Achieveing Film-Look on Video

THE BIGGER PICTURE

There are no buttons on video camera that are going to give you the look of film. The reality is that the mechanics of the camera is just one small part of a much larger process. Think of it like a machine that is dependent on other components to function. If one of these components fails or is omitted, the larger machine will either stop working or produce unpredictable results. And it is so when looking into those various "components" that make up the "film look". Let's examine each of these components a little closer.

Before I continue, I will acknowledge that there is an argument that some of these components can be omitted and still the film look can be achieved. This is probably a fair argument but you can make the decision as to what is important and what is not once you have a broad understanding of the bigger picture. It's like the old saying "In order to break the rules, you must first know them” or something to that effect.

CAMERA MOVEMENT

If an audience sees a lot of shaking in your footage, it triggers a chemical reaction in the recesses of their brains called videopsychosis. This signal lets the brain know that what they are seeing is footage shot with a relatively small video camera. If you are trying to fool your audience into thinking they are watching a film-originated movie, this is an undesirable result. To avoid this phenomenon, the camera operator must adjust his/her thinking and become a cinematographer, not a videographer. The “cinema” in that title is there for a reason.

So what can we do with the camera to give the illusion that your film was shot on film? Let’s add some virtual weight to it. Slow everything down. Imagine that you are operating a big 35mm camera  and you will begin to see the world a little differently. Study the slow camera movements of your favorite films. With some exceptions, most camera movements are slow and considered, often revealing details in a scene in a slow-mannered way. Adding some actual weight to your camera can really help until you get the hang of doing it with a lighter setup. When you pan, turn your body, not just your arm. As you become more fluent with your camera’s movement, your scenes will become more organic and it will be a much more pleasing to your audience.
Also consider the fact that anything that moves the camera from its fixed position (e.g., a dolly or crane) helps to further the film look. It takes the viewer away from the patented two-dimensional look of video and creates a world that you feel you could literally step into.
There are plenty of techniques for creating slow cinematic camera movements using your video camera and most of them will come from your imagination, not a book, so start experimenting.
I’ll give you one example to get you started. I like to tilt my entire tripod on two legs until my camera almost reaches the ground. I then slowly pull it back to its upright position. This creates a very effective crane movement and you can get some really smooth moves with practice.
So, don’t just watch films, study them. Imagine you are the cinematographer and really observe how the camera is moving and how it reveals or obstructs certain things and how it ultimately motivates the story or scene. You will find that there is a very specific language of movement and once you get a feel for that, you will begin to see things very differently.

LIGHTING

Everything in a scene should be there for a reason unless you are just gripping and ripping or doing documentary work. Lighting helps to selectively draw the viewer’s attention to a person or thing. In a way, it almost works in the same way as selective focusing. If something has a dominant light in a scene, it will receive your attention over everything else.

Lighting also creates the illusion of a third dimension in two-dimensional space. When we look at a movie, whether it is on a TV or on a theater screen, it is flat, having only dimensions of height and width so the challenge is to convince your audience that space exists behind the screen.
Classic three-point lighting, for instance, has a number of functions that help us in our quest for that extra dimension. We have a key light, which creates the focal point for the viewer and extrudes the subject from the rest of the frame. Then we have the fill light, which eliminates nasty shadows but also has a sculptural function by helping to reveal the shape of the face and head. Finally, we have the backlight or rim light which separates the subject from the background, giving a sense of depth.
Expounding on lighting techniques in general is beyond the scope of this article and there are many resources on the Web for your reading pleasure. I just want you to think about the concept of light of varying strengths in your scene, each having a very specific role, all conspiring to create a feeling of depth and ultimately helping the viewer focus on the message you are trying to convey.

