Saturday, May 24, 2014

How to get a Cinematic look on amateur videos.


Long time ago, when all of a sudden video makers decided to adopt 24 fps as their preferred shooting frame rate, experts alike claimed that it was only by shooting footage at this rate one could get that special cinematic look in his/her work. It sounded peculiar, to say the least. By design, PAL video runs at 25 fps. How could one wee-little fps turn a plain vanilla PAL home video into a film-like experience? What were the ‘physics’ and ‘physiology’ of the human perception that justified that claim? I initially assumed «they’d know better, wouldn’t they?» and went out to shoot at 24 fps too, just like the ‘pros’. Obviously, my shoots with my regular camcorder made my footage look anything but cinematic. Never mind the 24 fps. There must have been something more to it, I concluded. I was initially, in all my modesty, so convinced that ‘they’ were right and ‘I’ was wrong, that I subsequently doubted my own visual perception of films and videos. I seemed to be too ‘incapable’ of experiencing it. I could of course see the difference of ‘film’ based movies vs. regular TV programs, like news, reality TV, sports events, and the like. In general, lightly edited video footage with no special effects, especially news videos, looked very much like footage from better quality security TV cameras, if you know what I mean. Very truthful and real. However, ‘real’ is the single thing that fiction movies on film have never been exactly! Except for documentaries. Viewers see the difference. But still, the claim about the magic properties of 24 fps remained quite intriguing to me. Would that be any true at all? Were there ‘experts’ out there seeing the difference? Or was it again a case of the ‘emperor’s new clothes’? Someone launched the idea and the rest of us simply pretended that we indeed saw the difference... I just couldn’t grasp the logic. To this day, I can’t.

Recently, I decided to explore this further, and watched a training session about the very subject of cinematic look and feel by Videomaker . ‘Make your video look like film’, it was called. And the case was eventually and largely demystified to me.

Let me share here what I learned. The bottom line, there’s much more to the cinematic look than meets the eye. At a given moment, one of the speakers attempted to justify the 24 fps myth, but what he mentioned sounded like Alice in Wonderland rather than being a rational clarification at all. Let’s just forget them 24 fps, then... it’s got nothing to do with the so called ‘cinematic look’. It’s a long line of other reasons that make the difference.

I’d like to start by making a point about movies. In general, movies are traditionally a lot more disciplined that TV shows. They are not at all like TV ‘live’ events. That includes ‘reality’ shows too. Many of the traditional TV shows, especially the ‘live’ events, albeit well planned in advance to avoid surprise as much as possible, are still spontaneous and unpredictable. Not feature movies and TV series though. These are planned productions, designed, shot, edited and screened by experts long-long before they are released to the general public by the jungle drums of the Production and Distribution companies marcomms apparatus. Like, for years long before that. Movies have a plot and a story to tell, and, above all, an inspired and visionary message by the movie Director. All expressed in the cinematic look, and the atmosphere of the movie. The Director, assisted by the Cinematographer. The duo of ingenious artists, who make all this happen. In other words, movies are like fiction literature. There’s much more, thus, in a movie sequence of scenes, concealed between cuts, acting performances, scripts of narratives and actor body language. In movies, everything happens for a reason. Nothing is accidental or left to sheer chance. Even incidental street crowds are paid extras. However natural and unrelated they might look to us, the viewers.

Live TV programs are not like this. Lighting is abundant and seldom atmospheric, rather fun, happy and gay looking, many simultaneous cameras are handled by operators of cranes, off ground steel wires, dollies and steadycams, cameramen on headphones for receiving instructions about their live shooting, with the stage manager directing mixer panel operators which camera must be ‘on’ at any given moment. Everything is out there, revealed to us all, creating the sense that we, the TV viewers, are actually inside the studio too. That’s what ‘live’ audiences are by design meant to do. They are simple people like the rest of us, sitting in a studio and watching the show first hand. It’s like a theatrical play, but without concealed parts. Everything happens in the rhythms of a live event. In fact, this fundamental difference between live TV shows and feature films (incl. TV series) lies at the basis of what makes something look ‘cinematic’. I think...

