Grain - Guillermo Algora - Visual Effects Compositor

1. Primary Properties:

1.1 Intensity.

1.2 Size.

2. Secondary Properties:

2.1 Pattern Irregularity.

2.2 Softness.

1.1 Intensity:

Intensity is the amount of negative and positive multiplication of the image by the noise, resulting in the strength and color of the grain. The intensity is mainly dependent on:

• Luminance:
  
  

In film, the maximum intensity is around the mid-low values, followed by the mid-high. From there, it makes a sort of a bell curve with a dip (falloff) on both sides. The darkest areas are usually clean of grain because they have not been exposed to light (hence the silver particles cannot react). At the other end, the brightest areas (highlights) have received more light, causing a larger number of the silver crystals to react, which translates to a more uniform distribution of grain and rendering it less visible.

In digital, although the distribution is more uniform across the luminance curve, darks present the strongest intensity due to the digital interference being more noticeable against the blacks (there is always a certain amount of electronic interference without necessarily light), while the highlights on the other end, have the weakest perceived intensity due to the overall amount of light present. This ratio is referred as signal-to-noise ratio (SNR), which quantifies the relative strength of the image signal to the noise.

• ISO:

In both film and digital, the higher the ISO the stronger the intensity. In film, this is due to having larger silver crystals in the emulsion to increase sensitivity, while in digital due to the camera electronically amplifying the signal (voltage output from the image sensor) which also amplifies any electronic noise present within the camera's sensor circuitry.

Film grain is generally more color neutral and consists mostly of just brightness variations, while digital noise is made of both luminance and color differences (color "fizzle" between channels, often visible as random red, green, and blue pixels in the image). Film has the most intensity in the blue channel (due to its historical sensitivity to blue light), while digital tend to have more uniform noise characteristics across color channels.

1.2 Size:

Grain pixels clump together to generate grain of a particular size.

• High frequency (small): usually no more than a pixel.
• Medium frequency.
• Low frequency (large): may be large blobs of changing texture.
  
  

Size is determined by:

• Luminance:

Film: the size of grain can change according to the luminance of the image, with darker areas of the image having larger grain. The appearance of grain is finer in brighter areas because the silver crystals become fully exposed more quickly (the reaction completes quicker, leading to a smoother texture), whereas in shadows, the crystals may not fully develop (shadows are still maturing) before the shutter closes (leading to more visible grain).

Digital noise, on the other hand, generally tend to maintain roughly the same size (across the luminance curve) within a single RGB channel, being much more uniform than the clumpy film grain

• ISO:

In both film and digital, the higher the ISO the larger the grain.
Higher ISO film has larger crystals to capture more light, resulting in larger, more noticeable grain (these larger crystals develop into larger grain particles on the film, making the grain itself physically larger). With digital, high ISO settings amplify the signal from the sensor, but also amplify noise. The amplification process can cause the noise pattern to spread over larger areas, affecting clusters of pixels rather than individual ones, thus making the grain appear larger.

• Format:
  
  

The larger the film/sensor size, the smaller the grain size.
Smaller film formats tend to have larger (and more noticeable) grain because the size of the individual silver crystals (grains) is larger relative to the overall image size. In digital formats, the concept of grain is replaced by digital noise, which is influenced by the sensor size and pixel density. Larger sensors (such as those in full-frame cameras) generally produce less noise than smaller sensors (such as in crop-sensor cameras), similar to how larger film formats produce less visible grain.

2. Secondary Properties:

2.1 Pattern Irregularity:

Film has a more clumpier and irregular grain appearance as it results from the random distribution of silver crystals in the film emulsion. Each grain of silver crystal reacting differently to light exposure, creating a non-uniform pattern that adds a unique texture to the image. Higher ISO in film results in more clumps due to larger chunks of silver used to increase sensitivity.

Unlike film grain, digital noise often has a more structured and predictable pattern due to the uniform nature of the sensor's pixel array and the electronic processes involved. It can often exhibit patterning, in the form of noise grouping into little vertical or horizontal shaped clumps or in dot arrays. Some digital cameras even produce exactly the same noise pattern in each RGB channel, but with different intensity, and so require the same seed for their simulation.

There are three common types:

• Random noise is characterised by fluctuations in both intensity and color. There is always some random noise at any exposure length and it is most influenced by the ISO speed (higher the ISO, stronger the noise).
  
  
• Fixed pattern noise includes what are called "hot pixels," greatly exceeds . This type of noise is common in very long exposures (therefore could be on low ISO) and is unique he surrounding ambient noise which occur when a pixel's intensity far exceeds that of the fluctuations of the surrounding random ambient noise. It generally appears in very long exposures and is unique in that it will show almost the same distribution of hot pixels if taken under the same conditions (temperature, exposure, ISO).
  
  
• Banding noise is highly camera dependent, and is introduced by the camera when it reads data from the digital sensor. Banding noise is most visible at high ISO and in the shadows (when an image has been excessively brightened).

2.2 Softness:

Grain resembles a Gaussian like noise pattern (both in film and digital, although especially for the later at low ISO) because:

• A Gaussian distribution is symmetrical around its mean. This means that there is an equal probability of values being above or below the average value (mean). In digital noise, this translates to a similar chance of a pixel being brighter or darker than the "correct" value due to the random fluctuations in signal strength.
  
  
• Most values cluster around the mean, with fewer and fewer values occurring further away. This aligns with digital noise, where the majority of pixels will have a signal close to the intended value, with a smaller number deviating significantly due to noise.

And therefore, sometimes to match the plate grain a secondary Gaussian blur or sharpen applied specifically to the grain is required.