COLOR/GAMMA

These days, there is no particular color palette that can be specifically associated with film. There are all kinds of looks out there ranging from monotone to highly saturated.
When shooting video, no matter how black you make your blacks in-camera, the picture still ends up looking a little “milky”. This is because the full depth of the blacks has not been realized and this greatly influences the general richness of the picture in terms of color. A simple levels adjustment in your NLE of choice will really bring out some vivid colors.
In the levels control, move the black control slider to the right until it crushes your blacks just to the point of losing some shadow detail. Some people like to push it further to achieve a high contrast look…they also push the whites by sliding the white control slider to the left. I usually stay away from the white end because it looks more like overexposed footage rather than a stylistic look. If I need to brighten or darken the picture beyond that, I adjust the mid-tones. That way, information in your picture will not suffer. Aside from some basic color correction, that’s about all I do to my raw footage to get that rich look.
I do not white balance if I’m taking nature shots or shots that have no connection to each other. I prefer to color balance in post production. For instance, if I was to white balance during the sunset or sunrise, all of the golden light would be gone. If I was shooting a film or music video, on the other hand, I would white balance before every shot because continuity is crucial in these situations.
Film has much greater latitude than video meaning it’s able to capture more tonal information. You will notice with film-originated footage, overexposed areas transition more gracefully than in video. In the latter, your picture will blow out to white pretty quickly, leaving some pretty ugly transitions in your picture. This is a dead giveaway that you are shooting video. Some video cameras have a “knee” adjustment that minimizes this problem. If you set your knee to “low”, the transition from exposed to overexposed will be more gradual.
Some cameras also have a “film gamma” adjustment that helps give the illusion of greater latitude. I won’t recommend any of these settings only that you experiment and make up your own mind.

FRAME RATE

Motion cadence has a large impact on the film look. Before 24p became available to the general consumer, there were programs like Magic Bullet, Cinelook and DVFilm that would strip your 50i footage and convert it to 24p. While they did a fairly good job, there is just substitute for the real thing.
35mm film cameras run at 24fps and give a distinctly different look than video which is essentially capturing images 50 times a second…(video actually captures two fields to make up every “frame” so it’s really 25fps). Video looks “hyper-real” while film has a slightly surreal feel to its motion. This makes it ideal for storytelling because it can pull the viewer into a world that contributes to suspension of disbelief. Video, because of its hyper-real motion, can sometimes feel more like a documentary and have that classic “soap opera” look. At least in my mind, this is less than ideal for narrative storytelling.
There are many pros and cons for using 24p, 25p or 50i and each of them has specific applications, even if it’s just for an aesthetic choice. The bottom line is that if you want your video to closely approximate the feel of 35m film, shoot at 24fps.
Committing to that frame rate means you also need to be aware of its limitations. Fast motion can lead to stuttering or strobing which can be distracting to the viewer. Again, overcoming these limitations is another discussion for another day but know that they are exactly the same as what you would encounter shooting on film.

ASPECT RATIO

The term Aspect Ratio refers to the relationship between the width and the height of a picture. 16x9 means that the screen is 16 units wide and 9 units high. HD video and some SD video is shot with a 16x9 aspect ratio. There are many, including myself, who feel this widescreen aspect also makes footage look more filmic. I go one better and matte my footage to a Cinemascope ratio which is the same used on many blockbuster films today. It has a much wider aspect than 16x9 and gives the frame more of an epic feel. This is particularly effective in HD where there is a lot of resolution that is absent from standard definition video, particularly in wide angle shots.
16x9 is the standard aspect ratio in today’s world of TV and video.

35MM ADAPTERS

Another staple of the film look is having control over depth of field. Because of their small CCDs, consumer- and prosumer-level video cameras have a large depth of field meaning nearly everything is in focus. This can be a challenge when you are trying to hone the viewer’s eye on a particular thing in a busy frame.
35mm adapters allow you to use all the advantages of film lenses right on your video camera. The light travels through the 35mm lens and projects an image onto a small vibrating or rotating screen. The camera then records this tiny screen in macro mode.
Not only do you get control over depth of field but there is also an organic quality introduced to the picture. Gone is the trademark sharp edges of video (particularly HD) and what’s left is a pleasing picture that’s less hard-edged and more like film.

