What is film grain?

Film grain is further intensified here by bleach bypass. Click the image to enlarge it.

Film grain is often used to describe a few different concepts. For the film savvy viewer, film grain is the random grain-like texture seemingly overlaid on a scene captured on film. It is observed in a paper print, on a display or through projection. In this aspect film grain is somewhat related to film scratches and dirt specks. On a higher level, the grain texture is one element that distinguishes the film image from reality. This is true for stills, but even more so with cinema where the random nature of the grain manifests itself in consecutive frames, and the greater enlargement makes it more pronounced. A more technically inclined person with less sentiment for films would simply call this apparent image graininess noise.

In black and white negatives the light sensitive elements are usually silver halide crystals suspended in gelatin. When photons hit the crystals, they are converted to a latent developable state. Subsequently, lab processing dispenses with the unexposed particles. Film grain is commonly thought of as the remaining silver particles. This isn’t entirely correct. While observable film grain is a result of these image-forming particles, it is distinctly different from the particles themselves. The individual silver particles are so small they can’t be seen. What is perceived as grain is clumps of these particles and, more precisely, micro-variations in areas of relatively uniform negative density. In color film the silver particles are coupled with dyes; silver is removed in processing after development and only dye clouds remain. These dye clouds are the cause of graininess in color film.

Silver halide particles embedded in gelatin (left). Particles removed from gelatin for a better view (right). Note the 2 micrometers reference at the bottom. (images shamelessly lifted from this page)

Graininess is a subjective visual sensation. And it is highly dependent on scene tones, colors and details. All this makes it a bit hard to quantify. Traditionally, film stock density unevenness is quantified through measurements of density fluctuations. This objective quantity is called granularity. Granularity is measured with a microdensitometer in a small area of uniform density at 1.0 density above base. The microdensitometer usually has an aperture of 48 microns (0.048mm). The standard deviation from the average density gives us root-mean-square granularity. Standard deviation is very small, so it is usually multiplied by 1000 to bring it into whole numbers territory. Where there are very small silver particles, many of them are averaged and fluctuation is small. With large particles, there are less of them getting averaged, and fluctuations are larger. Modern film stocks (like Kodak Vision 3 stocks) have granularities below RMS granularity of 5, which is considered finer than extremely fine.

Film grain and film properties

In order to facilitate the capture of different shades, film uses silver halide particles of various sizes. But each particle needs the same number of photons for exposure, no matter what its size. So larger particles are exposed faster, and smaller particles need more light (or more time) to capture enough photons. This variance in particle size is responsible for the great dynamic range of film. In the dark areas of the image only the large particles are exposed, and in brighter areas particles of all sizes get exposed. That’s the reason grain appears coarser in shadows and low mids. Some cinematographers overexpose a bit in order to get the cleanest results, but this is stock specific.

There is a similar connection with film speed (sensitivity). Slower film is cleaner and finer grained due to its very fine individual grains. Fast films need larger particles to capture light faster, and thus exhibit coarser grain. Different developers can affect graininess, especially with black & white film. More notably, developers containing silver solvents lead to a softer grain look.

Film grain is also connected to image sharpness. While the relation is complex, especially in color film, fine grain stocks generally resolve more than coarser grained stocks. But there is more than resolving power to perceived sharpness. Film grain is noise and can mask image detail out. But it can also enhance tonality and fine detail by modulating tonal changes that are too miniscule for the brain to register. For those not easily scared by terminology: in this case film grain acts as stochastic noise and causes stochastic resonance.

Lenna clean (left), and with film grain overlaid (right). Grain brings out some of the more subtle detail and occasionally creates fake detail. (click to enlarge)

Film grain and the film look

Film stock manufacturers have always considered graininess a defect and have strived to decrease granularity. Filmmakers, on the other hand, often consider low to moderate grain an important aesthetic. Both for the pleasing qualities of its texture and for its subtle veil over reality.

But there are a couple of other properties not so obviously related to the film look. They aren’t as much a result of film grain as they are a consequence of film grains.

A digital image is made of rows and columns of dots (pixels). It is a matrix. So a digital sensor always samples uniformly the image delivered by the lens. This leads to aliasing problems with high frequency detail in the scene. Hence the need for anti-aliasing filters in the typical digital camera. These optical low-pass filters can kill very fine detail and also complicate the use of small symmetrical lenses with digital cameras with short flange focal distance (like Sony E-mount cameras). In contrast, the individual grains of film and, subsequently, the clumps that form visible film grain are placed randomly. This prevents any noticeable aliasing. There is an attempt to mimic this in the recent Fuji X-mount digital cameras. They use a pseudo-random color array for their CMOS sensors, and have dumped the anti-aliasing optical filter.

An example of aliasing (above) and the same image anti-aliased (below). Note the faux diagonal lines in the bricks in the top image. (click to enlarge)

The other effect concerns movies specifically and is also a result of the random distribution of grains in film. Because of this randomness, each consecutive frame in a film roll captures a slightly different image of the scene (for static or slowly moving scenes). While any single frame may lack some details, all frames as a whole can capture lots of fine detail. When film is projected at 24 fps, the brain integrates the individual frames’ contributions and sees the cumulative result. This lends an organic feel to projected film images.

Sensor noise

The closest relation to film grain in digital video is sensor noise. Unlike grain (usually looked at positively or ambivalently), sensor noise was widely considered a detriment to image quality. This is because digital sensor noise lacks grain’s inherent randomness of appearance and variation in size. Sensor photosites (pixels) are placed on a matrix and they are ordered and fixed sized. These properties translate to sensor noise. There are various causes of noise in sensors. Shot noise (photon noise), thermal noise, readout and reset noise, quantization noise, voltage variance noise, etc. all merge in a single combined manifestation. In earlier sensors noise would often manifest itself in patterns, and would appear quite objectionable. Newer sensors largely dispense with the fixed noise patterns and demonstrate a much more random noise structure. In CMOS sensors, debayering acts as partial anti-aliasing on noise and softens it. Video compression can further smear noise: this can be blotchy and ugly with heavy compression, but can be a positive when only slightly affecting noise. As a result, some recent digital cameras produce an organic noise structure that shares characteristics with film grain.

