We Measured Professional Listing Photos: What Makes Them Look Pro

The findings at a glance

These are the markers that separated published professional interiors from phone snapshots in our measurements. The sample size differs per marker, because not every photo can be read for every measure; the method section below explains each one.

What we measured, and how

The reference set. In September 2026 we collected 392 professional interior photos published on listings by Danish real estate brokerages, and measured them with our own scripts. We do not name the brokerages or republish their images; the numbers on this page are ours. Not every photo can be read for every marker, so each finding states its own sample: 378 of the 392 could be read for brightness, 93 living rooms for the near-white share, 200 for wall colour, and 80 interiors carried original EXIF data we could verify (matched to the original file, so no exterior shots slipped in). An earlier band of 74 interiors from 56 properties (28 living rooms, 17 kitchens, 12 dining rooms, 10 halls, 7 bedrooms) gave the colour and detail ranges.

The tools. Every photo was resized to 3000 px on the long edge with Lanczos resampling before measuring, so that differences in delivery size don't pose as differences in the photography. Brightness is CIELAB L*, where 0 is black and 100 is white; we read its percentiles (p1 is the level the darkest 1% of pixels sit below, p99 the level only the brightest 1% exceed). Camera lean comes from straight-line detection: a Canny edge detector followed by a probabilistic Hough transform finds the near-vertical lines, and the verdict is the length-weighted median of their signed deviation from vertical. Focal length comes from EXIF, converted to full-frame equivalent. Detail is the variance of the Laplacian, a standard measure of fine edges.

The phone side. Comparisons with phones come from October's own test captures, made with an iPhone 13, and from our own processing chain while we built it. That is a small set, and we say so in the limits section.

Finding 1: the verticals are straight, every time

Across the 392 professional photos, the median lean of the vertical lines was 0.00°, and it stayed 0.00° in every way we re-read them: re-encoded, resized by a pixel, cropped. Only 7 of the 392 (1.79%) leaned more than 1.0°. In 43 photos (11%) there were too few architectural lines to measure at all, usually because furniture filled the frame.

Two details matter for anyone repeating this. First, the measure has a resolution floor: professional photos we deliberately rotated by 0.25–0.60° still read 0.00°, so the true lean of these photos is somewhere below about 0.4°, not literally zero. Second, the sign matters. A sloping attic ceiling or the edge of a wide-angle frame leans lines in opposite directions, while a leaning camera leans them all the same way, so only the signed median measures the camera.

With a phone: in our first hand-held test captures, people held the phone anywhere from 0° to 8° off level, and the frame we delivered from them leaned 2.90° against the professionals' 0.00°. The other habit is pointing down: in our early captures 71.3% of the frame lay below the horizon line, against 50.4% in the professional set. Hold the phone upright, keep the horizon near the middle of the frame, and if the ceiling doesn't fit, step back instead of tilting up. Converging verticals are one of the effects that make an interior look wrong to viewers (Photography for Real Estate).

Finding 2: wide, but not as wide as the 0.5× lens

The 80 professional interiors with verifiable EXIF had a median focal length of 18 mm in full-frame terms; 10% were 15 mm or wider and 10% were longer than 25 mm. 89% were shot at 24 mm or wider, and only 8% were narrower than 26 mm, which is the 1× camera on the iPhone we tested with.

The iPhone's 0.5× ultra wide has a 120° field of view (Apple), about 13 mm in full-frame terms. That is wider than roughly 9 in 10 of the professional photos. Professionals stay away from that extreme for a reason: below about 22 mm, objects near the frame edge stretch, and this perspective distortion cannot be corrected afterwards the way lens curvature can (Photography for Real Estate).

With a phone: neither lens matches the professional view on its own. Shoot at 1× from a corner or doorway, and for a wider view take two or three overlapping 1× frames by turning the phone, not your body, and stitch them. Use 0.5× only when a room is too small for anything else, and keep furniture away from the edges of the frame.

Finding 3: bright is not the same as pale

Professional interiors are bright, but they get there by filling the shadows, not by pushing everything up. Measured on 378 photos, the brightness profile looks like this:

In the 93 living rooms, a median of 11.2% of the frame sat above L* 85 and 3.8% above L* 90, and the walls sat only 5.5 L* above the frame's median. The top half of that profile barely changes between room types (p90 between 83.9 and 87.5 across kitchens, living rooms, dining rooms, halls, empty rooms, bedrooms, bathrooms and utility rooms). The bottom half changes a lot: in bedrooms the darkest 10% of the frame sits below L* 20.2, in empty rooms below 53.9, because an empty room has nothing dark in it.