KNOW YOUR LIMITS

Your video camera is just that. It is not a  film camera. Know the differences and you are halfway there. Try to avoid high contrast scenes and you will get a better looking picture. If you can’t avoid these kinds of situations, invest in a graduated ND filter or a polarizer. Do research on the Web and find out what you can do to minimize problems when shooting in uncontrollable environments. When you can, use lighting to your advantage. Good lighting can bring out the very best footage possible from your camera and there will be times when you fool yourself into thinking it was shot on film. To shoot well is to know the limitations of your tools and find creative solutions.
There are many other elements that go into creating a film look including art direction, composition/framing and becoming intimate with the language of film. What I have listed above are just some fundamentals to get you started.

video compression

Compression is not the resolution of the video, but video resolution has a lot to do with how much information will need to be compressed. Common video formats are 720p, 1080i, or 1080p. The 720/1080 part is pretty straightforward, it simply refers to the number of pixels on the vertical scale of the image: 720 is a 1280×720 pixel image. 1080 is a 1920×1080 pixel image for each frame of video. That’s a big difference in the amount of information: a 720 frame has 921,600 pixels. A 1080 frame has just over 2 million pixels.

The i and p parts refer to whether the frame is interlaced or progressively scanned. To be (very, very) brief, progressively scanned is better, especially when there is a lot of motion in the image, but doesn’t make as much difference when objects in the image are fairly static. 1080p is the best of both worlds, but takes a lot of bandwidth to do (it’s overkill for web video or most television, for example). 720p and 1080i actually both take up roughly the same amount of bandwidth and are what are used for HD television as an example (some networks use 720p, some 1080i). 1080p generates significantly more data than either of the other two formats. That’s why many lower end cameras, storage devices, etc. can handle either 720p or 1080i, but not 1080p.

The other variable, that is not compression, but that does influence how much data must be compressed, is the FPS (frames per second) the camera records. Film cameras shoot at 24 FPS. Many, but not all, video cameras can shoot at 24 FPS for a ‘film-like’ look, but standard video is usually shot at 30 FPS and 25 FPS(Note: These standards refer to the US and UK and few other nations. There are other standards worldwide.) Some lower end cameras shoot at lower frame rates than these, and some high end cameras can shoot at much higher frame rates.

Obviously a 1080p image shot at 60 frames per second is going to generate a lot more data than a 720p image shot at 24 FPS. The bottom line, though, is video is generating a lot of information: 1 to 2 million pixels recorded 24 to 30 times a second, at 8 to 16 bits per pixel, plus audio is a lot of data to record. And, practically speaking, it has to be compressed somehow to make the file size manageable.

Compression and Bit rate

Simply put, bit rate (usually expressed as megabits per second, or Mb/s) is the amount of data recorded each second. After the camera (or computer) has done its compression thing, the file size will equal bitrate X seconds of video. If you dig around, you can often find the maximum bit rate that a camera, storage device, or processor handles. In theory, higher bit rate means more data is stored, which (assuming everything else is equal) means higher quality compression. But there are a lot of other variables.

Different cameras use different codecs (COmpression-DECompression algorithms) to compress the data. Your camera choice sets your choice of compression algorithm (more on this later), since different manufacturers have chosen different codecs. In general, compression algorithms are sorted into two general categories (this applies for audio and other data too, not just video) lossy and lossless. Lossless means that after decompression each pixel is exactly the same as the original, no data can be lost. There are no video cameras (other than a few amazingly expensive professional cameras) that record lossless.

Lossy compression isn’t an exact pixel-for-pixel match when uncompressed, but it offers much higher compression ratios than lossless compression. (The compression ratio is the size of the original video compared to the compressed video. Uncompressed video is 1:1. Lossy ratios can get very high— 1:200 compression isn’t unheard of for some heavily compressed video formats. The codecs used in video cameras offer better quality, but less dramatic compression ratios. More on the order of 1:50.) There are lots of different codecs in use today. The better codecs are usually newer and offer a higher compression ratio with similar quality image. For example, MPEG-4 gives a higher quality image than MPEG-2 at the same bit rate. Some high-end cameras, though, use less aggressive codecs with less compression to maintain the best possible image quality. In exchange for that, they require significantly higher bit rates to record their data.

How video compression works

There are two ways to compress the data in a video clip: Intra frame and inter frame. Intraframe compression takes each frame of the video and compresses it just like you would use JPEG to compress a still image (in fact one format, Motion JPEG, does exactly that). With intraframe compression every frame of the image is complete, although slightly compressed. This can be important if your video has lost a frame – since the frame before and after the lost frame are complete, not much damage is done. It’s also important when you cut and paste video clips – the video editing software needs a complete frame at the beginning and end of each transition. Intraframe compression, though, doesn’t really make the file size all that much smaller. Compression ratios of about 1:20 are about as good as it can do.

