Film grain is part of the image.

Grain makes more sense if we begin with the piece of film, not the texture. Film does not capture a perfectly smooth image and acquire grain later. The image itself is formed in a physical coating made from countless microscopic particles.

From a normal viewing distance, the tiny variations in that coating blend into continuous tone. Enlarge the image enough and the same variations become visible as grain. Tone and grain are not two unrelated layers. They are two scales of the same image-forming structure.

The interactive figure follows those four steps using black-and-white film, where the material change is easiest to see. Move through it once from left to right. The same patch stays on screen the whole time.

The drawing is schematic: one visible speck is not necessarily one original crystal. What we call grain is the larger pattern produced by developed clumps, gaps, local density differences, and, in colour film, dye clouds.

Colour negative film follows the same basic sequence, but its finished image is made from dye rather than metallic silver. Development forms dye around the activated sites. Bleaching and fixing then remove the silver and unused crystals, leaving the dye image behind. Section 04 returns to those colour layers.

Before following grain into the print, one more idea needs to be clear: film density and picture brightness are not the same thing. A negative reverses light and dark. Printing reverses them again.

Graininess and granularity are different words

Graininess means how grain looks and feels to a viewer. It changes with the scene, viewing distance, enlargement, contrast, colour, motion, and the person looking. Granularity is a laboratory measurement of how much density varies across a small test sample. Two granularity numbers are only directly comparable when the samples were measured in the same way.

The measurement is useful for comparing materials. It still cannot predict exactly how grain will feel in every shot, because the picture, processing, enlargement, and viewing conditions all change what the eye notices.

A cinema print is not usually the same strip of film that ran through the camera. The camera negative records the scene first. That negative then exposes a second strip called print stock, which is developed separately and projected for viewing.

Both strips are physical emulsions. The negative brings the first grain pattern into the print, and the print stock forms a new pattern of its own. The finished image can therefore contain two independent grains.

The negative's grain is copied into the print, but it does not pass through unchanged. The print changes the image's contrast, making some differences easier to see and squeezing others together. The print's own grain is formed later, in a different random arrangement. It is not a resized copy of the negative grain.

This map shows the shortest negative-to-print path so the two sources stay easy to follow. Some cinema workflows make additional film copies between the camera negative and the release print. Each extra piece of film can reshape the image again.

So before asking where grain should be strongest, ask what you are meant to be looking at. A direct scan reads the negative. A projected print shows the negative after a second film stock has reshaped it and added its own grain. Both can look like film, but they do not distribute grain across the picture in the same way.

A film print has a limited range. It can only become so thin at the bright end and so dense at the dark end. Near either limit, two nearby tones get pushed closer together, so the difference between them becomes harder to see. Through the middle, the print has more room to keep those differences apart.

Grain is made from small density differences, so the same rule applies to it. Negative grain usually survives most clearly in the midtones. Near bright white and deep black, the print squeezes more of that variation together.

Film curves give names to the two ends of this range. The thin, bright end of the print is its toe. The dense, dark end is its shoulder. When this article says the print “compresses” grain, it simply means that nearby density differences are pushed closer together and become less visible.

Colour film does not record a finished colour in one layer. It records three overlapping versions of the scene: one layer responds mostly to blue light, one to green, and one to red. Think of them as three image records sharing the same strip of film.

A subject colour changes those records by different amounts. A red object strongly affects the red-sensitive record; a blue object strongly affects the blue-sensitive record. During development, each record forms its own dye. Because this is a negative, those dyes are complementary colours. The visual below follows that reversal all the way back to the viewed positive image.

The dye names are less important than the relationship: a red patch and a blue patch are built from different mixtures of layer records. When those records vary independently, their visible grain cannot be the same grey speckle.

This is the practical meaning of chromatic grain. It is not arbitrary colour noise added for style. It begins as small, independent differences between the negative's colour records. Printing carries those differences into the colour variation we finally see.

In the deepest shadows, the dense print pushes many of those layer differences toward the same dark limit. The grain tends to look more neutral there. Through the midtones and lighter colours, the layers have more room to disagree, so the colour of the grain is easier to see.

This is why simply adding saturation to grey noise is not enough. It creates more colour, but it does not make the colour respond to the material underneath it.

These four ideas often travel together, which makes them easy to blur into one vague idea of “more film texture.” They describe different parts of the image, though. Keeping them separate makes every grain control easier to judge.

Stock and speed set the starting point

A film stock is a particular material recipe. Its crystal mixture, colour layers, and intended development process set the starting character of its grain. A faster stock is designed to record an image with less light. That often requires more sensitive material and leads to higher granularity, but stock design matters too. “Faster means bigger dots” is only a rough shortcut.

Exposure and lab processing can change how the same stock appears. A push usually means giving the film less exposure and developing it more to compensate, which can make grain and contrast more obvious. A pull uses more exposure and less development. The material has not changed, but the way it was exposed and processed has.

