Much has been made of achieving a “film look” with digital motion picture camera
technologies. While discussion has centered on providing a 24 frame per second
progressive image at sufficient resolution, there are other differences between what is
thought of as video and what viewers expect from film. Among these differences is the
limited dynamic range (contrast ratio or latitude) of most video recording. Eastman
Kodak has published experimental results showing high definition video delivering a
dynamic highlight range relative to an 18% gray of no more than 3 to 4 stops; compared
to the 15.9 stops of some color negative stocks2. Video camera manufacturers have made progress improving this dynamic range, including the introduction of knee and black
stretch (black gamma). Panasonic’s cine gamma and other methods have further
increased the effective dynamic range of CCD image capture, by changing the entire
characteristic response of the CCD imager.
Historic Differences Between Film & Video:
A colloquial statement made comparing film to video is that video can only display a
32:1 contrast ratio (about 5 stops) while film can achieve anywhere from 400:1 (almost 9
stops) to 50,000:1 (almost 16 stops). Part of the reason for this is that video standards
had been developed (and in some cases mandated) assuming they would be shown on a
CRT monitor in a lighted room, rather than projected in a darkened theater. Broadcast
equipment tends to sacrifice contrast ratio for brightness in the display, so even when
image capture technologies might have allowed for more dynamic range there was little
interest in doing so.
The SMPTE standards specify that a cinema projector should be capable of displaying a
contrast ratio of from 400:1 to 600:1 (9 stops more or less). It is highly unlikely that
any theater display technologies will significantly exceed this, as human eyeballs can
only perceive 10,000:1 (a little over 13 stops) at a given instant in time, and without a
few stops for the ambient theater light, people won’t be able to see their popcorn or find
their seats. Film stocks have been created where the negative can capture a 50,000:1
contrast ratio, and this allows for a great deal of leeway for setting exposure since only a
fraction of the latitude is needed to create a good release print. This makes film a
forgiving medium, but it does not change the capabilities of projectors or of the human
Film to video transfer (tele-cine) has been around about as long video-tape itself. Part of
the artistic process for modern tele-cine has been choosing what ranges of tonality in the film original to carry through to video, with its more limited dynamic range. Just as “pan and scan” techniques must be applied to get a 1.85:1 aspect ratio image on to a 4:3 screen, choices must be made when a 600:1 release print or a 50,000:1 negative is
transferred to be seen on a 32:14 TV monitor. Choices are made regarding when to bring
out more shadow detail, highlight detail or mid-range color saturation, depending on the
needs of the story. The deliberate adjustment of tonal range in the color correction
process is an important part of achieving a high-end film look.
In recent years film out (the process of transferring video or other digital recording to a
film print) has come into play, and in this case the limited dynamic range of traditional
video acquisition can be considered detrimental. A video limited contrast ratio can
certainly dissuade filmmakers from originating on HD or DV for a theatrical release.
Meanwhile new electronic display technologies with improving contrast ratios continue to be introduced. Clearly, there has been a strong incentive for increasing the effective dynamic ranges of new cameras.
Recent advances in digital video:
With digital signal processing it is easier to manipulate the cameras response over the full
range of the CCD. It turns out that a significant proportion of the full dynamic range of a
CCD imager is dedicated to highlight handling, in fact the highlight handling in most
CCD cameras represents a greater contrast ratio than the active range.
This overhead region of the CCD response is normally clipped white.
Film negative handles a wide tonal range in part because it responds more linearly to
tones in its mid range and more gradually in the highlights and shadows (the lazy S
curve). In other words, there are fewer tonal gradations in highlights and shadows but
only very extreme bright and dark areas are clipped white or crushed black. In this way,
film negative may capture 5 or 6 stops in the linear mid range and then deal with perhaps
2 to 3 additional stops each in both the highlights and shadows.
Alternately video systems provide extra color saturation within its 5 or 6 stop range for a
bright TV image. Video cameras have, more recently, developed technologies to get some detail out of highlights and shadows. Naturally when viewed on CRT monitors, these methods must use up portions of the full dynamic range of the display.
Black stretch: takes a little bit of the lower mid range to use in the shadows so blacks are not entirely crushed. This shadow detail comes at a cost however, and many blacks end up looking dark gray as black stretch effects are increased. Knee: can suppress objectionable hot spots. When using knee one may choose a point along the transfer characteristic below the full dynamic range (e.g., 85%) and set the response above this point to a more gradual tonal response farther into the CCD’s full dynamic range. When the knee is active, white on a chip chart no longer is 100% white but detail can now be seen between the white chip on the chart and the ultimate clipped white. When a knee is applied to the characteristic response, levels below the knee point are still exposed in the same way.