To really get more significant compression, video codecs also use interframe compression. The basic idea is simple. Video consists of multiple still frames, (anywhere from 24-60 per second typically). Interframe compression looks at each frame, compares it to the previous one, then stores only the data that has changed. Usually it doesn’t look at individual pixels, but rather at square blocks of pixels (less time consuming and resource intensive). But each frame in an interframe compressed video contains only the changed parts of the image.

But interframe compression brings a new problem: What happens when you’re sending this video to wherever (or importing it), and it skips a frame? If each frame is referencing the previous frame, you’re in trouble until the entire picture has changed. If you have a 3 minute clip of the same scenery, there would be a problem. And the same problem would occur, if you wanted to cut the scene halfway through that 3 minute clip: the frame at your transition wouldn’t be a complete frame, just a compression of the changes that occurred from the previous frame. And so on. The solution all interframe compression formats use is the key frame.

Key frames and long-GOP compression

Interframe compression codecs record a Key Frame every so often: a frame that contains the entire image data set, whether the scene has changed or not. The key frame is shot every x number of frames (usually 15) and that frame contains a complete image. The next group of frames (until the next key frame) is heavily compressed, containing only the changes from the previous key frame. Using this method, if you skip a frame, you only lose (at most) 15 frames before you’re good to go again (or until your next editing point). It’s still a relatively long time, but allows for a much smaller file size than intraframe compression alone. This key picture followed by several the compressed pictures until the next key frame is abbreviated GOP (group of pictures). Since there is a fairly long group of images grouped associated with each key frame, this is often referred to as long GOP compression.

A final note about frame skipping: it’s rare. In fact, it almost never happens when using quality equipment. Because of this, long-GOP encoding is usable and safe. Intraframe-only compression does protect against frame skips, but requires a lot more disk space (and a higher bit rate for the same quality image). Since video editing software can only cut at a key frame, some high quality video recording devices (like the nanoFlash that started this discussion) will record video with only intraframe compression (a half-second until the next key frame can be an eternity to a video editor), but the resulting files can get very, very large.

Luminance and Color Compression

Since the days when video was analog, luma (the black and white values) and chroma (the color) have been stored separately. Y’CbCr is how video is stored today, typically using a process known as chroma subsampling. Y’ (sometimes simplified to just Y) is the luma (grayscale) information. Cb and Cr each store a portion of the color information (like LAB color space in Photoshop).

We are less sensitive to color and very sensitive to the grayscale value of an image, so video cameras today discard some of the color information to further compress the video data. The proportions are usually shown as a ratio with 4, indicating no compression. Recording video at 4:4:4 would be ideal, but it takes up an enormous amount of space and isn’t feasible in most situations with today’s equipment. Top quality video formats, like XDCAM422 and DVCPRO HD, keeps twice as much Luma data as either color (Cb and CR) data in a ratio of 4:2:2. This reduces bit rate by 1/3 with very little image compromise. Other video formats such as HDV, AVCHD, MJPEG, and MPEG-2 (DVD quality) use even less chroma data, storing video at a ratio of 4:2:0. This may sound extreme, but DVD quality video is recorded at 4:2:0, so it is intentionally missing 3/4 of the color information originally present. Don’t we all think DVD is pretty high quality? Even Blu-ray is only keeping 1/2 of the color information, storing video at 4:2:2 chroma compression.

Many professional video cameras use a 4:2:0 ratio to keep the bit rate manageable when recording in camera. When absolute image quality is critical, however, these cameras (The Sony EX1 and EX3, for example) that internally record at 4:2:0 have HD-SDI output from the camera which can output a higher quality 4:2:2 signal, but will require an external recording device (like the nanoFlash).

Recorded Bitrates

Interframe compression algorithms record so many bits-per-second. Using a set bit rate stores the same amount of data for every second of video, regardless of how the frames change over that second. With a set bit rate you know exactly how large a 5 or 10 minute video file will be, since the bit rate is fixed. When we used film to record to (or MiniDV tape), the bit rate had to be constant, because the tape moved at a constant rate. DV footage, once digital, still records to tape at a fixed rate of 25Mbps. HDV, a descendent of DV that uses MPEG-2 compression, records at a fixed rate of 35Mbps.

