Now that our first 3D documentary is almost completed, compositor Jakub Kuczynski has time to give the lowdown on some of the challenges he faced editing Blowdown.
Challenge 1: Double-rendering
Because 3D is filmed in stereo pairs rendering for each eye is required. For a compositor with tight deadlines this process can be a painful one.
Challenge 2: Editing in anaglyph
3D compositing in anaglyph can be deceiving because it crushes depth perception.
The tendency is to compensate for this by creating more depth, but sometimes you can overshoot the mark – a discrepancy that becomes obvious when you view the footage on polarized monitors.
An extra step is then needed to make sure sure stereo pairs are aligning perfectly which, of course, means more time and inevitably more stress.
Challenge 3: Finding the happy balance between formats
Until that fine day when everyone is experiencing our documentaries in full stereoscopic glory it’s important to make sure shots work just as well in 2D as they would in 3D. That in itself is an art.
Here’s an example of one of Jakub’s most technically challenging shots – a 3D within 3D effect composited for Blowdown’s episode on the implosion of the Fonte Nova Stadium in Brazil:
And here’s how he brought it all to life:
But, of course, mastering challenges like these come with the rewards of creating 3D VFX everyone can get a kick out of.
It’s easy to tell when a 3D viewing experience goes sideways – sore eyes, headaches and a general feeling of awkwardness are unmistakable indicators.
Obviously we want to avoid these issues with our 3D content. And to fully understand how to do it right, you’ve got to understand how it’s being done wrong.
-Your eyes should focus easily and naturally when viewing in stereo. If you are getting headaches or your eyes cannot focus, improper alignment is the cause (DCR tip: Take off your glasses and try to spot an area of high contrast. You may see that a bright spot is a little higher for one eye than the other.)
-When items on screen are glowing or have an unnatural sheen it is due to exposure differences between the two cameras. It could be unmatched exposure or reflection issues with the beam splitter rig.
-Keep an eye open for bright objects. Glints, lens flares and spotlights create more technical issues that have to be considered during stereo capture. Glints off of metal objects can be messy and appear to be a different shape in each eye. Lens flares will “invert” and pull away from the viewer, which can be visually confusing. Spotlights can create star patterns that rotate differently in each eye making it uncomfortable to view.
-Try to determine if there is too much depth onscreen. Some say it is perfectly acceptable for backgrounds to be out of focus. Others maintain that if the viewer cannot easy focus on distant objects there is too much divergence. (DCR tip: Look for distant objects like mountains or spotlights; If the doubled-image of the mountain appears separated by many inches or feet the background is probably too far away.)
But if you can only view the material in anaglyphic 3D and 2D, all is not lost – use the anaglyphic 3D mode to analyze depth and the 2D mode to analyze footage quality.
Last but not least
If one of your 3D glasses lenses looks smudged it’s likely that the focus, zoom or shutters were not properly synced during filming.
And this time they’ll have the chance to actually see it in 3D.
It’s an obvious step – if you’re paying for 3D content chances are you’ll want to check it out at some point before you sign off on the show.
But with this emerging technology, even obvious steps are rarely easy.
So how did we get a cut out for the broadcasters to experience a full color 3DHD explosive demolition from the comfort of their plush office chairs?
First of all, our editor Brian Mann had to make sure all the shots in the cut were 3D legal and correct any colour discrepancies between the two eyes.
To create a comfortable 3D viewing experience, we also monitored the LR convergence throughout the edit.
This made for more work – dealing with issues that either didn’t exist before (converging shots) or that would have been tackled further along in the editing process (for example colour correction after picture lock).
To send this 3D version of the show to our broadcasters, we opted for two different delivery formats: a side-by side 3DHD version of the show on Blu-ray disc and a digital anaglyph SD version.
To create the Blue-ray disc we exported the CineForm 3D file from Final Cut Pro, burned it onto the Blu-ray disc via Adobe Encore that, unlike FCP exporting functions, allowed us to create more professional customized menus and shipped via courier.
Now all you need to view this form is any Blu-ray player (doesn’t have to be 3D) and a 3D-enabled monitor.
The anaglyph version was exported as a .mov file straight from FCP and then uploaded onto our server. It’s a seriously inferior experience, but will still give those who are unfortunate enough to not have a 3DTV handy a way to see it in some version of 3D.
Though you might want to get on that, dear readers. I’d hate to see you live in the flatlands of 2D when the 3D world is just within your reach.
As for the broadcasters … maybe someone should warn them— cause they’re in for quite a ride.
As 3D televisions (and eventually our first 3D documentary) make their way into living rooms near you, it’s time to lay down some helpful tips I found online to ensure you are getting the most out of your in-house 3D experience.
Ambience
In order to create the optimal 3D experience, you first have to create the environment for it to happen.
Dim the lights, cover the windows–black them out if possible. Essentially create yourself a tiny black hole.
By reducing ambient light you will eliminate ghosting and double vision that often botch the 3D experience.
When in doubt, darkness is always best.
Viewing Position
Until 3D televisions become 360 degrees of polarized celluloid, we’ll have to make due with flat screens that are better viewed head-on rather than from an angle.
Best then to keep the family or group of friends small so everyone gets a piece of the 3D pie.
Television Settings
Because 3D glasses are tinted, they’ll dim the movie if viewed in default settings made for 2D. So make sure to customize your settings and increase screen brightness.
3D is not normal television viewing—don’t forget it.
HD Screen
This goes without saying, but I’m going to say it anyway: 3D is a viewing format that belongs on an ample-sized, high-def screen.
The bigger the screen the richer your experience of the imagery—think GIANT pop-up book vs. OK pop-up card.