There are of course quite a few more contributing factors that make a video shoot look cinematic, like 'professional' so to say. Vimeo is abundant with such short film projects by starting and ambitious gifted amateur cinematographers and film Directors. Here are some of these factors, following ‘my personal’ ranking, based on what I heard.
  • Depth of Field.
  • Consistent Colour Correction.
  • Colour grading.
  • Camera movements.
  • Cuts.
Everybody among the ‘experts’ seems to agree that the depth of field (DOF) and associated bokeh in footage frames are at the top of the rank of cinematic look contributing factors. Most home video cameras shoot with almost everything shown rather ‘sharp’, especially in long shots, other than zoomed-in close ups. Full frame DSLRs used with a large variety of prime and zoom lenses have changed that look. Canon 5D Mark II and III cameras as well as other popular DSLRs have changed home video for ever, and are now even being used by pro’s for their distinguished handling of the DOF. Thus, satisfying the prime requirement of cinematic look. However, in shootout comparisons between top professional cinema cameras (Red, Arri, Panasonic, Sony, Phantom, etc), produced by Zacuto in 2010, 2011 and 2012, the 5D recorded footage sharpness, object definition, and lens resolution were rather on the ‘lower’ side. This has also to do with the compression applied in-camera, even if the attribute ‘uncompressed’ is used in its codec, as well as the quality of the glass, it goes without saying. I have often created video slide shows based on pictures shot with my 5D MIII and compared them with footage shot with same camera and lenses, in the camera’s native AVCHD compression, and even recorded for better results by an Atomos Ninja Blade external recorder in FCP’s native ProRes 4.2.2. codec. Either way, the delivery format for both (slideshows and video footage) being 1080p for Apple TV devices, I experienced in dismay quite a difference in sharpness and resolution between the two. Meaning that even the HDMI output signal is not capable of providing the resolution that a traditional good quality Canon lens is able to do in regular RAW photography. Good to know. 

When the footage becomes available to post-processing in one of the native codecs handled by Professional video NLEs (non-linear editors) like Avid, Vegas, and Final Cut Pro (FCP), then colour correction and grading become the subsequent cinematic look contributors. Through colour correction and matching one ensures the common look among shots, which is absolutely vital in making a home video look consistent and pro throughout. Colour grading is finally adding to the project’s special ‘atmosphere’. Do you want your ‘film’ to look like The Matrix, or CSI:Miami, or The Minority Report? Or even Django Unchained? There’s a preset for it. Either offered ‘free’ inside the NLE app or by acquiring plugin filters from suppliers like Tiffen. In conclusion, the colour look of a video project is very much a cinematic contributor!

Camera movements is next. There’s no way your work could look cinematic or professional if your footage is shot in handheld fashion without stabilising equipment at all, and all you can do is zooming-in and -out creating the impression you are shooting in the middle of a 8.0 Richter scale earthquake (especially in the zoom-in stand). Tracking and dolly camera movements add a lot of pro look and feel, especially if your glider/slider is in the vertical position, resembling a ‘crane’ shot, and apparent parallax effects are candidly visible in your shots. Motorised gliding adds even more pro quality.

Slow-motion is an additional bonus. Especially in shots where in part of the shot duration playback gets accelerated by dropping frames (4X is a good start), and the remaing part gets slowed down by the same factor (25%). This look is very pro indeed, and is often shown on TV ads. If a crane effect gets involved too, it becomes even sexier. It’s like your camera flies around a subject pointing to it, very fast indeed, and then brutally engages the slo-mo breaks! I love this style. Nowadays they even sell drones with embedded action cameras that are capable of shooting 60 fps in 1080i/720p. This is a frame rate that by definition, when conformed to 24 fps, offers a very smooth 25% slow-mo without even any added frames interpolation. If you add Twixtor effects to footage like this, then you can easily obtain slow-mo effects of qualities comparable to a Phantom 1000 fps capable professional camera! This is not always true though, as Twixtor often adds induced artefacts. The video mode recommended for best results here is ‘optical flow’. In this mode however, when subsequent renders reveal artefacts, interpolated in-between frames look too artificial and cause the shot  to look like ‘morphed’ and unreal. If it wasn't the case, why else should you go spend 100s of thousands of $ on a Phantom, then? Nevertheless, devices like the latest iPhone and point and shooters like the Canon S110 offer descent slow-mo with higher than 60 frame rates, the Canon even creating footage at 240 fps. The S120 offers higher resolution (720p?) but only reaches up to 120 fps. Very cool slow-mo that is, trust me.

