Video Best Practices
Picture performance for private cinemas
This document states how a private cinema picture should perform. Every figure is an outcome that you measure in the finished room, at the seats, with the room as built. None of it depends on a particular product.
It has two parts: projection, and fixed-pixel displays (LED walls, OLED and LCD panels). Each part is complete on its own, so some figures appear in both. SDR and HDR have separate targets throughout.
Each table gives a target and a limit. Design to the target. The limit is the worst result that is acceptable in the finished room.
Download the PDF
Email [email protected] and we will send it.
Part 1. Projection
A projector and a screen. The screen is part of the picture, so it has its own section.
1.1 Brightness
Brightness is the luminance of full white at the centre of the picture, measured from the reference seat. A dim picture loses shadow detail and colour. A bright SDR picture looks flat and tires the eye over a film. SDR cinema is graded at 48 cd/m². Home HDR is graded at 1,000 cd/m² or more.
| What is measured | Target | Limit |
|---|---|---|
| Full-screen white | 48–55 cd/m² (14–16 ftL) | 38–75 cd/m² (11–22 ftL) |
| TV-reference mode (optional, for broadcast content) | 100 cd/m² (29 ftL) | — |
| What is measured | Target | Limit |
|---|---|---|
| Full-screen white (peak) | 300 cd/m² (88 ftL) | At least 108 cd/m² (31.5 ftL) |
| Reference (diffuse) white | About 100 cd/m² (29 ftL) | — |
Dynamic tone mapping. HDR content is mastered brighter than the display can show. The display must tone-map it, and we recommend dynamic tone mapping only: mapping that adapts to the content scene by scene or frame by frame. A display that clips instead loses all detail above its peak.
An SDR picture is calibrated down to its target. Light above the target is headroom, not a fault.
1.2 Viewing geometry
The horizontal viewing angle is the angle the picture fills at a seat, measured across the full width of the widest format shown. It depends on the picture width and the seat distance together. The vertical figures set how far each viewer has to look up.
| What is measured | Target | Limit |
|---|---|---|
| Best seat | 40–50° | At least 36° |
| Back row | At least 36° | At least 30° |
| Front row | — | At most 60° |
| What is measured | Target | Limit |
|---|---|---|
| Angle from the eye to the picture centre, every seat | 0° to 10° above the eye | No more than 15° above or below |
| Angle up to the top edge, front row | 20° or less | 30° |
| Eye height on the picture, best seat | One third to one half of the way up the picture | Never below the bottom edge or above the centre |
Resolution. Above about 31° (HD) or 33° (2K), a viewer with normal sight sees the pixel grid. 4K allows about 62°. The angle targets stay the same; the resolution is the owner's choice.
1.3 Colour and greyscale
Colour is measured after calibration, at the reference seat, in the finished room. SDR has two tiers. The standard tier suits almost every system. The stretch tier is for the very highest end only: almost any system held to it needs a controlled warm-up, regular recalibration, and a spectroradiometer to verify it.
| What is measured | Target | Limit |
|---|---|---|
| White point | D65 | Within 0.003 Δu′v′ |
| Greyscale, 10–100 % | Neutral | Within 1.0 Δu*v* of white |
| Brightness curve (EOTF) | BT.1886 (gamma 2.4) | Exponent within ±0.10 |
| Primaries and test colours | Rec.709 | Within 7 Δu*v* |
| What is measured | Target | Limit |
|---|---|---|
| White point | D65 | Within 0.001 Δu′v′ |
| Greyscale, 10–100 % | Neutral | Within 0.5 Δu*v* of white |
| Primaries and test colours | Rec.709 | Within 4 Δu*v*; skin tones within 2.6 |
| What is measured | Target | Limit |
|---|---|---|
| White point | D65, within ±0.002 xy | ±0.006 xy |
| Gamut coverage | 99 % of DCI-P3 or more | At least 90 % |
| Colour error (ΔE ITP, mean) | 3 or less | — |
| Brightness curve below the tone-mapping point | PQ (ST 2084) | ±20 % below 0.02 cd/m², ±12 % to 1 cd/m², ±6 % above |
| Tone mapping above that point | Dynamic tone mapping, present and active | — |
| What is measured | Target | Limit |
|---|---|---|
| Brightness, sides against centre | SDR 80–90 %; HDR 85–100 % | At least 75 % |
| Colour at the corners | Within 0.008 xy | 0.015 xy |
A screen or panel with a slight colour tint is corrected by the calibration. Only the calibrated result is graded.