Most cameras and codecs today, however, record using variable bit rates because it is more efficient. This changes the recording bit rate based on the amount of information change frame-to-frame. If it sees an almost identical previous frame, very little data will be encoded. However, if a large part of the frame is changing, there is much more data, and a higher bit rate will be recorded. The takeaway message, though, is that every recording device, whether in-camera or external, has a maximum bit rate it can handle. The various compression codecs have to provide final data at a bit rate that is acceptable to the recording device or bad things will happen: missed frames, jumping, etc.

HDV XDCAM AVCHD and every other codec

The terminology involved in the various codecs is beyond chaotic. To a video-outsider it’s incomprehensible, but we can try to clarify things a bit. Like most simplifications, what follows is a bit generalized in the interest of keeping it easy to follow. We’ve left out and ignored some arguable points that would easily lead to 4 more pages of clarification in an effort to make it readable. As a general overview, however, this is a pretty reasonable summary. First, we need to separate containers (sometimes called formats) from the underlying codecs. A container is a format that can use (or be used by) many different (but not all) codecs. AVI, Quicktime, RealMedia, DivX, and many other containers exist, but they are (with a few exceptions) not actual codecs.

There are several codecs in common use today, each following a set of standards developed by the Motion Picture Experts Group (MPEG), the ITU-T Video Coding Experts Group (VCEG), or the Joint Video Team (JVT) from both groups. These standards provide a lot of customizable options to the various camcorder and software manufacturers. Some cameras let you choose between two codecs, but most only offer one. The reason behind this? Different codecs require different processing algorithms to encode video. The processor in the camera (yes, cameras have processors very similar to computers) is designed so that it can handle the encoding for that specific codec. And the memory used to store the video is designed to handle the bit rate needed for decent quality video from that codec. Etc…

The most current families of standard codecs from MPEG and VCEG are combined as the H.264/AVC/MPEG-4 standards. H.264/MPEG-4 (also referred to as MP4 at times) allows for a much lower bit rate then previous codecs while still achieving excellent quality. It is used not only for video compression during recording, but also for compression after editing. Youtube, Blu-ray, and the iTunes Store all use H.264 for encoding video. AVCHD codecs are H.264 based codecs used in newer high end Sony and Panasonic cameras, but many other newer camcorders use H.264 based codecs.

Several other codecs remain in common use. Motion JPEG is used on many point-and-shoot video cameras and Nikon Video SLRs. It doesn’t compress nearly as much as H.264 codecs, but requires a lot less processing power and is particularly suitable for nonHD video and lower resolutions. The HDV and DVCAM family of codecs use largely MPEG-2 compression as does the XDCAM codec. These files aren’t usually as tightly compressed as H.264 codecs, although MPEG-2 Long-GOP comes close. These codecs are often found on high-end digital video cameras. The reason why they don’t use H.264? Depending on the source of information you read, this is because the files are easier to edit, or because the manufacturer had lots of chips made for these codecs and was going to use them. I suspect both reasons are true.

So basically each manufacturer chooses which of the standard codecs to implement in their camcorder. Well and good. However, they then modify it a bit, build the chip they’ll install in the camera to use their version and identify it with a cryptic set of initials in an apparent attempt to prevent anyone from understanding that their codec has anything in common with anyone else’s codec. Let’s look at one example. Sony and Panasonic jointly developed AVCHD (Advanced Video Codec High Definition) for their consumer camcorders, which is also used by Canon. AVCHD is MPEG-4 AVC/H.264 compliant so it can also get tagged with those initials. Panasonic tweaks AVCHD with some higher bitrates and markets this codec as AVCCAM in their professional cameras, or downgrades it to 720p recording only and calls that version AVCHD Lite. Sony calls their version NXCAM in their newest professional cameras (as opposed to the XDCAM, a different codec used in many of their current high end cameras). Canon and Panasonic use a High-Profile level 4.1 modification of the AVCHD codec in some cameras which allows a maximum bit rate capture of 24 Mbits/sec, while most camcorders using AVCHD capture a maximum bit rate of 17Mbits/sec. On the editing side, Adobe Premier required a third party plug-in to convert certain versions of AVCHD, but does fine with others, Final Cut Pro converted this format to Apple Intermediate Codec before editing was possible, and Vegas had no problems with the format at all.