In general, images acquired digitally are cleaner than film, especially at base ISO speeds. Noise is mostly apparent in dark areas or when sensor gain is applied to increase sensitivity. In both cases visible noise is the result of lower signal-to-noise ratio.

Film grain in post

Film grain is often added to images in post in an attempt to get some of the characteristics mentioned above. Usually grain is applied to simply mimic the film look, but there are sometimes technical reasons behind the decision. Digitally acquired images can look clinical, being both clean and sharp. And more so with CG images. Overlaying a bit of film grain dirties them and adds some texture. With low tonal resolution images (such as 8-bit compressed video) film grain can act as dither and help cover banding issues. But adding film grain is not reserved for digital cinematography. Cinematographers routinely add grain in DI to movies shot on film, because latest film stocks with their extremely fine granularity can look scarily clean. Added film grain can be either software simulated, or scanned from actual exposed film stock. While the latter is the preferred method for most, there are some good synthetic noise examples around.

You can read the previous parts of the Cinematic Look series here:
Part 1: Aspect Ratio, Sensor Size and Depth of Field
Part 2: Frame Rate and Shutter Speed
Part 3: Dynamic Range

Cinematic Look, Part 4: Film Grain

Scene dynamic range

Dynamic range and dynamic range transfer is one of the often misunderstood concepts in video and film, maybe because it is a bit technical. Dynamic range is the ratio between the smallest and the biggest possible values in some signal. Here we are interested in the case when this signal is light. Scene dynamic range or scene contrast is the ratio between the luminance of the darkest blacks and the brightest whites in a scene. This ratio can get quite large in scenes with both bright sunlight and dark shadows.

Human vision has a curious characteristic. In order to accommodate large scene contrasts we don’t see light physically “correct”. We see exponential luminance increments as linear increments. We perceive the change from 10 cd/m² to 20 cd/m² as similar to the change from 200 to 400 cd/m². This means that the series of gray steps with luminances of 10, 20, 40, 80, 160,…cd/m² is perceived as uniformly changing. And the series of gray steps with luminances 10, 20, 30, 40, 50, 60,…cd/m² has the perceived differences between steps getting smaller. One important consequence of this logarithmic correlation of human vision to light is that the eye discerns small luminance differences in the darks better than in the highlights.

The logarithmic concept of light stops fits well with the workings of our vision and is widely adopted in photography. A surface is said to be one stop higher than another surface when the luminance of the first surface is twice the luminance of the second surface. So if a scene has a contrast ratio of 1000:1 it is said to have dynamic range of around 10 stops (2¹⁰ = 1024).

Film dynamic range

The dynamic range of film and digital sensors is usually smaller than high dynamic range scenes. And color reversal film has much smaller dynamic range than color negative film. For example, Kodak Ektachrome 5285, which is a reversal stock, has less than 9 stops of dynamic range. The captured dynamic range distribution varies a bit depending on the specific film negative stock but latest negative stocks like Kodak Vision3 5219 have dynamic range of over 14 stops. Color reversal film is usually much more saturated than negative film. Both high saturation and limited dynamic range make reversal film more of a specialty stock, appropriate for specific uses like ads or music videos. Movies are almost universally shot on negative film.

From the characteristic curve of film (Kodak 5219 in this example) we can note the following. There is a large linear part in the middle of the curve where equal exposure change results in equal density change. That’s where detail is captured uniformly and with the greatest tonal resolution. The slope of the curve in its straight part is called gamma. For most film negatives gamma is around 0.6. This means that one stop of light, or 0.3 log exposure, is represented by 0.3*0.6 = 0.18 density. So, in a way, film does dynamic range compression: as you can see from the chart, a spread of more than 14 stops (4.2 log exposure) is captured in a density range of less than 2.0 log D. Most of this is due to highlights and shadows compression as explained below. Note that film has different sensitivity to red, green and blue. This is taken care of during the printing process.

Dynamic range distribution of the Kodak Vision3 5219 film stock

I have also marked where 18% gray, 2% black and 90% white fall on the curve (for the green sensitivity curve). 18% gray or middle gray is what light meters use for light measurements. This is the shade of gray that falls perceptually in the middle of a black-to-white grayscale. 90% white is used as reference white in video and shows where diffuse white falls. Whites above this are generally specular highlights or in-frame lights. 2% black shows where the darkest detailed shadows fall. Below this, deep black with some tonal change is expected, but without real detail.

As we can see, there are around 3 stops below 2% where blacks are recorded, albeit compressed at the bottom and with less tonal resolution. And there are around 5 stops above 90% white for highlights. This is also the overexposure latitude. This latitude allows the cinematographer to overexpose in order to capture significant dark detail or to play with the look of the image during processing and printing. This allows for some contrast and grain modulation. Slight overexposure paired with pull processing (underdevelopment) and/or print down is common. The highest part of the curve is also compressed a bit, which means less tonal precision in this part. The point where shadows start to compress is called toe, and the point where highlights begin to roll is called shoulder.

It should be clear that the negative image is source material. If printed so that the curve is preserved, the image would appear very low contrast: washed and unappealing. That’s why release printing is done on high contrast positive stocks with gamma in the range of 2.5 to 3.0. This results in a print-through gamma of around 1.5 to 1.8. The print stock also does some further highlight compression through its toe. Blacks, on the other hand, are mostly unaffected due to the high maximum density over base of positive stocks.