The only other thing I’ll demand from here on in when it comes to 3D screens is they be dust and fingerprint free.
Super frustrating news: after some serious testing, it seems that YouTube 3D-enabled videos do not decode properly for playback on our 3D monitor when sourced by a MacBook Pro or PS3.
For the majority of people this may seem like an inconsequential technological glitch.
But for me, it’s a huge setback in our mission to get top-notch 3D content out into the universe.
Where we’re at
But after seeing the same footage in stereoscopic 3D I just wasn’t satisfied – I wanted to give people with 3D-enabled devices the chance to not only experience the stuff we shot, but to judge its quality for themselves.
So I had my editing team upload a side-by-side version of interior shoot selects to YouTube:
It seemed like a simple plan – put the videos on and set them to 3D.
We managed to get anaglpyh playback working fine, with the correct aspect ratio and eye orientation. We then pushed forward to see if modern stereoscopic 3D methods worked.
They didn’t. When we tried to play the clips using half-width, side-by-side, 3D they were a no go on our JVC GD-463D10U monitor. We tried using both a MacBook Pro with DVI to HDMI and a PS3 with HDMI to HDMI.
We also tried to play three other 3D-enabled videos on YouTube that were uploaded by other people and came up against the same issue. I’ve embedded these at the end of the post if you’ve got 3D-enables gear and would like to give them a go.
What’s going wrong?
From what my team can tell, the problem seems to be that YouTube does not map the pixels properly for TV playback.
A huge caveat – and my call out to the 3D-enabled – this is not to say the YouTube 3D function definitely doesn’t work. It just doesn’t seem to work on the equipment that we have access to – a Mac computer, a PS3, and a passive filter 3DTV.
More equipment than 99.999999 per cent of the world has … but still. Our tests were not exhaustive.
The next move
This setback has put me in a difficult position. YouTube has the potential to offer a free 3D online playback solution that’s more comprehensive than anything else. And I really want to get our stuff on there so people can check it out – especially since more and more consumers are buying 3D-enabled viewing devices.
But YouTube just isn’t working for us. So where do I go from here?
We could host future 3D content on our own web server or perhaps on another web video community like Vimeo, but doing so would seriously cripple our reach. Missing out on YouTube is clearly a doozy when it comes to exposure … it’s one of the top-searched sites in the world.
And there’s another downside: neither our web server or Vimeo offer the ability to toggle between different 3D delivery formats that YouTube (in theory) could. This means we would have to render out and upload many different versions of the same video – more work for my team.
We can do further YouTube tests, and see if we can work around the issues we’ve encountered so far. But this means potentially re-encoding and uploading new videos. It will probably require a great deal of time to invent workarounds, render new files, and upload the new tests – especially since my edit crew also has to meet the demands of film projects that are currently in production.
Also, further testing at this point feels like it could be a gamble. As far as we can tell Google has very limited support for the feature, so our only option is try, and try some more, to see if there is something undocumented that works, or try, and try some more, only to discover that it really doesn’t work after all – at least not yet.
The other option is to abandon YouTube. That means abandoning any desire or investment to attain any of the benefit of having our 3D content on the mega site – at least until the 3D feature becomes more mainstream and (hopefully) functional.
Or finally, the wild card option: you, dear reader, have successfully watched this type of footage on YouTube, have the magic solution to this seriously irritating problem, and can’t wait to share it with us …
There were some hiccups when we first got the software, but we worked out the kinks and it’s been pretty solid ever since.
Now our compositor has added the program to his arsenal – and it’s paying off once again.
Before incorporating CineForm, our VFX team gave our editor ProRes videos. They would then to be transcoded into CineForm files and muxed – two extra steps for our editor for each video every single time.
Now, he can read, write and export CineForm 3D files in Adobe After Effects, and deliver them – already muxed – directly to the editor.
I’ve outlined this technique in previous post for Vixias. For 7Ds we:
Set both cameras to self-time/infrared mode, which allows us to use an infrared remote (The Canon RC-5, in this case)
Position our box designed to receive any infrared signal and transmit it through a split cable to two infrared sensors.
Tape the infrared heads at the ends of the split cable to the infrared sensors on the two Canons and then use the remote to start them in sync.
And yes, because of the internal clock circuitry disparities we don’t get a 100 per cent accurate start and stop, as the cameras may not start recording on the same frame.
Our stereographer, Sean White, has found our 7Ds can be out of sync by one or two frames at 30 FPS.
To get around this we’re doing a physical slate at the start of each shot for our editor – this way he can toggle shots by a frame or two and sync from where the slate is.
Sean’s tested this system extensively.
He’s found that once you sync clips at the start they stay in sync for up to 12 minutes straight – much more time than we need to get the types of shots we’re going to capture with the 7Ds.
I’ve received a few questions re. how we’re syncing Canon 7Ds we’re using to capture elements of our first 3D documentary … so I’m blogging about it to share with everyone.
We’ve used DSLRs to get high res stills at set intervals for time lapses for years, but, of course, never in 3D.
For this 3D episode of Blowdown, the explosive demolition series we produce, the crew will use 7Ds for these time lapses – and also for establishing shots of the condemned sports stadium in Salvador, Brazil and, of course, the implosion itself.
Here’s how we’ve brought this system into the third dimension:
1) Splice cable so there are two heads on one intervalometer.
2) Attach heads to timer remote ports on two Canon 7Ds on side-by-side rig.
3) Sync using the one handheld intervalometer.
Voila!
I’ll provide more details re. our time lapse strategy later: wanted to get this bit up ASAP.