What Is Film Grain, and Why Does It Look Real?

Open any photo-editing app and you'll find a "grain" slider, usually sitting near contrast and saturation. It's a strange thing to add on purpose — grain is, technically, noise, an imperfection. So why do so many photographers and designers add it back in deliberately?

Where grain actually comes from

In film photography, grain is a physical property of the medium itself. Photographic film is coated with a light-sensitive emulsion made of silver halide crystals — the "grain" is literally the visible structure of those crystals once the film is developed. Faster film (higher ISO, meant for low light) uses larger crystals that are more light-sensitive but produce a more visibly grainy image. Slower film uses finer crystals for a smoother result but needs more light to expose properly.

Digital cameras don't have film, but they have an equivalent: sensor noise. When a digital sensor is pushed to a higher ISO to compensate for low light, it amplifies the electrical signal from each pixel — and that amplification introduces random variation, visually similar to film grain, usually called "noise" in a digital context.

Why grain reads as "real" to the eye

Grain and noise are technically defects, but they carry information that a human eye has learned to associate with genuine photography: the presence of grain implies a real sensor or a real strip of film captured actual light, at a specific ISO, in specific conditions. A perfectly smooth, noise-free image — especially one with even lighting and no texture at all — reads as synthetic, precisely because that level of smoothness almost never occurs in an unprocessed photograph.

This is part of why heavily denoised photos, or images from certain AI generators with no texture in them at all, can look subtly "off" even when every individual detail is rendered correctly. The eye is picking up on the absence of an expected imperfection, not on any one specific flaw.

Not all grain looks the same

Realistic grain isn't just uniform static laid over an image — that's actually one of the tells of a bad grain effect. A few properties matter:

• Luminance dependence. Real sensor noise is more visible in shadows and midtones than in bright highlights, because the signal-to-noise ratio changes with exposure. Grain that's applied evenly across an entire image, regardless of how bright or dark each area is, tends to look artificial.
• Size and softness. Grain isn't pixel-sharp static; it has a slight softness to it, closer to the texture of a fine-grained material than a hard digital pattern.
• Color behavior. Film grain typically affects luminance more than color, producing what's often called "monochromatic grain" — a texture that reads as brightness variation rather than random color speckling.

Getting these three details right is the difference between grain that looks like a genuine camera artifact and grain that looks like an Instagram filter from 2013.

Why designers add it deliberately

Beyond nostalgia, grain serves a real functional purpose in design and photography:

• It masks banding in gradients and smooth color areas, where digital images can otherwise show visible stepped bands of color
• It adds a sense of depth and texture that flat, digitally smooth images can lack
• It's a stylistic signal — a certain grain intensity and grain size is strongly associated with specific eras and formats (35mm film, high-ISO low-light photography, vintage printing)

Applying it to AI-generated or overly clean images

AI image generators, and heavily processed digital photos more broadly, often produce results that are unnaturally smooth — every surface rendered with even, artifact-free detail that no real camera sensor produces at typical settings. Adding realistic, luminance-aware grain back into these images is one of the most effective ways to make them read as authentic photography rather than a synthetic render.

That's the gap our Photo Humanizer is built to close. To be precise about how: it builds a softened monochromatic noise field and composites it with an overlay blend, which scales the effect by the brightness of the pixel underneath — so the grain lands hardest in the midtones and is compressed in both deep shadows and bright highlights. That is the film-grain distribution rather than the sensor-noise one, which is a deliberate choice explained in the next section. A light sharpening pass and a JPEG encode follow, so the texture goes through something resembling a real capture pipeline instead of sitting on top untouched.

For more on why AI images tend to need this treatment, see why AI-generated images still look "off".

Midtones or shadows? The distinction that matters

The section above says grain should be strongest in the midtones, and you will also see the advice that noise is strongest in the shadows. Both are correct, because they describe two different physical mechanisms that get lumped together under one word.

Film grain peaks in the midtones. Film is an emulsion of silver halide crystals, and each one either develops or it does not. In a blown highlight nearly every crystal develops, so the field is uniformly dense and grain is suppressed. In deep shadow almost none develop, so it is uniformly sparse. Visible texture peaks where roughly half developed — the midtones.

Digital sensor noise peaks in the shadows. A sensor counts photons, and photon arrival is random. The statistical spread scales with the square root of the count, so a bright area collecting 10,000 photons has about 1% relative noise while a dark area collecting 100 has about 10%. Less light, proportionally more noise.

They are opposites, and choosing between them is really choosing which look you are after. A film emulation should weight the midtones. A high-ISO digital emulation should weight the shadows and add a separate, coarser, coloured layer for chroma noise. Applying one recipe while describing the other is how grain ends up looking approximately right and specifically wrong.

Our tool models the film case. If you need physically accurate sensor noise for a particular camera, a raw processor with a camera-matched noise profile will do better. The full mechanism is in how camera sensor noise actually works.

Grain by format and stock

Grain is not one texture. It varies with film size, speed, and chemistry, which is why "add film grain" is an underspecified instruction.

Format sets the scale. Grain size is fixed by the emulsion, so what changes with format is how much you enlarge it. A 35mm negative blown up to A3 shows coarse grain; a medium format negative at the same print size shows much finer grain from the same stock, because it was enlarged less. Large format sheet film can look nearly grainless. If you are emulating a look, the format matters more than the stock.

Speed sets the coarseness. Faster films use larger crystals to catch more light, so ISO 3200 stock is dramatically grainier than ISO 100. This is the film analogue of the digital ISO relationship, though the cause is different — bigger crystals rather than amplified noise.

Colour behaves differently from black and white. Colour negative film has three emulsion layers of differing sensitivity, so grain differs by channel and is usually coarsest in blue. Black and white film has one layer, giving the cleaner monochromatic texture most people picture when they think of film grain.

Process affects it. Push-processing — underexposing and compensating in development — increases grain noticeably, which is why pushed low-light film has its particular gritty character.

How much is too much

The most common mistake is overdoing it, and there is a simple test: if you notice the grain, it is too strong. The target is texture you would only miss if it were removed.

A few practical constraints:

Match the scale to the output size. Grain is a property of the final image, not the file. Add it at output resolution — grain applied before a large downscale disappears, and grain applied before an upscale turns into blobs.

Match it to the subject. Grain reads as authentic on reportage, portraits, and low-light scenes because those were historically shot on grainy stock. On a bright, clean product shot it reads as an effect, because no photographer would have chosen fast film for that.

Keep it consistent within a set. Varying grain across images in one series breaks the illusion faster than getting the amount slightly wrong, because it signals post-processing rather than capture.

Do not stack it on an already-noisy image. Adding grain to a high-ISO photo that already has sensor noise produces a muddy double texture. Denoise first, or leave it alone.

When not to add grain

Grain is a texture choice, not a repair. It will not rescue a poorly composed image, and it will not make an implausible one plausible — grain over an image with impossible lighting or six-fingered hands is a grainy impossible image. Texture is the last thing the eye checks, after geometry, lighting, and anatomy.

There are also contexts where adding it is not a stylistic question at all. Documentary photography, photojournalism, evidence, insurance claims, and scientific imaging all have norms about what post-processing is permissible, and texture that implies a capture condition that did not occur can cross that line. Our terms rule out that use, and the reasoning is in content authenticity in the AI era.