Projector vs TV: the room decides it, not the spec sheet

Almost nothing on a projector spec sheet survives contact with an ordinary living room. The lumen rating is a brightest-mode bench figure, the contrast number is measured with the lights off, and the HDR badge describes a signal the machine cannot physically produce. The specification that actually decides the purchase is the dullest one on the page, the throw ratio, because it is the only one that tells you whether the projector fits the room you have. Measure the room first, then argue about panels.

The lumen number is three different numbers

Epson's Pro Cinema LS12000 sheet (CPD-61060, dated 1/22) prints 2,700 lumens twice, once as White Light Output and once as Color Light Output, and footnotes both: "Color light output measured in accordance with IDMS 15.4; white light output measured in accordance with ISO 21118." Two documents, two measurements. The white figure is a full white field read at nine points and averaged. The colour figure measures red, green and blue separately and adds them. A three-chip engine returns the same answer twice; a single-chip design with a clear segment in its colour wheel does not, which is why BenQ's TK700 page publishes 3200 ANSI lumens and no colour figure at all. Both numbers are taken in the projector's brightest preset, which Epson states plainly for colour light output: measured "in brightest mode".

Then come the figures that are not measurements. BenQ, which sells both kinds, writes that the Helmholtz-Kohlrausch effect "can result in a 1.3 to 2.4 time increase in brightness over a projector's measured ANSI brightness", and that light source brightness "measures its brightness directly from the light source itself before it has been converted to an image", before the panel, the wheel and the lens have taken their cut. XGIMI, which sells LED projectors, is blunter: "There is no standardized conversion ratio to turn LED lumens into ANSI lumens."

What those lumens become on your wall

Screen luminance is arithmetic: lumens divided by screen area in square feet, times screen gain, gives foot-lamberts. A 100 inch 16:9 screen is 87.2 by 49.0 inches, so 29.7 square feet. The LS12000's 2,700 lumens on a unity-gain white screen is 91 fL. SMPTE RP 431-2:2011, approved 6 April 2011, sets digital cinema reference white at 48 cd/m2 and equates that to 14.00 fL, which makes the conversion 3.43 and puts a 100 inch home screen at roughly six and a half times cinema white.

Stretch the same projector across the 300 inches its own sheet allows and the picture lands below a cinema.

Contrast is a property of the room

The LS12000 claims "Up to and over 2,500,000:1, Auto Iris on". LG's PH30N claims 100,000:1. SMPTE's reference cinema projector, the device the films were graded on, is asked for 2000:1 sequential contrast in a darkened room, and for far less on real picture content: "The nominal (Reference Projector) value for intra-frame contrast shall be minimum contrast of 150:1."

The reason the reference device looks so poor on paper is that RP 431-2 measures it honestly, in the room, with the doors shut. "It shall not be appropriate to subtract the room black (ambient light) when computing sequential contrast." The same document notes that checkerboard contrast is "reduced by many factors including projection lens flare, port glass flare, ambient light spilling on the screen and back-reflections from the room itself". Even a compliant cinema is allowed some stray light: for exhibition theatres "the ambient light level reflected by the screen should be less than 0.03" cd/m2, and against 48 cd/m2 of reference white that allowance on its own caps sequential contrast at 1,600:1 before the projector has contributed anything.

Your room is not a cinema. Take the reference viewing environment for HDTV, cited in Annex B of SMPTE ST 2084:2014 as "room illumination of 10 Lux". Ten lux arriving at that 100 inch unity-gain screen is 0.93 lumens per square foot, so 0.93 fL of black floor, and against 91 fL of peak white that is about 100:1. In a room lit to a broadcast reference standard. Switch on a table lamp and it drops again. The 2,500,000:1 on the box is a dynamic-iris figure with the lights off, and nothing you can buy converts it into contrast on a screen with light falling on it.

AVIXA put the rule into a standard. ANSI/AVIXA V201.01:2021 covers "permanently installed and temporary display systems and all displayed images regardless of AV technology", and it fixes where the meter goes: "System contrast ratio measurements are taken in the system's typical use case".

The screen is part of the projector

Ambient light rejecting screens work, within a geometry they choose for you. Elite Screens' CLR 3 material has 0.8 gain and a 170 degree viewing angle, "absorbs 90% of overhead lighting", and is marked "Not Compatible With Short or Standard Throw Projectors". Its CineGrey 4D material has 1.1 gain, a 70 degree viewing angle, "can reject up to 63% of off-axis lighting", and is "Not Compatible With Ultra-short/Short Throw or Triple Laser Projectors".

