Low persistence trades shorter illumination for less smear

What it is

A pixel need not stay lit for the whole frame. OLED headsets often flash for a millisecond or two at the end of the frame and go dark — low persistence: short illumination, short smear. Desktop LCD hold-type is the opposite: lit for the whole frame, so a turning head smears the pixel across time. The trade sits in duty cycle. At a given refresh, the smaller the fraction spent on, the less smear and the fewer photons you integrate.

It is how a frame is lit after refresh rate is already chosen. It does not fix the flicker-fusion point, and it does not take brightness loss as its main plot.

Why it happens

Persistence is how long a pixel is on inside one frame. The head is still moving in that window, so a lit pixel is not a point on the retina. Cut on-time from 11 ms to 2 ms and the painted length shortens roughly in proportion. That is why OLED VR would rather flash than stay on: a hold during a turn stretches high-contrast edges into ribbons.

The cost lands immediately in the photon budget. Same peak current, on-time a fraction as long, integrated nits fall. To get them back: raise peak (lifetime, heat, clipping) or drop refresh to lengthen the on-window — which hands back the persistence just shortened. LCDs add pixel response: even a short backlight pulse cannot finish if the crystal has not arrived, so low persistence is harder on LCD than on OLED.

Low persistence is therefore a smear switch, not free image quality. Duty cycle is “how long it is on”; refresh is “how often it comes on”. They are two knobs.

Studying it

High-speed film of a high-contrast edge, refresh and head speed fixed, sweep duty cycle, measure the angular width the edge sweeps in one frame. Subjectively, rate smear, and ask about flicker separately so a flash from too-short persistence is not scored as “sharper”.

Independent variables: duty cycle (persistence time), refresh held fixed, head speed, panel type (OLED / LCD). Dependent variables: angular smear width, smear ratings, flicker detection.

Include a hold-type, fully-on frame as baseline. Without it, low-persistence gains are eaten by refresh-rate gains in the write-up.

Where it stops holding

Head still, content still: persistence length is almost invisible; the gain lives in motion. Pulse too short at a refresh that is not high enough, and the periphery flashes first — people will trade “sharper” back for “not flashing”. If LCD response is on the same order as persistence, shortening the backlight only darkens; the smear stays. A video-see-through camera that samples hold-type has already painted motion blur that a low-persistence display cannot wash out.

Applying it

• Default moving content to low persistence; do not run head-turn scenes on a desktop-style full-frame hold.
• Keep duty cycle and refresh inspectable apart. When persistence changes, hold refresh still, or you will not know which knob cut the smear.
• If persistence is already short enough for peripheral flicker to be reported, raise refresh before cutting the pulse again.
• How to check: hold 90 Hz, change only on-time, turn the head on a high-contrast vertical line. Ribbon to hairline means low persistence is working; hairline plus peripheral flicker means duty cycle has crossed the line — do not go shorter.

Related

• Same group: N1.12.1 Refresh rate jointly determines flicker visibility and motion continuity · N1.12.3 When the eye pursues a moving object, smear is paint on the retina · N1.12.4 Raising refresh rate lowers available brightness
• Nearby: N1.05 Motion-to-Photon Latency · N1.11 Render Latency and Motion-to-Photon Time
• Search terms: low persistence · duty cycle · hold-type blur