[ This article was first published in the December, 2004, issue of
Larry’s Monthly Final Cut Studio Newsletter. Click here to subscribe.
Updated 1/23/05: Hoo, boy! I should have expected this, but this
article touched a nerve. Lots of comments from a variety of folks are
reflected in this article. Updated 4/15/05 with correct name of makeup. ]
This article grew out of an idea suggested by King Dexter, a student and long-time reader. King writes:
Many, many FCP users are independent small-fries with hand-to-mouth budgets and limited networks for resources and exposure. Many very creative artists must necessarily work with the lower end formats such as MiniDV. Most recognize there are pros & con’s and understand that there are pitfalls to beware of when working in say, mini format. I’ve seen some great disappointments surface with regard to expectations that don’t come to fruition due to ignorance of format-specific limitations. Is there a way that you might list obstacles, pitfalls, limitations, etc. that one might want to be aware of when beginning a project using less than high-end DV? It’s not that this kind of information is not available, but man do you have to search high and low and in all the crevasses to find it. How nice if there could be a Larry Jordan 1-10 (or 20, or 30) check list and comparison to consider when ramping up to create in low-end DV. Hopefully this kind of information would inform and encourage rather than discourage those who, for whatever reasons, must choose to work with less than multi-gazillion dollar cameras.
Larry replies: While this list, comparing the benefits of “professional” standard definition video formats (such as Digibeta or Betacam SP) with DV could be quite long, for me, it boils down into four main trade-offs:
- Clarity or apparent focus
- Highlight and shadow detail
- White levels
DV uses a very compressed and restricted system for displaying colors, which is called 4:1:1. (Click here to read an article explaining this in more detail.) In summary, what this means is that in DV the color for 4 pixels is described using a single number. A computer creates fully uncompressed (i.e. 4:4:4) color. However, as Graeme Nattress points out, “There is no fully uncompressed 4:4:4 SD tape format. All broadcast tape formats, DVCpro50, Digital Betacam are 4:2:2. This means that DV contains half the color resolution of SD.”
Philip Hodgetts adds:
This is somewhat misleading or incomplete.
To clarify: the numbers 4:4:4 are the ratio of [the] sample of the YUV color channels within the datastream. Graeme is correct that there is no tape format that records 4 U and 4 V samples for every Y sample recorded.
What I think you meant to convey in that last sentence is that “…DV contains half the color resolution of professional standard definition formats”. DV has only 1/4 the color resolution of the luminance resolution of any SD format. In 4:2:2 Y = 720 x 480 (or 486); U & V are 360 x 240 each; in 4:1:1 Y is still 720 x 480 but U & V are only 180 x 120 each but offset so that they “fill in” a bit for each other.
Not a big thing, and the general principle is right, it’s just me being picky. 🙂
[Larry again.] For this reason, if your program involves a lot of chroma-key compositing, meaning blue or green-screen work, or lots of color correction or color tinting, there just isn’t the color information in your signal that exists at higher quality levels, such as Betacam SP, DigiBeta or DVCPRO-50. All these professional formats work in a color space where, in general, colors are averaged across 2 pixels, rather than across 4. While still not ideal, these formats provide four times the color information when compared to a DV signal.
DV cameras, in order to reduce costs, use smaller chips to convert the image from the lens into a digital signal, than a professional camera. In a DV environment, a Sony PD-150 camera, which costs around $3,000, for instance, uses three 1/3 inch chip to convert light to digital signals; some DV cameras only have a single imaging chip. At the high end, a Sony DVW709 Digibeta camera, which costs around $45,000, uses 2/3 inch chips. The larger the chip, the better able the camera is to resolve the image.
Second, the optics the lens uses are far less able to sharply define the image. Good lenses are really, REALLY expensive — for good reason, they make your image look great. A DV camera can show that your talent has hair. A good Betacam camera can show the flow and texture of the hair. A good HD camera will show individual strands of hair. Good lenses, which can cost tens of thousands of dollars, make this possible.
Finally, the auto-focus in most DV cameras is laughable. Far too often, when I see a DV image, I see that the auto-focus avoided focusing on the talent, but, instead, focused on the background. Canon cameras with automatic lenses are notorious for this, which is one reason I am not a fan of the XL/1. Where ever possible, turn off the auto focus and auto iris when you shoot. This avoids having the camera continuously “search” for focus, or “pump” whenever the camera pans or someone walks into the shot.
DV cameras don’t have the ability to display a lot of detail in dark areas or light area. Dark areas, especially scenes supposed to be at night, tend to get muddy, with lots of subtle details lost. In very bright areas, say, curtains covering a window, the detail of the curtains gets lost because the light from the window is so bright.
This is not necessarily the fault of the DV spec, but more the fault of the imaging chip, poor lenses, and general lack of quality components, which is required to make a camera that sells for 2-4 thousand dollars.
This lack of detail in highlights and shadows, is often what gives DV it’s traditional “video” look.
Ed Scott adds:
I look at the paragraph above in a slightly different way. First, you set camera exposure to show whatever highlight detail is needed. Second, you add fill light (bulbs or reflectors) to a level that lets you see the amount of detail you want in the shadows. This is easier to do this viewing a monitor. The fact that a professional camera has wider latitude or dynamic range does not prevent this technique from working with a less expensive camera, though as you say, the image cannot look quite as good.
It is admittedly a small point, but it would often be possible to significantly improve the images recorded on DV cameras by adopting the exposure and lighting techniques used by professionals.