Format changes enlargement

Format, also called gauge, is the physical size of the recorded camera frame. An 8mm negative must be enlarged much more than a 35mm negative to fill the same screen, so its negative grain appears larger. The print is a separate piece of film at its own size. Changing the camera format does not secretly resize the print stock.

Grain is not a blur control

Sharpness here means how finely the film can resolve detail. Stock design affects both sharpness and grain, but they remain different properties. Increasing a grain-strength control should not quietly soften the source at the same time. If it does, you cannot tell whether you prefer the grain or the hidden blur.

Real grain also contains a mixture of fine, medium, and larger variations. Engineers call this a broad spread of spatial scales. If every mark belongs to one repeated blob size, the texture announces its pattern and starts to look synthetic.

Motion-picture grain is not a still texture being moved across the frame. Each exposed frame presents a new patch of emulsion, so its microscopic arrangement is new. The picture content may hold still; the grain pattern should not.

Repeating a short grain clip, holding one pattern for several video frames, or sliding a fixed texture eventually reveals a rhythm. It stops feeling like material and starts feeling like an animated layer.

The illustration above isolates the timing. The clips below show it in matched Resolve renders, with the grain model on and off.

Watch the cheek, hair, and out-of-focus lights rather than the edge of the face. The grain-on image has local activity without a crawling sheet detached from the subject. Human vision also tends to be less sensitive to rapid colour changes than to equally strong brightness changes, which is one reason chromatic grain can feel gentler in motion.

Everything so far can be turned into a practical checklist. A convincing digital model does not only need to generate a pleasing texture. It has to preserve the relationships that made the texture: what material formed it, what picture sits underneath it, how large the film was, and when the frame changed.

Why the common shortcuts fall short

Most weak emulations solve the visible symptom—put some texture in the image—without rebuilding the relationships behind it. That creates a few recognizable failure modes.

A five-minute test for any grain tool

1. 01Feed it a neutral ramp. The texture should change meaningfully from shadows to midtones to highlights.
2. 02Compare saturated red, green, and blue patches at the same brightness. See whether the texture changes with the colour underneath it.
3. 03Pause on a frame, then advance one frame at a time. The whole pattern should renew without revealing a loop or slide.
4. 04If the tool models a projected print, change the camera format. The negative grain should change size while the print grain keeps its own scale.
5. 05Toggle grain off. Fine detail should not reappear merely because a blur was secretly tied to the control.

The checklist does not demand one particular technique. A scan of real film can provide useful source material. A generated pattern can be useful too. A scan is still only a source, and an algorithm is still only a method. What matters is whether the whole system reproduces the relationships above.

To do that, a model has to know more than where to place bright and dark specks. It needs to know which film material is forming the image, how that material responds to the subject, what happens when the negative is printed, how large the frame is, and when a new frame begins.

How Latent builds that chain

Latent is Ohqay's film-process model for DaVinci Resolve. Grain is formed inside the same negative-to-print process that creates the image's colour and tone. It is not added after the image is finished.

The negative begins with grain measurements for the selected stock, or a clearly named related-stock estimate when direct data do not exist. Its three colour records vary independently. Those variations travel through the print, which reshapes them and adds a second, unrelated grain pattern of its own. The physical scale follows the stock and camera format, and every timeline frame receives a newly generated pattern.

That structure is why the v0.61.40 chart changes texture across both colour and tone. There is no instruction that says “make green grain yellow.” The colour appears because a green patch changes the film's three records differently from a red or blue patch, and the print carries those differences into the viewed image.

What is measured, and what is estimated

Several negative stocks have directly published granularity measurements. Others use a documented estimate from a related film family. Modern print-stock measurements are less complete, so the absolute amount of print grain is a reasoned calibration, not a claimed laboratory measurement. Sources establish which colour layer is coarser; the exact size differences between layers are bounded estimates.

Those limits matter. A physically structured model can still contain estimates. Naming them makes it possible to improve the model when better evidence appears instead of hiding judgment behind the appearance of precision.

The visual examples are original Resolve renders made with Latent. The process descriptions and measurement language draw on the following primary and technical sources.

• Kodak VISION3 500T Color Negative Film 5219/7219 technical information

  RMS granularity method and the variables that affect perceived grain.

• Kodak glossary of motion-picture terms

  Graininess, granularity, emulsion, development, and related definitions.

• Kodak guide to push and pull processing

  How exposure and processing affect granularity.

• Kodak film format choices

  Negative formats and their practical image characteristics.

• Realistic Film Grain Rendering, Newson, Faraj, Delon, and Galerne

  A stochastic-geometry approach to simulating photosensitive grains and density variation.

• Steve Yedlin, On Film Grain Emulation

  Why scanned plates versus algorithms is the wrong dividing line for judging fidelity.

This guide follows a direct projected negative-to-print path unless another path is named. A scan or a different film chain can distribute grain across tone differently. Last reviewed 2026-07-26.