I work on the manufacturing side of LED fill lights, so I hear the same complaint every season from brand owners and from their customers: "I bought a light, turned it all the way up, and my video still looks worse." The light is almost never broken. It is behaving exactly the way a small, cheap, battery-powered panel behaves, and the spec sheet never told the buyer which numbers actually matter.
Brightness Is Measured at the LED, Not at Your Face
The big number on the box is lumens, and lumens measure total light leaving the emitter. What your skin actually receives is illuminance, measured in lux, and the two are separated by distance and by optics. Move a panel from 30 cm to 60 cm and the light on your face does not halve, it drops to roughly a quarter, because illuminance falls off with the square of the distance. This is why a light that looks blinding on the desk produces a dim, flat result once it is on a tripod arm across the room.
It gets worse. The lumen figure on a budget panel is frequently the theoretical output of the LED dies at the die level, not the measured output of the finished lamp. Once the light passes through a diffuser, a fresnel layer or a soft panel cover, you lose another thirty to fifty per cent. A softbox does not add light, it converts light into a larger, gentler source, and it is paid for in brightness. So a lamp rated for a hard, bare output will always under-deliver once you make it pleasant to look at.
CRI and R9 Decide Whether Skin Looks Alive
Colour rendering index is the number that separates "bright enough" from "my face looks ill." CRI Ra is an average across eight pastel test colours, and the deep red sample, R9, is deliberately not part of that average. Human skin is full of blood close to the surface, so it is precisely the R9 region that decides whether a face looks warm and alive or grey and slightly green.
This is the single most common corner cut in cheap panels. A lamp can honestly claim Ra 80 while its R9 sits near zero, and the result is a complexion that no amount of colour grading fully repairs. Panels built with a proper red phosphor or a dedicated red emitter push R9 above ninety, and the difference is immediately visible on camera. If you are lighting anything where colour accuracy is sold rather than shown, such as apparel, cosmetics or food, ask for R9 specifically, not just Ra. Broadcasters also lean on TLCI, the television lighting consistency index, because it weights the way cameras see colour rather than the way eyes do.
One more manufacturing detail matters to brand owners: LED dies are binned by colour and flux. Tight binning costs money. Loose binning means two lamps from the same production run sit at slightly different colour temperatures, which shows up the moment a customer runs a two-light setup and one side of the face is warmer than the other.
The Colour Temperature on the Box Is an Estimate
A panel marked 5600 K is usually somewhere within three to five hundred kelvin of that number, and that is considered acceptable in the industry. Two lamps from different batches can therefore disagree visibly, and mixing them in one frame produces a colour cast that is difficult to correct without degrading skin tones.
Bi-colour lamps create their adjustable range by mixing a warm white and a cool white emitter array. That has a consequence nobody warns buyers about: at either end of the range one array runs at full power, but in the middle both arrays run at roughly half. Total output is therefore lowest in the middle of the adjustment range, exactly where most people set the dial. If you want a bi-colour panel that holds its brightness across the whole range, it has to be over-specified at the ends, which costs LED count and money. Cheaper designs also mix imperfectly, so the beam shows a warm half and a cool half, or casts two-tone shadows behind the subject.
PWM Dimming Fights Your Shutter
Most inexpensive lamps dim by pulse-width modulation, switching the emitters on and off faster than the eye can see. A camera sensor does not see it that way. It reads line by line, so when the switching frequency and the shutter scan rate are not harmonically related, you get horizontal banding that crawls up the frame. It is much worse at high frame rates: a 120 fps slow-motion clip is far more likely to catch the dark part of the cycle.
The clean answer is constant-current dimming, where the drive current is reduced instead of the duty cycle. The trade-off is that at very low current the colour point of an LED drifts, so a lamp dimmed to ten per cent on DC can shift noticeably in tint. Good designs push the PWM frequency high enough to be invisible to common shutter speeds, or use DC dimming with colour compensation. Cheap designs do neither, and the buyer discovers it the first time they shoot slow motion.
Soft Light Needs Area, Not Power
Hardness is not about brightness, it is about the apparent size of the source relative to the subject. A small bright panel close to the face is still a hard source. It produces specular hotspots on the forehead and nose, sharp shadows under the chin, and a reflection in every pair of glasses. Making light soft means making it physically larger, which almost always means diffusion, which means losing output and needing more power to compensate.
This is also the honest critique of the ring light. Because it sits almost on the lens axis, it lights the face from every side at once and removes the modelling shadows that give a face shape. Skin texture flattens, blemishes actually become less visible, and the face reads as a flat disc. The catchlight in the eye is a ring, which is the giveaway. A ring light is a good tool for even, forgiving coverage and a poor tool for a dimensional portrait.
Placement Beats Wattage
The most common physical mistake is putting the lamp on the lens axis. Magnetic or clip-on lights that grip the phone sit within a few centimetres of the lens, which guarantees flat frontal light, no modelling, and a higher chance of red-eye. Off-axis placement at roughly thirty to forty-five degrees, slightly above eye level, produces far better results from the same number of watts.
The second mistake is using a small light to fight a window. Daylight through glass is orders of magnitude brighter than a battery panel. Balancing a backlit subject against a bright window needs hundreds of lux at the subject, which is not what a pocket panel delivers. The fix is rarely more lamp; it is closing a curtain, changing the camera angle so the window is side light instead of back light, or moving the subject.
Batteries Change the Story
Built-in lithium cells are where the printed specification and the real specification diverge most. Peak output is easy to quote and impossible to sustain. A panel that pulls hard from a small cell heats the cell, the driver derates to protect it, and within a few minutes the light is running below where it started. Cells also lose capacity with heat and with charge cycles, so a lamp that was adequate on day one is visibly weaker a year later.
Ask for sustained output after ten minutes rather than peak output at switch-on. Ask whether the lamp can run at full power while connected to a charger without cooking the cell. And check that the quoted runtime is measured at the brightness you will actually use, not at the lowest setting.
What to Ask For
Ask for measured lux at a stated distance, not lumens. Ask for CRI Ra and R9 separately, and for TLCI if the lamp is for video. Ask for the tolerance on colour temperature across a batch, and whether a two-light kit is colour matched. Ask whether dimming is PWM or constant current, and test it at your highest frame rate. Ask for sustained output after ten minutes at full power, not peak. Size the panel for the distance you will actually shoot at, and buy more area rather than more watts if you want soft light. Then place it off axis and slightly high, and let it sit one to two stops below the ambient rather than trying to overpower the room.
None of this is exotic. It is simply the set of numbers that determine whether a fill light flatters a face or makes it look worse, and none of them appear on the front of a retail box.
ODMKit develops LED fill lights and camera accessories for brand owners, on OEM and ODM programmes from Shenzhen. Full product range and manufacturing capability.
Top comments (0)