1.4 Contrast and black
Contrast is measured in the finished room. Full on/off contrast compares a full white picture with a full black one. Checkerboard contrast compares the white and black squares of one picture, so it includes light that the room returns to the screen.
| What is measured | Target | Limit |
|---|---|---|
| SDR full on/off | 2,000:1 or more | At least 1,200:1 |
| SDR checkerboard | 150:1 or more | At least 100:1 |
| HDR black level | 0.005 cd/m² or less | 0.05 cd/m² |
| HDR checkerboard | — | At least 150:1 |
| Room light on the screen, picture black (SDR / HDR) | 0.01 / 0.002 cd/m² or less | — |
| What is measured | Target | Limit |
|---|---|---|
| Front third of the side walls and ceiling | Dark, 10 % reflectance or less | — |
| The rest of the room | The designer's choice, with the contrast cost stated | — |
| Stray light seen from any seat (exit and step lights) | — | 3.4 cd/m² |
In a real room, projection rarely reaches the SDR checkerboard target.
The room does not have to be black. Light-coloured finishes reflect picture light back onto the screen, and that lowers checkerboard contrast. State this cost in the design, so that the owner can weigh picture performance against the look of the room.
1.5 Screens
A screen is assessed by measurements taken in the finished room. The design states the measured data of the specified screen.
| What is measured | Target | Limit |
|---|---|---|
| Brightness at the sides, best seat | 80–90 % of the centre | — |
| Brightness anywhere on the picture, every seat | — | At least 75 % of the centre |
| Light through an acoustically transparent screen | With a full white picture, nothing behind the screen is visible from any seat | — |
| Texture, holes, sparkle and moiré | None visible from the front row, on a full grey field, with the installed projector | — |
Gain is measured, not assumed. Measure gain against a calibrated diffuse white reference standard, with full-field red, green, blue and white, using a spectroradiometer, and state the method with the figure.
Light through the screen. A black backing layer on the screen is the usual way to meet this outcome. A black wall alone rarely does, because speakers, brackets and sockets behind the screen still catch the light.
Texture and holes. Perforated screens usually need a longer front-row distance than knitted or woven screens.
Ambient-light-rejecting screens. They help a room that cannot be made dark. In a dedicated cinema they often cost viewing angle, uniformity and texture. The outcomes above decide, as for any screen.
Part 2. Fixed-pixel displays
LED walls, OLED and LCD panels. The picture is made by the display itself, so contrast and HDR brightness reach far higher than projection, and the pixels, modules and seams need checks of their own.
2.1 Brightness
Brightness is the luminance of full white at the centre of the picture, measured from the reference seat. A dim picture loses shadow detail and colour. A bright SDR picture looks flat and tires the eye over a film. SDR cinema is graded at 48 cd/m². Home HDR is graded at 1,000 cd/m² or more.
| What is measured | Target | Limit |
|---|---|---|
| Full-screen white | 48–55 cd/m² (14–16 ftL) | 38–75 cd/m² (11–22 ftL) |
| TV-reference mode (optional, for broadcast content) | 100 cd/m² (29 ftL) | — |
| What is measured | Target | Limit |
|---|---|---|
| Full-screen white | 300 cd/m² | At least 200 cd/m²; at most 310 cd/m² for comfort |
| Peak white in a small area (10 % window) | 1,000 cd/m² | At least 600 cd/m²; at most 1,000 cd/m² |
| Reference (diffuse) white | About 100 cd/m² | — |
Dynamic tone mapping. HDR content is mastered brighter than the display can show. The display must tone-map it, and we recommend dynamic tone mapping only: mapping that adapts to the content scene by scene or frame by frame. A display that clips instead loses all detail above its peak.
The display must not add a dynamic brightness boost beyond the content as mastered.
An SDR picture is calibrated down to its target. Light above the target is headroom, not a fault.
2.2 Viewing geometry
The horizontal viewing angle is the angle the picture fills at a seat, measured across the full width of the widest format shown. It depends on the picture width and the seat distance together. The vertical figures set how far each viewer has to look up.
| What is measured | Target | Limit |
|---|---|---|
| Best seat | 40–50° | At least 36° |
| Back row | At least 36° | At least 30° |
| Front row | — | At most 60° |
| What is measured | Target | Limit |
|---|---|---|
| Angle from the eye to the picture centre, every seat | 0° to 10° above the eye | No more than 15° above or below |
| Angle up to the top edge, front row | 20° or less | 30° |
| Eye height on the picture, best seat | One third to one half of the way up the picture | Never below the bottom edge or above the centre |
Pixel pitch. To keep the pixel structure invisible from the front row, the pitch in millimetres should be at most the front-row distance in metres divided by 3.44. This is the 1 arcminute limit of normal sight.
Resolution. Above about 31° (HD) or 33° (2K), a viewer with normal sight sees the pixel grid. 4K allows about 62°. The angle targets stay the same; the resolution is the owner's choice.