Last but not least, it’s the cuts. The real pro look of cuts you can especially observe in multi camera shots of scenes with lots of actor narration. "In the Blink of an eye" is a classic book (Amazon.com) on how to do cuts. Great film editors are distinguished by the ways they handle and resolve them. Cuts are like typography. When they are effective, you'll never notice them. They come simply natural and help the story move on. No more no less. I am personally obsessed with cuts, and I have very little experience about how to do them well. A known Cinematographer (A.S.C. level) offered during the Emmy award Zacuto documentary ‘Light and Shadows’ the following advice: «...always pay attention to other people’s cuts and learn from them. Learn what works and what doesn’t. Try and try again». One such case is narration with multicam shooting. After syncing the shots, editors offer different perspectives of the same stage by slicing and dicing camera shots and presenting multiple cuts of the event. To show the speaker and his/her listeners, as well as their reactions and body-language. The best cuts occur when a narrator changes sentences. The moment he/she pauses to take a breath. Typically not in the middle of a sentence and definitely never in the middle of a word! Also, cutting on the beat of background music, especially in music videos, is a must!

And a final remark about shot duration. Many amateur videos suffer from shots lasting far too long. Like, seconds long, even minutes. This way, home video looks very much like security footage from a CCTV system. Any frames not adding value (additional info) should be thrown away. You should of course record longer shots than you'll use to obtain long enough handles and be able to select the best parts in post, but your selected between cuts duration should be far shorter. Professional editors, especially when cutting narration multicam shots, often keep duration below the 3 secs, with ‘action’ cuts often below one second! I won't be surprised if someone told me that, we, as viewers, are practically able to experience even one single frame shots (!) flashing in front of our eyes. Professional editors do often work in sub second territories for emotional ‘effect’ in action sequences. If you still doubt, use a stopwatch to time them next time you watch a MTV clip, if you can at least, and see how right I am.

Monday, May 5, 2014

Amateur video with Pro ambitions

5D Mark III with Ninja Blade (shot with an iPhone)
Almost 30 years ago I became quite interested in electronics equipment for consumers for shooting home video. My kids had the right age, and the idea that I could 'create' short video clips, and them being able, years later, to watch themselves on TV, provided the necessary motivation to spend a shedload of dough and effort on this feat. It was around 1986-87 when I realised that home video had already been on the market for a few years under the VHS recording standard and using rather bulky 'shoulder supported' equipment. I joined the crowd about around the launch by Sony of their own proprietary 8mm capturing system (was it really called that way?), whereby traditional recording technologies were replaced by CCD capture elements that promised far better quality. Anyways, the following ten years I shot a lot of family footage and even edited and created few descent clips with titles and background music and all that, based on analog tape recorder workflows. Although my gear was bulky and quite expensive, its quality was barely 'fine' (very low resolutions, hardly reaching 300+ horizontal lines at best), not to mention the fact that during analog workflows each additional copy of source footage had its quality degraded to the point it couldn't even fool a newborn in its cradle... Sheer unacceptable. Also, the lack of timecode and genloc capable equipment at consumer prices, absolutely necessary to synchronise two source signals in order to create simple 'resolve' transitions, meant that only plain cuts were possible. As for those cuts, frame accuracy was a dream for the future. We were happy enough with sec accurate cuts... Editing was a handcraft then. You had to know the sec you had to hit the stop and record button. My Gosh, I spent hundreds of hours in frustration.

Nevertheless, even so, the clips came out relatively reasonable (much we knew then) to watch on our TV sets of the time. Happy to watch ourselves on TV was a real motivator to keep us going! Something only the privileged could afford, TV anchors, actors, and artists; reality TV had not been invented yet those days. Those amateur clips look very much like a bad joke compared to what even amateurs of the worst kind can achieve nowadays with little money to spend on equipment. Come to think of it, it's simply beyond imagination the delta achieved in home video quality that we have seen in the last 30 years indeed. This was all made possible of course with the advent of inexpensive computer equipment, digital video processing with affordable capturing hardware and editing software and last but not least, the recent appearance of those big ass HDTV sets, which, in less than ten years, have led to the mercy of junkyards anything related to CRT built TV sets.

Indeed, today's 'amateurs' are challenged with a much more luxurious feat of producing video in Full HD (some are even modestly trying the new 4K standard), using relatively inexpensive cameras with possibilities that would only feel like wet dreams to pioneering amateurs like myself, if we were told then what was to come! The 'age of analog' as one could term the 80ies and the 90ies, was one where no digital captures and edits existed (certainly not in the consumer world), everything was recorded on lousy and rather expensive tapes, when quality suffered at least 40% into a first gen copy and 60-70% into the second gen (with the 3rd gen only good for junkyard viewing), when Photoshop and Premiere were slowly and modestly revolutionising the early digital editing space, and when the commercial TV channels worked with Broadcast TV standards that barely provided, if I remember well, any more than 576 lines of horizontal picture resolution, only possible to achieve by using millions worth of studio equipment. It was the video Dark Ages then, it is the Age of video and digital media Enlightenment right now (called Multimedia in the nineties, remember?). Don't ask me what it will be in another ten years, I can't think of anything reasonable to speculate about a future even that close. That's quite true if only one looked at the progress we simply witnessed in the last 4 years, even in the aftermath of our recent financial tsunami.