We learned this the hard way. An early version of our own processing matched the professional median almost exactly (L* 71.0 against 71.8) and still looked pale: 33.4% of the frame was above L* 85, a level only 3 of the 93 professional living rooms reach. Brightening the whole frame slides the histogram up without changing its shape, so the walls and the window run into white while the corners stay dark. Lifting the shadows towards the professional profile instead brought the near-white share down to 8.6% at the same overall brightness.

With a phone: add light where the room is dark, not everywhere. Open the curtains, switch the room's lights on, and in an editing app lift the shadows rather than the overall exposure. If the walls are already close to white, a brighter photo will only be a paler one.

Finding 4: neutral walls, restrained colour

Measured on 200 professional photos, the walls came out at a* 0 and b* +3 in CIELAB: neutral on the green–red axis and a touch warm on the blue–yellow axis (the middle 80% between b* +1 and +6). In the 74-photo band, the middle half of the photos had a median colour saturation between 0.03 and 0.09 on a 0–1 scale. Nothing is punchy.

Phones work against this in two ways. Mixed light is the first: a warm bulb in a daylit room tints everything near it. The second is that the phone re-measures colour and exposure for every frame. In one of our captures, the same white wall in two frames taken seconds apart differed by a colour difference (ΔE) of 21.5, with a lightness of 56.2 in one and 47.1 in the other. Inside a single professional photo, the same wall varied by only 0.25–1.06.

As captured

Delivered by October

With a phone: switch off warm lamps that aren't a feature of the room, or make sure every light in the frame is the same colour. Press and hold on the screen to lock exposure and focus before a series of frames, so every frame of the room is measured the same way.

Finding 5: the phone is sharp enough

This surprised us most. On the same detail measure and at the same size, the iPhone's own unprocessed frame scored 889, against a median of 510 for the professional photos (670 for professional living rooms). The phone's JPEG compression cost at most 3.8% of the finest detail. In other words, the camera captures more fine detail than the reference photos contain.

Where detail gets lost is processing. Heavy noise reduction and one-click enhancement smooth fabric and wood into plastic: one third-party enhancement step we tested removed 31.7% of the texture in a single pass. Absolute sharpness is also a poor yardstick on its own: in the 74-photo band it ranged from 46 to 1919, because a bookshelf scores far higher than a plain wall.

With a phone: you don't need RAW or extra megapixels for a listing. Shoot in good light, hold still, and go easy on smoothing and enhance filters.

Finding 6: one in six professional photos has a styling fault

We also labelled every one of the 392 photos for whether the room was ready: 323 were, 69 were not. That is about one in six published professional photos with an unmade bed, an open wardrobe full of clothes or toys across the floor. The bar for a good listing photo is "tidy", not "perfect".

Our own early test captures did far worse: only 7.4% of 135 frames were ready. Of the 125 that weren't, two causes accounted for 70%: an unmade bed (57) and open storage such as a wardrobe or cupboard (30), followed by clothes on furniture (13), clutter on the floor (11) and crowded surfaces (7). Most of these are ten-second jobs before the shutter.

What simple rules got wrong

Before we measured the professionals properly, we built a quality check from commonly quoted thresholds. Run unchanged over the 392 professional photos, it rejected 163 of them (41.6%). A rule against "milky" blacks rejected 114 photos that were simply bright, empty white rooms; a saturation ceiling of 0.28 rejected 11 warm rooms. Rebuilt on the 392, the check now uses five tests (blur, camera lean, noise, burnt-out highlights, a frame that is too dark) and rejects 9 of the 392 (2.30%).

The lesson for anyone grading photos, human or software: a professional photo is not a set of slider values. Straight lines, a sensible lens and the shape of the light carry it; colour and contrast vary widely between good photos.

Limits of this study

• One market. The reference set is Danish. As a check we measured 120 photos from US agent listings: they shared the same centre on all twelve markers we compared, with a wider spread, and 19% fell outside the Danish brightness band.
• Small phone sample. Phone figures come from October's own test captures (iPhone 13) and our processing chain during development, not from a broad survey of phone photos.
• What we could not measure. Camera height could not be measured reliably: our instrument returned a value on 0 of 605 photos. Lean below about 0.4° reads as 0.00°.
• Different samples per marker. Each figure states its own sample; they are not all from the same photos.
• Point in time. Measured in September 2026 on photos as published on the web.

How October uses these numbers

The October app is built around the first three findings: it keeps the phone level with an on-screen level, has you turn only the phone so the frames can be stitched into one wide photo, and takes the shots itself. From that capture an AI image model draws each finished photo, aiming the light at the professional profile above; the model is told to keep walls, floors, windows, fittings and the view as captured, and each photo is checked against the capture and redrawn if the room changed. The photos are AI-generated, so check each one and label it as AI-altered where your MLS requires. Photos start at $49 a place (up to 6 rooms), $99 with a walkthrough video, and your first place is free. See pricing or the real estate page.

Questions