The existence of a toe and a shoulder is the cause of one the defining characteristics of film, and consequently, of the cinematic look. The relatively large dynamic range paired with the compression of the extremes is the reason of the pleasant look of material shot on film in terms of range distribution: highlights seemingly roll off forever without clipping and there is a notion of tonality in the deep shadows.

Film preserves detail and tonality in highlights and shadows, with a pleasant roll-off in specular highlights

An interlude: gamma encoding and end-to-end gamma

Gamma is another often misunderstood area. The fact that the word is used for at least three different concepts in the image-making realm doesn’t help either. In the case of film gamma is the slope (or the tangent) of the linear part of the characteristic curve. In digital, gamma is used both as a synonym of transfer function or transfer curve, and as the value used for the exponent in the special case of power-law gamma encoding/decoding.

The dynamic range of the human eye is around 10 to 15 stops in a given moment of time, depending on lighting conditions. Displays and projection have smaller reproduction capabilities. Projection usually has intraframe contrast ratio of 150:1 or smaller. Good monitors may have intraframe contrast of around 1000:1. So the highlights and blacks compression above shoulder and below toe allows for squeezing a higher dynamic range into the smaller dynamic range of the reproduction system. It is, in essence, a case of tone mapping.

System gamma, end-to-end gamma or print-through gamma (in the film case) all describe the gamma of the whole process: from scene to the final deliverable. Replicating scene light would suggest system gamma of 1. But this is only true if the viewing conditions were equivalent to scene conditions in terms of light. This is rarely the case. Projection flare, low absolute projection luminance (less than 50 cd/m²) and the relatively dark viewing conditions lower display contrast significantly and make blacks appear brighter to the eye. The higher system gamma adds some contrast and combats these limitations. For example, film negative gamma of 0.6, intermediate film gamma of 1 and print film gamma of 3.0 lead to a composite gamma of 0.6 * 1.0 * 3.0 = 1.8. For the brighter viewing conditions in offices and homes an end-to-end gamma of around 1.2 is considered sufficient.

Linear light transfer (above) and power law gamma encoded transfer (below). Mid-gray background for reference. Note the limited tonal resolution in the blacks with linear encoding.

Gamma encoding in digital images serves a different purpose. Consumer grade images are universally 8-bit. If light is encoded linearly the dark stops have very limited precision: 2 is a stop higher than 1, 4 is a stop higher than 2, 8 is a stop higher than 4, etc. There are almost no values to encode intermediate shades. On the other hand, there is an excessive amount of values in the upper end: in the top stop between 128 and 255, for example. So linear encoding is both inefficient and losing important information in the shadows. Power-law gamma encoding addresses this by applying a transform (usually a simple power function) to the input signal. The eye still needs linear light in order to see the correct image so the display applies the reverse curve and linearizes the output. Decoding (reverse) gamma values between 2.2 (sRGB) and 2.6 (digital cinema) are used, depending on the expected viewing conditions.

Dynamic range of digital video

Digital sensors are more straightforward than film in terms of captured light representation. The quantized signal from the sensor’s analog to digital converter is linear. If a photosite (pixel) is capturing twice the light than another photosite, then its quantized value will be twice larger. Most DSLR cameras capture RAW images quantized to 14 bits. For a typical DSLR camera with slightly above 11 stops of dynamic range, 14 bits allow for some decent tonal resolution even with linear encoding. But things start to get complicated when the raw data have to be stuffed into less bits for recording.

Canon DSLR Standard picture style: dynamic range over light stops (log exposure)

All DSLR cameras and consumer video cameras output 8-bit video. Stuffing 11+ stops of dynamic range into 8 bits can’t be done linearly simply because the coding space lacks resolution. The typical compromise results into a gamma encoded S-shaped (over stops/log exposure) transfer curve. The top 8 to 9 stops of the RAW dynamic range are selected for transfer because they are cleanest. Some sort of a knee is usually implemented with the highest 1 to 1.5 stops getting compressed. The knee is very similar to the shoulder of film. It simulates a roll-off in the highlights, slightly increases the overall dynamic range and also allows for a bit more tonal precision in the mids where the most important tones are. The resulting image is sufficiently contrasty and ready for the consumer display. But it is not really supposed to be post-processed.

Again, I have marked 2% black, 18% gray and 90% white on the dynamic range chart for the Canon DSLR Standard picture style. Note that there is around one stop over 90% white available for highlights. Compare this to the excessive overexposure latitude of film. Shadows are better represented although the low stops are lacking in tonal resolution. An attempt to contain the highlights on exposure will often result in crushed blacks in high contrast scenes.

This type of consumer-ready transfer function plus the limited dynamic range of early digital cameras have led to the notion that digital video is too contrasty, highlights are hard clipped and the blacks are crushed and lacking detail. This is exactly what many people mean when they say that an image looks “video-ish”.

Blown highlights due to limited dynamic range. The front girl actually wears a chequered shirt. Also note the blown pink shirt in the middle. Like Crazy was shot on the Canon 7D. Compare to the shot from A Serious Man above.

A rare case of hard clip actually complementing the graphic presentation of a film. Sin City was shot on the Sony CineAlta HDC-F950.