2.3 Colour and greyscale
Colour is measured after calibration, at the reference seat, in the finished room. SDR has two tiers. The standard tier suits almost every system. The stretch tier is for the very highest end only: almost any system held to it needs a controlled warm-up, regular recalibration, and a spectroradiometer to verify it.
| What is measured | Target | Limit |
|---|---|---|
| White point | D65 | Within 0.003 Δu′v′ |
| Greyscale, 10–100 % | Neutral | Within 1.0 Δu*v* of white |
| Brightness curve (EOTF) | BT.1886 (gamma 2.4) | Exponent within ±0.10 |
| Primaries and test colours | Rec.709 | Within 7 Δu*v* |
| What is measured | Target | Limit |
|---|---|---|
| White point | D65 | Within 0.001 Δu′v′ |
| Greyscale, 10–100 % | Neutral | Within 0.5 Δu*v* of white |
| Primaries and test colours | Rec.709 | Within 4 Δu*v*; skin tones within 2.6 |
| What is measured | Target | Limit |
|---|---|---|
| White point | D65, within ±0.002 xy | ±0.006 xy |
| Gamut coverage | 99 % of DCI-P3 or more | At least 90 % |
| Colour error (ΔE ITP, mean) | 3 or less | — |
| Brightness curve below the tone-mapping point | PQ (ST 2084) | ±20 % below 0.02 cd/m², ±5 % to 1 cd/m², ±3 % above |
| Tone mapping above that point | Dynamic tone mapping, present and active | — |
| What is measured | Target | Limit |
|---|---|---|
| Brightness | SDR within ±5 % of the mean; HDR within ±9 cd/m² | — |
| Colour | Within 1.3 Δu*v* | 4 Δu*v* |
| Greyscale | Held to the colour tolerance down to 5 % stimulus | — |
A screen or panel with a slight colour tint is corrected by the calibration. Only the calibrated result is graded.
2.4 Contrast and black
Contrast is measured in the finished room. Full on/off contrast compares a full white picture with a full black one. Checkerboard contrast compares the white and black squares of one picture, so it includes light that the room returns to the screen.
| What is measured | Target | Limit |
|---|---|---|
| SDR full on/off | 4,800:1 or more | At least 2,000:1 |
| SDR checkerboard | 1,000:1 or more | At least 200:1 |
| HDR black level | 0.0005 cd/m² or less | 0.005 cd/m² |
| HDR checkerboard | 5,000:1 or more (10,000:1 stretch) | At least 1,000:1 |
| Room light on the screen, picture black (SDR / HDR) | 0.01 / 0.0005 cd/m² or less | — |
| What is measured | Target | Limit |
|---|---|---|
| Front third of the side walls and ceiling | Dark, 10 % reflectance or less | — |
| The rest of the room | The designer's choice, with the contrast cost stated | — |
| Stray light seen from any seat (exit and step lights) | — | 3.4 cd/m² |
Face reflectance. Measure and state the reflectance of the display face. A dark face returns far less room light than a projection screen, so lighter room finishes cost less contrast. The stated reflectance sets the checkerboard contrast that the room allows.
The room does not have to be black. Light-coloured finishes reflect picture light back onto the screen, and that lowers checkerboard contrast. State this cost in the design, so that the owner can weigh picture performance against the look of the room.
2.5 Pixels, seams and modules
This section covers what is true only of a display made of pixels, modules and cabinets.
| What is measured | Target | Limit |
|---|---|---|
| Pixel structure | Not visible from the front row, on a full grey field or fine detail | — |
| Scaling | No scaling artefact visible when the source does not match the display | — |
| What is measured | Target | Limit |
|---|---|---|
| Bright seams | None visible from any seat on full white, mid-grey and near-black, after seam adjustment | — |
| Dark seams | None visible from any seat on the same three fields, after seam adjustment | — |
| Seam brightness, measured | Within ±2 % of the screen average | — |
| Cabinet gap, position and flatness | Within 1/10 of the pixel pitch between adjacent cabinets | — |
| Adjacent modules | Brightness within ±1 % (±2 % at or below 0.1 cd/m²); colour within 0.001 Δu′v′ (0.0025 at or below 5 cd/m²) | — |
| Adjacent pixels | Brightness within ±4 %; colour within 0.025 Δu′v′ | — |
| What is measured | Target | Limit |
|---|---|---|
| Steps on one-code-per-step ramps, from the front row | None visible | — |
| Near-black greys | Neutral | — |
| Brightness at every seat, against straight on | Within 20 % | 25 % |
| White colour at every seat | Within 0.005 Δu′v′ of straight on | — |
| Contrast at every seat | No more than 20 % lower | 50 % lower |
| Flicker, scan lines and aliasing | None visible from any seat | — |
| Reflections | No light source seen reflected in the display from any seat | — |
Seams. Align the cabinets mechanically first. Use the processor's seam adjustment after that, and record it per seam. A seam that shows only from the side seats is a flatness fault; fix it mechanically. Check the seams again at handover, after any module or cabinet change, and at an interval that the design states.
Flat panels. Every outcome in this part applies. A large panel suits a room where the best seat is within about 3 m. Image retention and uneven wear from letterbox bars are costs to state.
How to measure
- Measure in the finished room, with the lights as they are used for viewing.
- Measure brightness, colour and contrast from the reference seat unless a row says otherwise. Check the visual outcomes from the seats each row names.
- Use a spectroradiometer, or a colorimeter profiled against one for the display in question. Colorimeters misread laser and LED light by more than the stretch-tier tolerances.