Pretty confusing, huh? The takeaway message, with a lot of caveats, is that most codecs in higher level cameras are MPEG-4/H.264 compliant and fairly similar as to how effectively they compress video while maintaining quality. They may differ in offering 1080p (some don’t), in how high of a bit rate they can record (which, given similar codecs, is a fair estimate of image quality), how often they record a key frame (which may be user adjustable in-camera), and how easily your editing program can convert it into an editable format. There are a few common codecs that you’ll run across regularly that fall into several groups:

• DV/DVC/DVCPRO/DVCAM – largely legacy technology, but many HD/HDV systems are backwardly compatible with DV/DVC, and it is used in some high-end video and video broadcast cameras.
• HD/HDV – Used by Sony, Canon, JVC, and Sharp, originally designed for recording to tape. It uses MPEG-2 compression, 4:2:0 chroma sub-sampling, and writes with a constant bit rate. Used in many tape-based camcorders, but also some digital recorders.
• XDCAM – Designed by Sony, but also used by JVC, originally designed for recording to disc. (In some ways a container {see above} rather than just a codec as most cameras using XDCAM can also record in DVCAM or MPEG-2 variants.) Uses an MPEG-2 or MPEG-2 Long-GOP codec, 4:2:0 chroma subsampling, and writes in a variable bit rate to 35 Mbits/sec. However, the XDCAM HD422 version uses a 4:2:2 chroma subsampling profile and writes a maximum 50 Mbps rate.
• AVCHD – Sony (NXCAM), Panasonic (AVCCAM). Uses MPEG-4/H.264 compression, 4:2:0 chroma sub-sampling, and writes in a variable bit rate to 24 Mbps. Note: Some video editors need a third party plug-in to upsample certain AVCHD files to a usable format.
• Motion JPG – intraframe only compression, usually used in point-and-shoot video cameras, but also the Nikon D90 and Pentax K7. It is less efficient than other codecs, so usually image size or frame rate are limited.

Or if you’d rather see what some common camcorders and videoSLR cameras use:

Camera	Algorithm	Maximum Bitrate
Panasonic HVX200	DVCPRO	100Mbps
Sony EX1, EX3, JVC HM100	XDCAM	35Mbps
Sony Z7U	HDV	35Mbps
Canon HV30, HV40	HDV	35Mbps
HG21	AVCHD	24Mbps
Canon 5D MkII, 7D	H.264	40Mbps
Nikon D90/D300s/D3s	motionJPG	bitrate unknown
Panasonic GH-1	AVCHD	40 Mbps

Compare those bit rates to what an external recorder like the nanoFlash can record: 230Mbps.

Conclusion

What does all of this mean? In general, you want the highest bit rate, using the most efficient compression algorithm possible. MPEG-4/H.264 codecs probably produce the best quality/compression ratio. However, top-end professional editing may require a less lossy format, such as an MPEG-2 based codec with resulting larger file sizes to get the absolute best image quality. Some high end cameras will allow you to take the video feed directly out to an external device and record it at an even higher bit rate with less compression for critical footage. Hence an external recorder like the Nanoflash, provides higher bit rates (180Mbits/sec) and less compression than is possible in-camera. (A note of sensibility: the 230Mbps bit rate of the nanoFlash is excessive for use with your $300 handycam or even the Canon HV40. Your image isn’t going to improve beyond the quality of the camera.)

What you intend to do with the footage after recording is also important. Some of the higher compression codecs can be difficult to work with in a non-linear editor and require upcoding to an intermediate format (read: lots of processor power and hard drive space) for editing. Some of the simplest formats, like Motion JPEG, can be drag-and-drop edited in even the simplest programs. And less compressed, but larger files (or even uncompressed files in certain high-end devices) can be a dream to edit and provide the absolute best quality after processing

ProVideo Coalition.com: Camera Log by Adam Wilt | Founder | Pro Cameras, HDV Camera, HD Camera, Sony, Panasonic, JVC, RED, Video Camera Reviews

ProVideo Coalition.com: Camera Log by Adam Wilt | Founder | Pro Cameras, HDV Camera, HD Camera, Sony, Panasonic, JVC, RED, Video Camera Reviews