Dynamic range distribution of the Arri Log C transfer curve

Recent high-end digital cameras have much better dynamic range capabilities and rival the best film stocks. Access to the full dynamic range is enabled through either linear RAW video (12 bit or more) or some (near) logarithmic transfer function. Both linear RAW video and log video are production formats and require post-processing for presentation. The idea of log space video is to provide a near flat distribution of coding values over exposure. Such a distribution provides both the full camera dynamic range and better tonal precision in blacks and highlights. Thus log curves are close to film characteristic curves, allowing for easier intercutting of digital video and scanned film footage. For example, the Arri Log C transfer function encodes around 14 stops of dynamic range from the Arri Alexa camera. Similar transfer curves have been constructed for many cameras, including DSLRs. It is worth noting that accommodating a large dynamic range into a limited coding space (such as 8 bits) results in limited tonal precision. This makes the practicality of true 8-bit log curves somewhat dubious. A 10-bit film scan allocates around 90 coding values per stop in the flat part of the characteristic curve, 10-bit Arri Log C allocates around 75 values. Whereas an 8-bit transfer curve like Technicolor’s CineStyle for Canon DSLR cameras allocates around 27 values per stop. That’s why low precision flat curves should be used with care and with understanding of the tonal precision trade-off. You can read more on 8-bit flat transfer curves here.

The previous parts of the Cinematic Look series can be found here: Part 1 on Aspect Ratio, Depth of Field and Sensor Size, and Part 2 on Frame Rate and Shutter Speed. And the next part is on Film Grain.

Cinematic Look, Part 3: Dynamic Range

Film frame rate

The frame rate of the motion picture specifies the frequency at which sequential frames are captured. In a film camera, this is the number of frames per second that pass through the camera gate and get exposed. In the days of silent film both motion picture cameras and projection cameras were hand-cranked. Cameramen took pride in their ability to crank at a steady pace. Common knowledge is that 16 frames per second (fps) was the prevalent rate of cranking. But this is not exactly so. Cranking speed varied wildly between 10 and 26 fps. Cranking speed varied even across the reels of a single movie. Projection was an art in itself. The projectionist had to watch the action closely and correct cranking speed when needed. On top of that, theater owners sometimes required projection speed to be increased in order to squeeze more showings in a given time frame – a practice that could lend a slapsticky feel even to the most serious drama.

Standardization came with the arrival of sound and the need to have sound in sync with picture. And ever since, the accepted standard for cinema frame rate is 24 fps. There wasn’t anything special about this number, other than being the approximately average projection speed from a bunch of sampled theaters in 1926. Nevertheless, it has come to define some of the prominent characteristics of the cinematic look.

Shutter angle and shutter speed

The second temporally important property is the period of exposure for each frame. In a motion picture film camera the negative is not exposed continuously. A window of time, with light shut out, is needed in order for the sprocket wheel to pull the next frame in the film gate and prepare it for exposure. In a film camera this is what the rotary shutter does.

Schematic of a half-moon rotary disc shutter positioned beside the film gate

The rotary shutter is an arc shaped mirror (called “half-moon”) that rotates in front of the gate. Its pivot is placed either beside or underneath the gate. When the gate is covered, the mirror serves to reflect the image from the lens into the ground glass, so that the cameraman can see it in the viewfinder. At the same time the next frame is being loaded into position behind the shutter. When the shutter is rotated away from the gate, light exposes the negative. The exact shape of the arc defines the shutter angle. The shutter angle is specified in degrees and describes the size of the cut out part of the shutter disc.

In simple cameras the shutter is shaped as a semicircle (and the shutter angle is 180 degrees). In more advanced cameras the shutter angle (and thus the shape of the shutter) can be changed. The shutter rotates with a constant speed and makes one revolution per frame. So for the standard cinema frame rate of 24 fps that means 24 revolutions. Bigger angle means longer exposure time. Smaller angle means shorter exposure time. In photography and cinematography the exposure time is often called shutter speed, because the exposure time is the time the shutter stays open per frame. The shutter angle can easily be made small, but large angles are harder because of the need for next frame advancement.

Various shutter angles and their respective exposure time as a fraction of total frame time

There is another popular rotary shutter type, called “butterfly”. It consists of two segments (90 degrees each, for 180 degree shutter angle) positioned opposite of each other. Butterfly shutters rotate at half the speed of a half-moon shutter. That is, a full revolution exposes a couple of frames. Panavision Panaflex cameras usually have this design.

Early cameras usually had a fixed shutter angle. The angle itself varied between cameras. For example, Bell & Howell Eyemo 71k used a 160 degree shutter; the 16mm B&H Filmo 70 DR used a 204 degree shutter. Eventually a shutter angle of 180 degrees ended as standard or “normal”. For a frame rate of 24 fps this equals shutter speed of 1/48 sec. The shutter angle can be converted to shutter speed with the following formula:

The following table conveniently lists some useful shutter angles converted to shutter speeds.

Motion blur and strobing

There are a couple of artifacts that arise from the rotary shutter and the 24 fps frame rate.
As the camera only ever sees half of the time (for a typical 180 degree shutter), it doesn’t capture the scene continuously. This means that fast moving objects, and especially objects moving across the frame, will exhibit jerky movement. This is called strobing. The defect is also very noticeable during pans.

The relatively slow shutter speed and fast motion in the frame result in motion blur

The other artifact is also related to motion. Because of the relatively slow shutter speed (1/48 sec for 180 degree shutter angle and 24 fps), fast moving objects blur in the frame, because the longer the exposure, the more movement is captured. This is what we call motion blur. Note that while strobing is in essence an artefact related exclusively to the no full time shutter, motion blur results from the (relatively) long exposure of each frame.

Smaller shutter angles (shorter exposure) exhibit more pronounced strobing effects. Bigger shutter angles (longer exposure) increase motion blur. Faster frame rates can smooth out the perception of strobing, even with shutter angles smaller than 180 degrees. Faster frame rate also decreases captured motion per frame and decreases motion blur per frame. Note that small variations in shutter angle (and shutter speed) are imperceptible for most of the audience. Most people won’t notice any difference between a 180 degree shutter angle (1/48 sec shutter speed) and a 144 degree (1/60 sec shutter speed).