Recording video on a DV camera uses a digital standard for white level. This allows the camera to record whites at 109% when measured on the FCP waveform monitor (109 IRE to you engineer types). SD cameras, however, only record whites at 100%. For this reason, you need to decrease your white levels when integrating DV with SD footage. If your ultimate output is DV or DVD, this white level adjustment is not a big deal. Compressing your video for DVD will automatically clamp the whites at 100%.
However, if you are recording to any form of Beta for ultimate broadcast, this white level is a really big deal. Consequently, you need to pay attention to your white levels throughout the editing (but not production) process and adjust them after the fact.
Kit Laughlin wrote to point out that audio, too, is important.
IMO, the most important consideration for an indie filmmaker choosing a DV camera is not the picture quality, important though that is. The emotional content of a program is carried on the sound track — and you know how intolerant of poor sound all viewers are….
So, you might care to add a comment about this in the next newsletter: which DV cameras can actually record really good sound? This has three corollaries: how do you get good sound into the camera, how do you hear that on location — where you might have some capacity to re-record a poor take? There is no such thing as “We’ll fix it in post” in this reality if the original sound is thin, has a poor signal to noise ratio, or intrusive b/g sound. Last, how do you edit the recorded material and do justice to it? Computer speakers cannot do this job!
Jay Rose in his excellent book (Producing Great Sound for Digital Video) tested a number of cameras, and the Panasonic GY-500 was the best available then (2003). His recommendation was one of the reasons I bought mine — the sound is nearly as good as a Tascam DA-P1 (the DAT recorder). I used to use the DAT to record double system sound when I was using the Sony PD-150.
Another advantage is that the Panasonic is that it takes 1/2″ mount broadcast lenses, and quite a variety of them, from Fujinon and Canon. The optical difference between these lenses and prosumer ones has to be seen to be appreciated. The focussing is precise and the lenses are fast too (the Canon I use is ƒ1.7 wide open, and is sharp wide open). With 1/2″ CCDs (another reason to avoid prosumer cameras, which use 1/3″ ones, mostly, is that you absolutely need this speed to be able to control depth of field (DOF), THE creative control tool for camera folks. I use two ND filters to assist me in this too. (For newbies, the Depth of Field is a function of the ƒ stop being used and the focal length of the lens being used. The smaller the sensor, the greater the DOF for any focal length lens equivalent. By this I mean that if you know 35mm photography, and you are trying to reproduce the look of, say, an 85mm lens (typical portrait), the smaller sensor of the video camera will only need a much shorter (wider) lens to reproduce the same proportions. “Wide” (short focal length) means more depth of field.
So, to be able to avoid almost everything being in focus—the typical “video” look on the early prosumer cameras, you need to be able to separate b/g from foreground—and to do this you need a fast lens. To cut the light down so you CAN open up the lens (the lower ƒ number for any focal length, the correspondingly shallower DOF you see), you will need a range of Neutral Density, or ND, filters. The numbers refer to the number of stops of light cut by the filter: an ND 2 will cut the light down to a quarter (so, from ƒ11, say, to ƒ5.6, two stops. For real b/g separation, you need to be shooting at around ƒ2 or ƒ2.4.)
[In reference to] the corollaries: the indie film maker MUST use a high quality shotgun microphone, and it MUST be on the end of a boom swung by someone who: 1) can point it and keep it just out of shot, and 2) is wearing high quality headphones connected to either the camera or, better, the mixer he/she is wearing. I use a 2 channel Wendt X2 for this purpose, and it is an awesome performer. I recommend BeyerDynamic DT-250s as a rugged, on-location can (these are a sealed headphone, so no spill). Any of the Sennheiser or similar microphones recommended for ENG or film work will do a decent job, assuming they are shock-mounted to the boom. The combo of the mic, mixer and good recording capacity of the camera will blow the audience away compared to much of the sound you hear on indie films.
And last but certainly not least, what are you going to actually listen to all this good sound on in your editing suite? A proper set-up is essential; the interested reader can search LAFCPUG for “editing suite” as a start, but I recommend near-field speakers including sub-woofer, in a well-baffled room (near-field set-ups don’s need quite as good acoustics as standard monitoring set-ups, as signal to background sound is higher as you are much closer to the speakers). Bigger rooms are way better than small ones for the same reason. I use and recommend KRKs from Huntington Beach; excellent and relatively inexpensive. These are powered speakers, so you need a small mixer (mixer runs out of the analogue outputs from the deck or camera at line level; speakers are plugged into one of the line pairs as one of the sources and the speakers are driven out of the “Main” outputs. Volume is adjusted on the “Main” pot.
Another reason professional video looks so good is that time is taken for good lighting. It is impossible to overemphasize the benefits of good lighting in a production. Even cheap indie films can be improved by using softer lights and bounce cards. But, here are two other tricks that I swear by that can immeasurably improve the look of your productions:
- Use filters to warm and soften interviews. Experiment with a 1/4 warm black Promist filter from Tiffen or Schneider. You’ll be amazed at how it changes the look of your interviews for the better.
- Use matte make-up on your interviewees. I am a HUGE fan of T. Zone matte makeup, by Lancome. This $25 tube of magic makes everyone look great. Best of all, as it doesn’t add color, guys are willing to use it who would never be caught dead in makeup.
DV is a great format and many, many fine productions are made using it. This is just a list of what it doesn’t do well. If your projects don’t require precise color, DV will work great. If they do, you may want to look at a different format – a good one that FCP handles well is DVCPRO-50. Cameras and decks are less expensive than Betacam and image quality is very, very nice.