Aggressive knee adjustments attempt to mimic the film “shoulder” (the highlight portion
of the film’s negative’s transfer characteristic). By lowering knee point and tweaking the knee slope the effective dynamic range can be increased. Knee adjustments are possible in most manufacturers latest professional production cameras. And while digital cinematic productions have used the above methods to achieve latitude acceptable for some large theatrical distributions, knee adjustments have the following limitations when used to significantly increase dynamic range:
• Any part of the scene that is illuminated at the knee point may exhibit artifacts
1- he knee area is not gamma corrected like the mid range (it is linear)
2-The only way to get a contrast ratio approaching a typical film stock would be to
set the knee point extremely low;
3- This leaves an extremely small mid range
4-This creates a relatively large knee area that is not gamma corrected
5-This removes some of the levels of tonality from the midrange, and this can cause banding
6- Knee adjustments can be difficult to make, because potential artifacts introduced by extreme knee settings are unpredictable and difficult to detect on a field monitor
In summary using knee adjustments is very good for enhanced highlight handling for TV
applications, but must be approached with caution when attempting to significantly
increase the dynamic range of the camera beyond that of traditional video monitoring.
When the final product is to be shown on a CRT display, the highlight handling and
increased latitude made possible through knee adjustments should be more than adequate.
Cine Gamma, The Next Step:
Panasonic’s Cinema Series cameras all use knee and black stretch when shooting for
NTSC, ITU 601 DV or ITU 709 HD standards. Normally these are destined for CRT,
DLP or LCD viewing.
In addition the Cinema Series cameras have one or more cine gamma modes, which takes the above techniques a step further and a step closer to the behavior of film.
The FILM REC mode is radically different from VariCam’s other shooting modes
and although many of the basic principles apply to the other cine gamma features, the
changes are more subtle and require little change to shooting style when compared to
standard video modes.
VariCam’s FILM REC mode:
FILM REC provides an alternate approach to using the knee functions for highlights
(Black Stretch still works the same way), and addresses the problems encountered as the practical limits of knee adjustments are reached. Because FILM REC pushes the entire
characteristic response below where it could be pushed using knee adjustments, there is
no knee in FILM REC.
In the mid range of a video transfer characteristic, a power law is applied to the response, giving the line a slight curve, and this cancels the non linearity of the response of a CRT monitor (a non linearity that is shared by most display technologies). This is known as gamma correction and is adjusted using the master gamma setting. Gamma correction has the added benefit of applying a decreasing value to tonal changes as the extreme highlights are approached. Gamma correction is not applied to the knee area when kneeis in use.
The cine gamma transfer characteristic applies a series of power law curves joined to
form a seamless more gradual function that applies gamma correction throughout the
transfer characteristic. The CINE GAMMA transfer characteristic is capable of
increasing the contrast ratio to better than 1,000:1 (10 or 11stops). As with the knee, this means that white is not seen as 100IRE on a video display because “white” will contain some details. The dynamic range of FILM REC is comparable to the contrast ratio of some film prints, and is enough to utilize the full contrast range of a theatrical projection.
User adjustment in the FILM REC mode is simplified as well. The somewhat confusing
knee adjustments are replaced by one dynamic range adjustment known as DYNAMIC
LEVEL, which is set from 200% to 500% of standard video dynamic range. The transfer
characteristic of the VariCam shooting in the FILM REC mode is designed to simulate a
film positive in the middle “linear” range.
Reducing Dynamic Level increases the image contrast similarly to increasing the master
gamma. If the full 500% is not needed for the scene content, more tonal levels can be
used by the image if the dynamic level is decreased. This will also yield a more saturated
Viewing FILM REC:
Because we have not confined our contrast range to something a CRT monitor can
display we cannot necessarily set exposure using a field monitor as the only reference.
For a DP accustomed to shooting film, this lack of instant feedback should not pose a
problem but still bears some explanation.
If we are capturing more than 10 stops of latitude (in the camera imager as well as on
tape) we cannot possibly view all of this in a monitor that displays only 5 or 6 stops. For
better or for worse the monitor is biased towards the top of the tonal range, not the center.