Top: 180 degree shutter will produce typical motion blur and strobing. Middle: decreasing shutter speed increases blur and decreases strobing. Bottom: increasing shutter speed decreases blur and increases strobing.

Motion blur and strobing are the main ingredients of what people call the dream-like effect of cinema. They set apart the cinematic image from the crisp and fluid reality and reinforce the unreal feel of cinema. This makes them an important characteristic of the traditional cinematic look.

Digital cameras

In digital video cameras (and DSLRs) there is no real need for a rotary shutter as there is no film negative that needs moving around. Nevertheless, some high-end digital cinema cameras like the Arri Alexa Studio and the Sony CineAlta F65 use rotary shutters to closely simulate the exposure process of film cameras while letting light on the sensor. But the vast majority of digital cameras only have electronic shutters. The camera simply reads out the sensor (or parts of it, for CMOS sensors) simultaneously ending its exposure and resetting it for the next frame.

One advantage of the electronic shutter is the ability for full time exposure (or 360 degree shutter angle equivalent). Rotary shutters need to be closed for some time after each exposure so that the sprocket wheel moves the next negative frame into position. No such movement is necessary in digital cameras. This allows for exposure times as low as 1/fps seconds. A disadvantage is the jello effect (image skew) that may happen with fast pans or fast moving subjects. This is a characteristic of the rolling shutter – typical for CMOS sensors – and is the result of partial sensor readout: parts of the sensor continue exposing while other parts are being read. High-end sensors with their fast readout times tend to minimize this defect.

Setting the digital camera to 24 fps and 1/48 sec shutter speed emulates the way film cameras work pretty close. Some digital cameras (especially DSLRs) don’t have an option for 1/48 sec. But setting the camera to a shutter speed of 1/50 sec will give results virtually indistinguishable from 1/48 sec.

The higher frame rate debate

Recently there has been a push towards higher frame rates, especially in connection to 3D movies. Peter Jackson is shooting The Hobbit in 48 fps and James Cameron will allegedly shoot Avatar 2 and 3 in 60 fps. The reasoning behind this is that higher frame rates result in more fluid and crisper image compared to 24 fps and the typical 180 degree shutter angle. This proposed change has been met with polarized reactions. Many find this image bland, TV-like and lacking the dramatic feel of 24 fps (at 180 degree shutter). Nevertheless, a higher frame rate will probably be accepted as standard alongside 24 fps. If high frame rates take off this may lead to a shift in perception about what is considered cinematic in terms of temporal characteristics. But for now, 24 fps and a 180 degree shutter angle define the traditional cinematic look. This is the look games (in cutscenes) and video emulate when trying to be cinematic. You can read more in the articles on high frame rates and 3D movies and frame rate as artistic choice.

The next part of the Cinematic Look series is on Dynamic Range.

Cinematic Look, Part 2: Frame Rate and Shutter Speed

To better illustrate this we need to go back in time. Since the arrival of sound cinema frame rate has been fixed to 24 fps. As such, the choice of frame rate was not an option for filmmakers. There were tries to introduce higher frame rates but these never really took off, mostly because higher rates require more film stock. And this obviously means the price of filmmaking would go up. With digital movie making this is less of a concern – storage is cheap.

But we should actually go further back in time. Before talkies. It is a common misconception that silent films were shot at 16 fps. Cameramen claimed that they had hand-cranked at this speed; some cameras even had indicators for 16 fps to help hand-cranking. This myth was busted by Kevin Brownlow in an article from 1980. During his work on restoration and conversion to tape of silent films he found that 16 fps was not the norm. Cranking speed varied widely between 12 and 26 fps. And higher (than 16 fps) speeds were common. There was also a tendency towards high frame rates mostly based on the habit of theater managers to have their projectionists cranking at higher speeds in order to squeeze more shows in the busy evening schedule. This would inject a slapstick feel even into the most serious drama. To counter that, directors and cameramen would increase speeds to ctach up, hoping that projection would look about right.

It is popular knowledge that the frame rate of talkies (24 fps) was selected as the minimum rate that can yield a decent optical track. This is not exactly so. When Western Electric had to select the frame rate for their process they conducted a survey in a bunch of movie theaters. 24 fps turned out to be about the average projection speed in these theaters. So they selected it. In fact, rival sound processes had the frame rate as low as 21 fps. The point is, there was not anything really scientific or special about the choice of 24 fps. It was largely arbitrary. So it is to a large degree coincidental that the established look of cinema has the dream-like qualities it happens to have. This was not intentional. But all this by no means diminishes these qualities. One way or another, they are here, and we’ve come to love them.

Home Sweet Home has the frame rate rising from reel to reel. And then the last reel is very slow.

Now, how conscious were the directors of the silent era of the artistic side of frame rate is debatable. Maybe they had external reasons for specific choices. For example, low fps means less film stock used. Which leads to a lower price and extended scenes in a reel. Maybe the variations were unintentional. Nevertheless, this provokes additional thought.

So back to the topic at hand. Democratizing frame rate choice can be a good thing. While too many available frame rates may lead to chaos, having available 24 fps plus a higher rate (48 or 60 fps) is something that deserves consideration. Frame rate can be used as a mean to further an artistic vision. The obvious example is with films striving for perceived realism, or films going for a documentary feel. No doubt some of them can benefit from more fluid and crisp action. This could help make the viewer a part of the scene. To make them experience it in a more visceral way. Other (most?) films are better suited for the traditional cinematic illusion. And I believe that existing movies should be left alone and not upconverted to high fps. But having the option to go for a higher frame rate with a film increases the creative possibilities. The same way a filmmaker can choose film or digital, a specific film stock, lighting style, framing, etc. they would be able to choose a frame rate that promotes their vision in the best way possible. It is also good to mention that speeds lower than 24 fps are being used in 24 fps movies for artistic effects. This is achieved by either undercranking the camera, or shooting 24 fps and then dropping frames. In editing, frames are duplicated as many times as needed to achieve correct speed when projected at 24 fps.

