline, set FCP to Loop Playback and press the backslash key to Play Selection. Now when you adjust the controls in the Inspector, you’ll see how your changes are affecting your clip for better or worse!
Well, FCP’s automatic choice for Inertiacam didn’t help the footage at first. No choice for Tripod Mode means the only option we have is to increase the amount of smoothing. But, if we change the Method from Automatic to Smoothing, we get a few more controls.
Translation Smooth controls the horizontal and vertical movement of the shot
Rotation Smooth controls the rotation of the clip (as if the cameraman is tipping the camera left and right rather than panning or tilting)
Scale Smooth controls smoothness of any zooming for tracking into or out of the subject.
Because we don’t move in or out of our meerkat interviewee, I’ve reduced scale smooth to 0. I’ve also lowered both translation and rotation smooth values from their default 2.5 values to minimise the pulsing of the background. There’s still a fair amount of wobble in the shot, but it feels more calmer than before.
I didn't adjust the BetterStabilizer from the previous example, I thought the results were too good - but looking at the settings we see there are much simpler controls. There are a number of presets which you can select based on how your footage was shot which seem to adjust the Stabilise Skew, Autoscale and Strength controls. I’m assuming the skew option is similar to the rotation control above, but it might be to do with rolling shutter, I couldn’t find any documentation about it to be honest.
One limitation I’ve noticed here is that these simple controls are either on or off, with an overall strength slider that is limited to a maximum of 100. In the FCP X stabiliser and L&L you can increase the values above the slider limits by dragging the numbers up or down.
But you can really go to town on adjustments in Lock & Load. Here we can adjust the overall strength of the stabilisation as well as control the horizontal movement independently from the vertical, whilst controls for rotation and zoom seem to reflect similar controls in FCP X. You can even adjust the speed L&L adjusts the smart scaling - in fact I found my lowering this value from its default gives better results by slowing that scaling speed.
I tried these techniques to stabilise another shot from a project I worked on a few months back:
In the above case, each stabiliser was manually adjusted to try and get the best results. You will notice the same “smearing” issue on some frames as we experienced on the duck clip.
Finally, if you’ve stayed with me this far it would be remiss of me not to mention Lock & Load’s more advanced feature. In the Lock & Load Effects browser you’ll notice there’s a option for “Advanced Tracking (ROI)”. This means that you can set up a Region Of Interest using a rectangle to focus the analysis on to either just this area (inside) or to exclude this area (outside). This is one of those more advanced features that is often worth the price of the plugin alone and it’s great to see it available here as part of the standard plugin. You may want to spend a little more time reading the Lock & Load Help document on this to be able to get the best out of it.
So what have we learned? Well, hopefully I’ve tried to show you that, like most processes in video editing, there doesn’t really seem to be a one-click solution for everyones’ problems. I have to say that I usually find the built-in stabiliser up to most of the tasks that I give it, though understanding how to tweak the controls is just as important to get the best out of a feature.
The controls in BetterStabilizer would suit someone who’s more comfortable choosing from a set of presets rather than making detailed adjustments. I personally find these controls slightly limiting but I was amazed at its ability to stabilise the meerkat shot and may probably reach for BetterStabilizer the next time I have an interview that need stabilising.
Also, the fact that you can apply BetterStabilizer directly to an edited multicam clip in the timeline is a huge bonus and would have helped me out on a recent job had I known. It’s just a huge shame that the analysis isn’t a background task and can take noticeably longer than its rivals.
Finally though, Lock & Load would seem to be the tool for the professional to add to the built-in stabiliser. The controls in L&L mean you can really tweak the stabilisation to try and get the very best out of your shot. However, as we’ve also seen, the quality of those results often depend on a number of factors, including how it was originally filmed (the smearing seen on a number of the examples I’ve used).
Ultimately is there an outright winner? Unfortunately, as is often the case, it seems not. It would depend on the type of footage you’re trying to stabilise, the results you’re aiming for and the time you’re willing to tweak the settings. And though it sounds a little disingenuous to say this at the end of a lengthy review - and if you’ve made it this far I really hope that you’ve found it useful - but sometimes the solution to those shaky shots can be as simple as getting the cameraman to use a tripod more.