WTF is the BFD about the RED?
Over the last few years, there’s been a lot of talk about the RED camera. But what is it, and why is it so buzz-worthy?
For almost ten years now, digital technology has been heralded as making art possible on a low budget. Not only filmmaking, but also music, animation, and even photography and still art. The term is the “digital revolution”, and it’s the technology that makes it possible to make a feature film without having millions of dollars.
But the digital technology has had its limitations.
Standard feature films are shot on 35mm film stock. In the digital age, film is usually scanned into a computer for editing and visual effects work, but the film itself is an analog medium. Its resolution is technically infinite, as you can magnify it more and more and always discover more detail; the limitation becomes the size of the “grain” on the image. Film is usually scanned at 2K resolution, which is just slightly larger than the largest standard high definition resolution of 1920×1080, although visual-effects-heavy films will often scan at 4K, and in rare instances even 8K. But the finishing and final output that is “printed” back to film is usually 2K.
Film has a wide “dynamic range”, or “latitude”, meaning that it can record detail in both the dark and the light areas of the image across a wide range of brightness. I’ve heard between 14-17 stops of dynamic range can be captured on film; I don’t know for sure, but it’s high.
Video, on the other hand, even newer digital video, is not quite so hot. Until the recent creation of HD camcorders, most camcorders have been standard-definition — 720×480 in NTSC countries, like the USA, and 720×576 in PAL countries, like much of Europe. For the non-numerically inclined, if the resolution of a typical film scan is the size of a postcard, standard definition video is the size of the stamp.
High definition recording has improved the issue of visual resolution, more or less, but many of the problems inherent to video persist.
Whatever the resolution of the image, the dynamic range is very limited in video. To film’s ~15 stops of latitude, video offers about 3. That means that it cannot record a wide range of brightness across an image and you must make a choice as to what is more important.
If you want to keep detail in the brighter areas of an image, you need to “stop down” the exposure so that the details are not lost. But that means that the less-bright areas of the image will become extremely dark, and details in the dark parts may all simply be lost and become black. Digital cameras can also be “noisy”, meaning they will have random color/brightness fluctuations at a per-pixel level, that is more visible in dark areas and becomes extremely visible if those areas are artificially brightened.
If you expose to keep the shadow detail, then the bright parts of the image will “blow out” and become a mass of pure, flat white. So what you often must do when shooting video is light everything very bright, so even the shadows aren’t too dark, expose for the bright areas of frame, and then enhance the tonal range of the image in editing.
Consumer-level video also has reduced color sampling. I could get caught up writing LOOOOONG posts on different color schemes that could make your head spin, but suffice it to say that in video, every pixel has its own unique brightness value, but may not have its own color value recorded, and instead will be interpolated or averaged together from surrounding pixels.1 Film does not have this reduced color sampling; every pixel gets its own brightness AND color value (in technical terms this is referred to as 4:4:4 color).
Another, larger disconnect between a digital camera and a film camera is the size of the imaging plane. As mentioned above, most high-budget “Hollywood” films these days shoot on 35mm film, which is 35mm measured diagonally across each frame.
Instead of a film frame, a digital camera has a digital sensor. Your standard home video camcorder — even an HD one — has a sensor 1/3″ in size. Converting between measurement standards, that’s about 8.4mm. Even the high-end super-professional Sony F-950 camera, on which they shot the last two Star Wars films, only has a 2/3″ (16.8mm) sensor, and only records HD resolution.
If you’ve ever taking a photography course, you may be familiar with the pinhole camera. Simply (and probably somewhat incorrectly) put, the smaller the area that light has to pass through and strike, the fewer light rays you have to deal with, and the easier it is to focus the image sharply.
So you know how film has that cool look where some parts of the image are in focus and the background will be out of focus? And how video seems to have everything in focus pretty much all the time? That’s why: video is recording onto a smaller imaging surface than film, getting less of the scene in frame and having a deeper depth of field (“field” referring to “the space in front of the camera within which an object will appear to be in focus”).
So video loses to film — and loses hard — in terms of resolution, dynamic range, color sampling, and imaging plane.