But then again, having too many variables can confuse people and may also lead to wrong or arbitrary choices. Still, the latter is not really a good argument. The lack of restraint or of understanding of a specific variable of filmmaking is not an excuse. Confusion amongst the audience on the other hand is something that should be considered. The audience of the silent film wasn’t conditioned to a specific frame rate in the way that we are with 24 fps. Getting used to high frame rates may lead to 24 fps movies being rendered unwatchable for some viewers. But perhaps this is a risk worth taking.

Then there is the option of variable acquisition frame rate. Even the critics of Jackson and his 48 fps endeavor admit that his aerial, scenery and establishing shots look spectacular. Think of Discovery or National Geographic. Variable frame rates can give us the best of both worlds: motion blur and strobing for dramatic impact, and crisp, fluid images for landscapes and panoramas. This is easy to achieve technically once theaters start supporting high frame rates. For example, a 48 fps master can simply double frames for the 24 fps sequences. This will fully preserve low frame rate aesthetics where necessary.

Frame Rate as Artistic Choice or What Can We Learn from Silent Films

But 48 fps actually comes with benefits. Well, for 3D at least.

Cinema and reality

Motion blur and strobing are the main ingredients of what some label the dream effect of cinema. These artifacts of relatively low frame rate and less-than-360 degrees shutter angle separate the cinematic image from reality and their abnormality reinforces the unreal nature of cinema. This is something cinema viewers have come to appreciate. This characteristic of cinema has stayed more or less the same after the arrival of sound, after the introduction of the wide screen and after the advance of CGI.

The Hobbit breaks new ground with its 48 fps presentation

Quake players would often run the game with a frame rate into the hundreds for maximum fluidity and responsiveness

3D movies

3D changes things for cinema. Most importantly, it blurs the boundary between the fantasy of cinema and reality by tricking the brain to think that what it sees on the screen is not a screen at all, but deep and three dimensional. This is in direct contradiction with the otherworldliness of cinema, as defined by its most famous image artifacts. So, in a sense, in the moment we put 3D in the equation, we clash with the dream-like nature of cinema. And even more: one might say that with 3D we give up the dream nature of cinema. But there is more than just philosophical reasoning.

Problems with non-coinciding eye convergence and eye focus aside, 3D simply does not coexist well with strobing and motion blur. The already strained eyes get additional load while trying to discern depth plans because of the lack of crisp object edges and the jerky movement. Selective focus does not help either because the eye can’t wander freely through the scene (but that’s unrelated to frame rate). So 3D tells the brain “This is real!”. On the other hand, the motion blur and strobing artifacts obstruct this perception.

So we can argue that 3D simply is not cinematic. With its life-like aspirations 3D is more suitable for interactive representations of reality like games and, ultimately, virtual reality. But – like it or not – 3D is here anyway.

So there is really only one logical direction to go from here in the 3D cinema case. Motion blur and strobing need to go. 48 fps will no doubt help tremendously in terms of both fluidity and crispness. And 3D cinema becomes reality – for good or bad – thus resolving the contradiction. And if we are lucky, 2D movies will stay 24 fps so that we have our opposition to reality.

The familiar fluidity of TV and video games is the main reason younger generations will most likely embrace 48 fps without questions. That, and probably being more open to changes. They have sunk so many fluid images through their TV and games entertainment, that actually it won’t be surprising if the strobing film look is going to be considered weird and severely outdated in a few years. And the hyper-realism of the “be in the scene” factor will no doubt sell the concept to many others.

(There are some further thoughts on the topic and an interesting reference to silent films in my follow-up article on frame rate as artistic choice.)

Cinema and Reality, or Why 48 fps is Good for 3D Movies

Before we start, let’s make it clear what does “cinematic look” actually mean. For us, cinematic look is what audiences have come to expect from a motion picture in terms of appearance, or in other words in terms of visual perception. This is the image we have been culturally conditioned to consider as cinematic through decades of exposure to movies. And this is what we are trying to replicate with digital cameras when aspiring to achieve the “cinematic look”. Note that the cinematic look is historically “film look” as movies were almost exclusively shot on film for some hundred years.

The aspect ratio

Silent films were generally shot in 4:3 (1.33:1) aspect ratio

For years wide format moving images were associated with cinema and the more squarish 1.33:1 format was linked to TVs. At least, that’s what we were conditioned to imagine. With the introduction of wide TVs things change a bit. HD and full HD television screens now have aspect ratio of 1.78:1 (1920×1080 pixels for full HD; 1280×720 pixels for HD). In comparison, the most popular cinema aspect ratios are currently 1.85:1 and 2.39:1 (often labeled 2.35:1 for historical reasons), for flat and anamorphic projection respectively. Which means TVs are now much closer in geometric appearance to cinema screens. Cinematic ratios weren’t always like this, though. Silent films were 1.33:1. Talking pictures were 1.375:1 for some twenty years till 1952. For various reasons, around this time the widescreen revolution in cinema happened with the forementioned wider aspect ratios becoming prevalent.

Artistic reasons for choosing a specific aspect ratio notwithstanding, it is correct to assume that wide image is nowadays associated with cinematic appearance. Up to some limit, at least. Incidentally, using anamorphic adapters on HD cameras may yield images with extracted aspect ratio of up to double the HD ratio (or 3.56:1), which can be way too wide, so some cropping of the sides is recommended in such cases. But use of anamorphics on consumer and prosumer digital cameras is a topic for another article. The bottom line is, 1.78:1 video is already quite cinematic in geometric appearance. But if one so desires, it is safe to consider a slight cropping of the top and the bottom to bring it to 1.85:1 or even 2.39:1. Have a look at this article for more in-detail thoughts on aspect ratio choice for a specific project.

