Noise in dark interiors, and how far to clean it

Noise is the speckled, grainy texture that appears in dark areas of a photograph. In property work it turns up in basements, north-facing rooms, evening shoots and the shadow areas of any bracketed set, and the temptation is always to clean it more aggressively than the image can bear.

Where it comes from

Two main sources. Raising ISO amplifies the sensor's signal along with its noise, so higher ISO means more visible grain. And lifting shadows in editing amplifies whatever was recorded in those dark areas, which was very little, so the noise that was always there becomes visible.

The second cause matters more in property photography than photographers expect. Even a low ISO frame develops visible noise in the corners of a room once the shadows are lifted to show detail, which is exactly what a listing photo requires.

Small sensors produce more of it, which is why drone frames and phone photos show noise sooner than a full-frame camera. That case is covered in drone photo editing for property.

Why it matters commercially

Noise reads as low quality even to viewers with no photographic knowledge. A noisy image looks like a phone snap taken in a hurry, and that impression transfers to the property and the agent.

It also compounds. Portals compress uploaded images, and compression handles noise badly, so a slightly noisy file can arrive on a listing page looking considerably worse than it did on your screen.

The over-correction problem

Noise reduction works by smoothing variation between neighbouring pixels. The difficulty is that fine detail is also variation between neighbouring pixels, so the process cannot distinguish grain from texture beyond a point.

Pushed too far, the result is the plastic look: walls that are perfectly smooth in a way no painted surface is, carpet that has lost its pile, wood grain reduced to a vague pattern, and foliage outside the window turned into green mush.

The plastic look is worse than moderate noise, because noise reads as a technical limitation while over-smoothing reads as manipulation. Buyers do not diagnose it, but they register that something is artificial.

How much is right

The working standard is that noise should not be the first thing you notice at normal viewing size, and texture should survive at high zoom. In practice this means:

• Treat areas differently. Flat surfaces such as walls, ceilings and skies tolerate more smoothing because they have little real texture to lose. Detailed areas such as carpet, wood, brick and foliage need much less.
• Preserve edges. Skirting boards, door frames and furniture outlines should stay crisp. Noise reduction that softens edges has gone too far.
• Watch colour noise separately. The coloured speckling in shadows can be removed quite aggressively without harming detail, unlike the luminance grain.
• Consider the output size. Images destined for portal display at moderate size tolerate more cleaning than files that will be viewed large.

Preventing it at capture

Everything above is easier if the file arrives cleaner.

A tripod is the main tool, because it allows long exposures at base ISO instead of high ISO handheld. A four second exposure at low ISO produces a dramatically cleaner file than a fast exposure at high ISO, and in an empty room nothing is moving.

Adding light helps too. A bounced flash lifts the shadows optically rather than digitally, so there is less to amplify later. Opening blinds and turning on lamps raises the ambient level, though it introduces the colour mixing discussed in fixing colour casts in interior photos.

And bracketing properly matters: including a genuinely bright frame gives the editor real shadow information rather than requiring them to lift a dark one. These habits are collected in shooting for your editor.

Sharpening interacts with it

Sharpening amplifies local contrast, which amplifies noise. Applying strong sharpening to a noisy file makes the grain crisper and more obvious, which is the opposite of the intent.

The order matters: noise reduction first, sharpening after, and sharpening sized for the output. Over-sharpened property photos develop halos along edges and a gritty texture across flat walls, and the combination of heavy noise reduction with heavy sharpening produces the worst of both, smooth surfaces with crunchy edges.

Noise differs between the frames of a bracket

A detail that matters when blending. The frames in a bracketed set have different noise characteristics, because the brightest frame used a much longer exposure than the darkest one.

When those frames are combined, the shadow areas typically come from the brightest exposure, which is the cleanest source available for them. This is one of the practical arguments for bracketing properly rather than lifting shadows from a single mid exposure: the bracket supplies genuinely clean shadow data instead of amplified noise.

It also means noise reduction is better applied after blending rather than to individual frames, since treating each frame separately can produce visible differences in texture between areas that came from different exposures. If you are blending your own work, that ordering is worth adopting.

When to accept it

Some frames are noisy and cannot be made clean without destroying them. A dark basement, an evening interior lit only by lamps, or a phone photo from an agent all have a ceiling.

In those cases the honest approach is moderate cleaning that keeps texture, accepting some grain rather than smoothing the image into plastic. A slightly grainy photograph that still looks like a room beats a clean one that looks like an illustration.

A studio worth using will tell you when a set's ceiling was set at capture rather than quietly delivering over-processed files. If you want that assessment on your own difficult frames, send a dark room through the contact page. Single-exposure work of this kind falls under image enhancement.