The RED camera records a 4K image (four times the size of the largest HD standard resolution2) with 11.7 measured stops of dynamic range (and they’re working with the camera/sensor firmware to improve it) with pixel-to-pixel color accuracy, with an imaging plane the same size as 35mm film (technically Super35). It uses standard 35mm lenses like you’d find at any motion picture rental supplier, and it’s a fully-digital system.
Now, the RED isn’t QUITE the only system out there with these capabilities. Panavision created the Genesis, on which such films as Superman Returns, Apocalypto, Superbad, and The Other Boleyn Girl were shot. The Genesis also has a Super35 sensor and can also use standard cinema lenses, and records a 1920×1080 HD image.
Arri, another film camera supplier, came out with the D20, a 2K camera with a 35mm sensor and lens use, and ~10 stops of latitude.
The Dalsa Origin is also a 4K camera, also with a 35mm sensor and 35mm standard lens use, with supposedly more than 12 stops of latitude.
So given that it’s not the only fish in the sea, what’s the big fuckin’ deal when it comes to RED?
In a word, accessibility.
The three cameras listed above are rental-only cameras, all renting for around $3000/day, not including lenses. The Dalsa’s “base package” is $5000. They are relatively inaccessible to most low-budget filmmakers, who can buy a Panasonic HVX-200 for the same price as a two-day rental of the Genesis.
RED, on the other hand, is available for purchase direct from the manufacturer. The base rate, the price for the body of a RED camera, is $17,500.
That may still sound like a lot, but it’s less than a week’s rental of the Dalsa camera.
The comparison is somewhat unfair, given that that price is JUST for the RED body — no accessories, no batteries, nothing — but it still remains that you can get a fully decked-out RED system for $50,000 out the door, which is less than you’d spend renting any of the other systems for just one month. The RED is modular and upgradeable, meaning that the investment in the purchase does not depreciate as quickly, and is certainly much more worthwhile for the serious filmmaker than renting cameras for every project.
It’s also a much cheaper rental, as a result. If you decide that buying isn’t for you, current market rate for a RED system — body, batteries, accessories, recording media, and usually even a heavy-duty tripod — is just under $1000/day. That’s a third of the cost of any of the competitors.
Speaking of recording media, the other cameras have their own proprietary recording devices — either massive RAIDs to which the camera must remain tethered, or expensive portable units that you can rent extra.
And while the RED does offer a large-capacity RED Drive, it can be attached to the camera for portable operation. More than that, the camera is capable of recording to high-transfer CompactFlash cards. The same kind you can buy at a photography store. A non-proprietary format.3
So whereas a roll of film would normally be several hundred dollars, plus processing, and only useful once, you can buy a CF card, or even the drive, and use it again and again with a pure digital stream.
Another thing to consider: 4K resolution data is a lot of data. It’s difficult to deal with a 4K stream (something like a data rate of 400 Mbps) without having a really high-end, custom configured system. But RED uses a very clever wavelet technology, REDCODE, to fit all that data into a stream of 27 MB/s. That’s pretty much the same as HD, which means most off-the-shelf computers today, particularly Apple, can work with it natively.
I reserved a RED about 18 months ago and was just recently informed that my number is up for purchase. It may take me a while to get it, but I’m working on figuring out some loan options here, and once I do get it I may be posting/gushing about it a great deal more, so it was important to talk about why this camera is SO AWESOME.
- This is why, if you’ve ever tried to pull a greenscreen key off of DV, it looks like stairsteppy dogshit. The edges are not strongly-defined in the color channel, but are instead a gradient between the green and whatever color your foreground is. The keyer is actually doing a good job of separating the image via color, it’s just that your color information isn’t good. The edges look good because the brightness information is all there, and the way the eye processes images fools it into reading it as a sharp-edged color image. On DV, keying on the brightness, aka luminance, info is the way to go. If you care to learn more about the what and wherefores of color sampling, you can read up here.↩
- The RED utilizes a Bayer pattern sensor, which requires complex algorithms to interpret the available color data. Some have said that this limits the actual resolution of the camera, and indeed shooting some resolution charts has shown that the camera can’t QUITE resolve a 4K level of detail. But it’s close-enough-for-government-work to call it a 4K camera.↩
- Red does sell their own branded CF cards and doesn’t “guarantee” cards that you buy yourself, but as long as your cards match the proper specs, you can technically buy generic media for a high-end system. And that’s a big deal. ↩