The maximum video signal delivers the maximum brightness level of the monitor, so a
properly set monitor is displaying the top 5 or so stops of the full dynamic range of the
camera. Panasonic recommends that a scene shot in FILM REC not be exposed much
differently than if it were shot in VIDEO REC (perhaps only 1/3 of a stop) but you will
observe when switching between the two modes that FILM REC looks underexposed in
the monitor by comparison (perhaps by as many as 2 _ stops). In other words, the image
will look dark, dull and the darker areas may not be visible on the monitor when viewing
FILM REC. The diagram below Illustrates how the full dynamic range of the monitor is
biased to the top portion of the video signal, and it also shows how everything being
recorded can be seen on the monitor in VIDEO REC and how much of the shadow detail
cannot be seen.
If you increase the monitor brightness (please don’t leave it out of adjustment!), you may
find additional detail in the shadows. These details in the shadows should reproduce on a
film-out and could likewise be brought out in color correction at the expense of highlights
or overall saturation. The monitor is still showing what any other monitor will show if
raw footage is played back, so if you will not be going to film-out or color correction,
you may want to expose for that image (i.e., overexpose), even though you will not be
utilizing the full range of the camera. Anything clipped white on the monitor will be
clipped on any reproduction, but areas that look too dark can be recovered in post.
Practically speaking, this means that the convention of setting normal (18% gray)
exposure to 60IRE or 70IRE must be abandoned since it will put most of your subjects in
the highlight area.
So what can be done with material recorded using FILM REC? The original intention of
the feature was that data might be handled like a film negative (or more precisely,
positive). Either it is color corrected to be sent to a film out, or it can be color corrected
tape to tape and optimized to be shown on another electronic medium, (just as is done in
tele-cine). By delivering superior dynamic range at image capture FILM REC delivers
the one-source-many-uses possibility previously unique to film. The following
workflows have been used by VariCam productions using the FILM REC mode.
1. The DP sets up each shot adjusting the dynamic range, black stretch and other
camera parameters until the light levels (determined by the cameras internal spot
meter “Y-GET” function) register where the DP’s experience tells him they
should be. From here a special look up table for the Arri Laser is used, such that
the scenes are accurately reproduced during a film out without any color
correction. This would be a workflow similar to using color reversal, and has
been successfully utilized in Japan. A look up table (LUT) created for this
purpose can be obtained from Panasonic. Using a custom LUT, a film out was
created that very closely approximated the density curves of 35mm print, derived
from 35mm negative
2. Another operational mode, favored by some DPs in the USA (many of whom are
unfamiliar with HD or video) is to simply treat FILM REC as a 640 ASA film
stock (depending on how parameters are set 400 ASA or 800 ASA might be more
appropriate), and light using a light-meter, as if film were being exposed. By
shooting tests beforehand the DP can learn the highlight and shadow behavior
(these tests would be necessary when approaching any new film stock). Color
correction using a high-end, system proceeds as if it were a digital intermediary,
having skipped the 2K scanning process. For any video deliverable, color
correction would be approached similarly to tele-cine for a DVD or broadcast
3. Another approach would be to keep a waveform monitor on set, and light as if
shooting for broadcast, getting maximum information within the full camera
imager dynamic range and color correcting in post to match lighting scene to
scene. The difference for FILM REC would be that normal exposure would be
kept closer to the middle of the range.
4. Lastly, it is still possible to simply ignore the fact that the camera is recording a
wider dynamic range than the monitor can show and simply use FILM REC to
achieve a desired look. By exposing to the highlights the user has a way of
shooting that can deliver superior highlight handling without making knee
adjustments. This is similar to how Cine Gamma features are used on the standard
definition cameras with Cine Gamma modes. A helpful item to keep in mind here
is that video is like a film positive (not negative) so when in doubt you should
close the iris a bit rather than opening it as would be wise when exposing film.
More workflows may develop as the technology gets into the hands of the creative
community and complimentary technologies come into the marketplace.
Many VariCam users choose HD digital acquisition over 35mm or 16mm film because of
a limited budget. When working on a shoestring, the post-production budget may be
unknown, or non-existent. Part of the case for digital cinematography, is the ability to
quickly and inexpensively create material that can be shown to potential financial
partners, in hopes of getting further investment for post production. Independent film
crews have been shooting using FILM REC optimized for the field monitor, knowing that
the cost of a film out might be covered only after the project has been projected digitally
at a film festival. This mode of operation was not anticipated by the manufacturer (as is
often the case) but certainly is capable of producing some good images.