Cinema aspect ratios

Note that in the above two paragraphs we are talking about the aspect ratio of what you see on screen. This is not necessarily the same ratio as the recorded image, especially when the recording medium is film.

Sensor size and depth of field

In terms of aesthetics the most apparent property linked to sensor size or film frame size is depth of field. Popular understanding is that smaller sensors yield greater depth of field and, conversely, large sensors have less depth of field. Technically, if we shoot the same composition with two differently sized sensors, and:

1. with lenses covering the same angle of view, i.e. wider lens for the smaller sensor and longer lens for the larger sensor;
2. with the same camera-subject distance;
3. with the same aperture diameter;
4. enlarge the result to the same print or screen size (or resize to the same video pixel size, let’s say 1920×1080);
5. and use the same criterion for sharpness (i.e., the circle of confusion is proportional to the sensor size),

then both pictures will have exactly the same depth of field. Your experience tells you otherwise? The tricky part is number 3). Same size aperture does NOT mean same f-number because the f-number equals the focal length divided by the aperture diameter. Which means that for the conditions above, the longer lens (used on the larger sensor to achieve equal angle of view with the small sensor) will shoot at a bigger f-number in order to maintain the same physical aperture.

On the other hand, if we retain all conditions but change 3) to “at the same f-number” (which, by the way, should also, more or less, preserve the same exposure, considering we shoot at the same ISO), then the smaller sensor will indeed yield greater depth of field. This is because the wider lens (used for the smaller sensor) will have a smaller aperture opening at this equal f-number. So we can conclude that lenses with equal angles of view, shot at the same f-number on differently sized sensors (or film frames, for that matter) manifest different depth of field, with smaller sensors giving pictures with greater DOF.

For this article we will ignore other properties related to format size. In the digital case these include dynamic range, sensitivity and noise (all three are, more precisely, connected to sensor pixel size). For film, different film frame sizes (but from the same emulsion) will show different grain sizes when projected on the same screen (or printed on the same release stock).

Camera aperture and projection aperture

Motion picture film cameras have a rectangular film gate in front of the negative, which defines the portion of the frame getting exposed. This portion of the negative (and often the gate itself) is called camera aperture. When projecting a release print in a movie theater another gate is used in front of the film called projection aperture. The projection aperture is slightly smaller than the camera aperture to allow some safety margin for imperfect alignment of the film roll. It is essentially a window in the camera aperture area. So the audience in the theater never sees the full image as it was recorded but a slightly cropped version. Camera and projection aperture may also vastly differ when different aspect ratios are involved in shooting and projection: for example, when a movie is shot in 1.33:1 but projected at 1.85:1. All this makes film frame area measurements somewhat ambiguous. In this text for consistency when talking about various film formats we will mean the standardized camera aperture unless “projection aperture” is explicitly stated.

Frame sizes and aspect ratios of popular film formats

A frame from Kodak Gold 35mm stills negative film

Stills photographers coming to videography sometimes wrongfully assume that 35mm motion picture frames are the same as stills 35mm frames. A full frame for stills photography is sized 36mm x 24mm, or, rather, this is the exposed area of the frame (or the camera aperture). Note that film rolls are oriented horizontally in a stills camera with film perforations at the top and the bottom of the frame. On the other hand, film used for motion pictures is (usually) oriented vertically (perforations at the sides). For 35mm film the longer side is around 24mm and the shorter side around 18mm. The exact frame height depends on how narrow is the frame line. The standard negative pulldown (or film pulldown) for movies is 4 perforations per frame (4-perf): the camera sprocket wheels pull four perforations from the film roll for each frame.

35mm silent frame

35mm Academy format frame, shown with the space reserved for soundtrack

3-perf Super 35 frame

Techniscope is a 2-perf format

Silent film utilized the full area of the frame for recording images because there was no need to leave space on the negative for sound. The camera aperture of silent film was 24.89mm x 18.67mm (.980″ x .735″) with 1.33:1 aspect ratio.

For talkies the image area shrunk in order to accommodate the soundtrack on the release print. In 1932 sound pictures camera aperture was set to 22.05mm x 16.03mm (.868″ x .631″), with the projection aperture set to 20.1mm x 15.24mm (.825″ x .600″) and 1.375:1 aspect ratio.

Wide screen formats varied a lot through the years. Anamorphic formats utilized a frame size similar to the Academy format but in order to achieve widescreen ratios anamorphic lenses were used to squeeze the image while shooting and then unsqueeze it on projection. But we will leave anamorphic formats out for this article and focus on flat formats as they relate easier to digital sensors. The current wide standard for shooting flat is Super 35. Super 35 is a production standard, meaning it gets resized when printed. This also means there is no need to leave space on the negative for sound as the printing is not 1:1. Super 35 was originally a 4-perf format sized 24.89mm x 18.67mm (.980″ x .735″). This is a 1.33:1 format so frames were matted down (to 1.85:1 or 2.39:1) for release. This also means a lot of the frame was wasted, so currently a 3-perf version sized 24.89mm x 13.87mm (.980″ x .546″) is used in order to maximize frame utilization. This saves around 1/4 stock length compared to 4-perf. 3-perf still gets cropped a bit when printed but wastes much less negative than 4-perf.