What we are seeing in today's professional video and movie industry (for ads, SF TV series and feature movies in particular) is the excessive use of a 3D rendered 'reality', making us believe that we are in fact 'seeing' and 'hearing' everyday objects, landscapes, animals, plants and humans, whose only actual 'real existence' are stacks of megabytes in binary storage, interpreted by computer apps and lightening fast hardware. All made by software developers for us to enjoy with our real senses and eventually made to believe they are as 'real' as anything else around us. The new 'reality', call it virtual, fake, augmented, whatever you like, is there to be. It is the intellectual product manufactured by quite an ingenious 3D developer community, who turned their ideas and dreams into factual user-interface outputs that hit our visual and hearing senses as if they existed indeed. In this sense, 'The Matrix' was quite a visionary achievement about our world of the future, one has to admit for sure.

If you come to think of it, any given 'amateur' today, with not so terribly deep pockets, but with the right motivation and skills, can simply undertake a low budget video production hands down, which, with the exception of additional 3D rendered reality and effects, can challenge any professional studio otherwise, and that in terms of visual quality of his/her production. Everything is out there to go buy, at least if you know what you're doing, and in case you don't, simply google it. It's not only accessible to the happy few studio professionals, but also to gifted amateurs of all sorts. Full HD and even 4K capable cameras, studio lighting, sliders and gliders, cranes and dollies, professional digital capturing devices, super computers, peta level storage cloud servers, whatever. It's all there to grab, produce, shoot and edit your marvel and put it out there for potentially the whole mankind to watch... Like we saw in that hilarious Korean singer in a video clip that's gone viral not too long ago. YouTube and Vimeo are loaded with self-made productions in the thousands, not to dare say even millions. I don't mean useless smartphone snap amateur video of humans making fool of themselves trying to get their five minutes of fame, but reasonably serious video, with some good footage and maybe a modest plot to go along. A documentary perhaps? You name the stuff you need and Amazon.com with eBay, together with hoards of digital online suppliers simply got what you need and will sell to you for money, delivered to your doorstep 'courtesy' of UPS, DHL, DPD, TNT, and the likes.

A few such specialized items for video capture, normally used by Pro video shooters, are manufactured and sold by ATOMOS. I came across them during a recent Photography Professionals Industry Fair in Brussels. I had no idea what they were talking about and the whole fuss around it. Until I got closer, that is... Their products are simply professional digital recorders that capture directly onto high capacity disks (preferably SSD) the video signals of a large variety of video camera equipment, Professional and Prosumer, including many DSLRs. They do this via HDMI or via common for Professional equipment so-called HD-SDI connectors. The ATOMOS recorders are basically compressing the captured signals quite loosely or even store them uncompressed. Today's Terabytes are much cheaper than the Gigabytes 15 years ago, so who cares? They typically use Apple's ProRes or Avid's DNxHD codecs, and by so doing they can preserve the maximum possible quality in the video signal, becoming literally the RAW equivalent of video, or, let's say real close.

Having recently decided to freshen up and update my video skills after many idle years, and maybe end up shooting some descent video footage again, now that I got plenty of retirement time, I purchased the ATOMOS Ninja Blade recorder to attach to my 5D Mark III, which is considered to be one of, if not, the best DSLR camera for mobile professional footage capture, admitted even by the industry experts. I have also completed an online Lynda.com course on FCPX fundamentals (a great package FCPX it is, if only you knew how-to use it) and even added Apple's Compressor to my editing kit for my export encodings. I am typically obsessed about image sharpness and resolution, colour quality, and smooth video movement, so I had to try for myself what is practically available in the 'gifted' amateur space to achieve the best possible results. Initially, I struggled a lot to get this stuff working at all, but I've eventually been able to get a result real close to what I hoped for. You can watch a test here for yourselves (it's just a test of my so many). Select its highest HD quality available and preferably watch it full screen on your Mac, or even better on a HDTV set, if it's capable of Youtube playback. Mind you though. YT needs to recompress uploaded footage too, and the quality damage they caused in this test clip is in the range of 25-30 %, I'd argue... So, don't get too excited, my original clip that came out of Compressor configured for the Apple TV device is way better than what you'll watch here!