Read those two together. Each screen rejects light arriving from the direction the other one needs, so the screen is not a later accessory decision, it is chosen at the same moment as the throw type. Gain above 1.0 is not free brightness either: CineGrey 4D buys its 1.1 by narrowing the useful viewing angle to 70 degrees.

Throw ratio, and whether the thing fits

Throw ratio is projection distance divided by image width. A 100 inch 16:9 picture is 2.21 m wide, so the ratio range on a spec sheet converts directly into a range of places the projector may stand.

Epson's own table agrees with the arithmetic: 50 inch to 300 inch pictures "at 58" to 62'4"". A long-throw model in a 4 m room can be ceiling mounted behind the seats and nowhere else. The LS12000 has powered lens shift of -96.3% to +96.3% vertically, which is what lets it sit off the screen axis; a projector without shift must be placed on that axis or use keystone correction, which reshapes the image inside the panel and so uses fewer of its pixels.

Lamp, laser and LED, and what "life" means

Epson spells out what the laser number means: "Approximate time until brightness decreases 50% from first usage." A 2,700 lumen projector at hour 20,000 is a 1,350 lumen projector, which on that 100 inch screen is 45 fL rather than 91. Hisense quotes "25,000+ Hours" for the L9Q's triple laser at 5000 ANSI lumens. Nothing here fails on a given date; it fades, continuously, and the spec sheet tells you when it will have faded by half.

HDR is where the physics stops

SMPTE ST 2084:2014, approved 16 August 2014, defines the PQ curve and "assumes a peak luminance level limited to 10,000" cd/m2. Grading works far below that ceiling but far above a projector. Dolby's Color Grading Best Practices Guide version 4.0, dated 16 December 2019, has colorists working on 1000 nit and 4000 nit mastering displays and adding an optional trim "at 600-nits to check the mapping at the average peak luminance level of current consumer HDR televisions". VESA's DisplayHDR CTS 1.2, released 7 May 2024, certifies 1000 cd/m2 for DisplayHDR 1000, measured on an 8% centre patch.

That patch is the whole difference. A TV's peak is a small-window peak, driven far above what the panel could sustain across a full field. A projector has no such reserve, because its peak white is its full-field white: the 312 cd/m2 computed above is what a 1% highlight gets and what a 100% white frame gets. So every HDR title on a projector is tone mapped down, by roughly three against a 1000 nit master and by more than twelve against a 4000 nit one. Epson's answer is to hand you the dial: the LS12000 offers "16 steps of control" over the HDR curve. That is a manual decision dressed as a feature.

Input lag is the part projectors fixed

BenQ publishes "4.2 ms (1080P 240Hz), 8.3 ms (1080P 120Hz)" and "16.7 ms (4K 60Hz)" for the TK700. The 4K number is exactly one frame at 60 Hz. Lag on a modern projector is a property of the processing mode, not of the projection technology, and it is the one line of the comparison where the projector is not at a structural disadvantage.

The running costs nobody quotes

The LS12000 draws 311 W in Normal mode, 243 W in Medium and 204 W in Eco. A thousand hours of Normal is 311 kWh, which at the EU household average Eurostat reports for the second half of 2025, EUR 0.2896 per kWh, is about EUR 90. The lamp model is worse on both counts: the Home Cinema 1080 draws 345 W in Normal and 235 W in ECO, and its lamp is a consumable at USD 76 against a 6,000 hour Normal-mode rating.

Then the filter, which no comparison article mentions. Epson's LS12000 manual: "It is recommended that you clean the air filter every 5,000 hours. Clean it more often if the projector is in a dusty environment." Ceiling mounted, that is a ladder every couple of years, plus an ELPAF62 filter when tapping the dust out stops working. Add the screen, which for any room with windows is an ALR panel and a second purchase.

When the projector is the better answer

Size, mostly. Samsung's 115 inch Micro RGB (MRN115MR95FXZA) lists at USD 29,999.99, and the LS12000 covers 50 inch to 300 inch. Above roughly 100 inches the projector stops competing with televisions and starts being the only affordable option, which is a real and sufficient reason to buy one.

The second is a room you control. If the room can be brought near the 10 lux of a broadcast reference environment, 91 fL on a 100 inch screen is a genuinely cinema-grade SDR picture at six times the luminance a film was graded at, and the contrast arithmetic stops working against you. Film and television graded for a dark room is exactly the content that survives the trip.

The wrong answer is a bright room with windows you cannot cover, a display that runs all day and burns rated hours continuously, or a purchase justified by HDR. Screen size and seating distance give you the foot-lamberts, the throw ratio tells you where the box goes, and the light you cannot switch off sets the contrast ceiling, all three before a model number is chosen.