Various wide-screen apertures have been used through the years. VistaVision was a 8-perf horizontal format developed by Paramount in the 50′s and similar to 35mm for stills. With camera aperture sized 37.7mm x 25.17mm (1.485″ x .991″) it offered great image quality. Most productions were matted and printed down to standard size 1.85:1 format vertical prints for theatrical release. Despite the exceptional quality VistaVision didn’t pick up because of the higher stock costs in comparison to anamorphic formats. Since the 60′s it’s been used mostly for special effects work requiring greater resolution.

On the other side of the spectrum was Techniscope introduced by Technicolor Italy in the early 60′s. This was a 2-perf production format meant to save film stock by sacrificing a bit of image quality. It used a camera aperture sized 22.05mm x 9.47mm (.868″ x .373″). Techniscope pictures were shot flat then printed with 2x vertical enlargement factor to be projected anamorphically. Being 2-perf, during production it used half the stock compared to 4-perf but resulted in larger grain and less clarity.

North by Northwest, like other Hitchcock movies from the second half of the 50's, was shot in VistaVision

Then there is also 16mm film, which is widely used for documentary, TV and occasionally for cinema work, especially for indie films. Super 16 (which is the analog of Super 35 in the 16mm world) has camera aperture sized 12.52mm x 7.41mm (0.493″ x 0.292″). Recent award winning films shot in Super 16 include The Hurt Locker, Black Swan and The Wrestler.

Digital sensors

For a long time TV cameras used analog pickup tubes to convert optical images into electric signals. The imaging area of the tube is usually 2/3 of the diameter of the tube. Standard tube diameters included 1 inch, 2/3 inch, 1/1.8 inch. You probably notice similarity with digital sensor size categories. Indeed, sensor sizes like 2/3″, 1/1.8″, 1/2.3″, Four-Thirds, etc. are named like this for historical reasons related to analog TV and video cameras. Each of these has an imaging diagonal roughly equal to the imaging diameter of a tube of that size (remember, the imaging diameter of the tube is about 2/3 of the overall tube diameter). For example, a typical 2/3″ sensor will have a diagonal of around 11 mm. This is roughly equal to 2/3 of 2/3″.

Modern HD digital video cameras normally use a 16:9 sensor. The typical aspect ratio of the digital sensor in a photo camera is either 3:2 (mimicking film stills) or 4:3. But when shooting video with a photo camera only a 16:9 portion of the sensor is used with pixels in the top and the bottom getting discarded. One consequence of this is that crop factors often used for comparison of photo camera sensors are not always accurate in terms of video. Modern video is predominantly widescreen and cropping practically always happens at the top and/or the bottom thus keeping the original width of the image unchanged. That’s why cinematographers and videographers often use the sensor width when comparing sensors in terms of DOF instead of the usual diagonal measurement as used in crop factors for stills.

Considering this, the following table lists some film format frame and digital sensor sizes with only the approximately 16:9 (or wider, where 16:9 is not applicable) area taken into account, sorted by width.

So how do we interpret these in terms of depth of field?
It is a common understanding that TV and video have relatively big apparent depth of field. And we can easily see why this is the case. First, TV cameras tend to use relatively slow zoom lenses with relatively small apertures. Second, and more important, as seen above they have a much smaller imaging area compared to both film and large digital sensors.

One can often read on forums statements like “I like Canon 5d Mark 2 because of its cinematic DOF”. Statements like this can be attributed to years of visual opposition TV vs Cinema and the consequent automatic generalization: TV has lots of DOF, cinema has shallow DOF. This is not always true. The correct statement is “cinema can have shallower depth of field than TV”.

One look at the table shows that Full-Frame DSLRs actually have much larger sensor size than the typical widescreen motion picture frame size (i.e. Super 35 and classic widescreen). This means they may demonstrate excessively shallow DOF compared to motion pictures shot on film when pictures are shot at the same f-number/exposure (see above). APS-C sized sensors are actually much closer to the typical film frame size. No wonder that digital cinema cameras that claim “Super 35″ sized sensors actually utilize APS-C sensors. This doesn’t mean that APS-C sensors are better than Full Frame. There are other reasons to use sensors larger than APS-C: low light sensitivity, dynamic and color range, overall image crispness (this last one is often “lost in compression” in DSLR video). And the shallow depth of field fetish, of course.

Sergio Leone shot Once Upon a Time in the West in Techniscope, with a frame size smaller than APS-C

A small non-technical digression. There is also the aesthetic side of DOF. Some of the greatest films in history sought to get deep focus by either using wide lenses exclusively or pooling tons of light on set and shooting at small apertures. Others ended with relatively bigger DOF for technical reasons: smaller film frame sizes (compared to Super 35 and APS-C). It is telling (and a bit ironic) that Paramount promoted their VistaVision process as a deep focus vehicle, mostly based on the availability of a 28 mm lens – one of the widest at the time. It was not shallow focus that tempted filmmakers, but rather image clarity, depth and wide angle possibilities. Generally, movies are supposed to represent objects in relation to their surroundings. This requirement implies sufficient DOF in order to visualize these relations. Selective focus is certainly a great tool for isolating subject matter and commanding the viewer’s eye. But shallow DOF is just that: a mean, not a goal. Very shallow DOF can surely be a good aesthetic for certain scenarios (recently Tinker Tailor Soldier Spy relied on shallow DOF, mostly by utilizing longer lenses), but these tend to be the exception, not the norm. So we can nevertheless argue that shallow depth of field is not an implicit characteristic of cinema because there are lots of influential movies heavily utilizing deep focus shots. There are other properties more intimately associated with the cinematic look. Some of them are in the focus of the next part in this series.

We can safely conclude that in the DSLR realm (and in the digital sensor world, in general) APS-C is the closest representation of the cinematic look in terms of DOF.

Cinematic Look, Part 1: Aspect Ratio, Sensor Size and Depth of Field