The Lowest Possible Input Lag Retro Setups for Fighting Games and Rhythm Games

A Super Turbo player in Hong Kong is two frames into a punish window on a 4K OLED and the jab whiffs. The opponent recovers. The round ends. The screen, not the player, lost it. The lowest-lag retro setup for fighting and rhythm games is a known, measurable chain: a system outputting native RGB or component into a scaler with sub-millisecond response, then into a modern panel running its fastest Game Mode. The physical constant is a frame at 60Hz, 16.67ms. A fighting-game punish window can be as tight as two frames. Rhythm titles like Dance Dance Revolution or Beatmania judge timing windows in milliseconds. Every device added to the chain, every buffer, every conversion, shaves off time the player cannot get back.

A Sony PVM-20L5 professional CRT monitor is the zero-lag benchmark. The electron beam draws the line the moment the signal arrives. No buffer. No scaling. No handshake. A consumer Trinitron is functionally identical for response-time purposes. Any digital scaler, even the best, adds time. The question is how much.

OSSC: The Microsecond Reference

The Open Source Scan Converter (OSSC) is the reference for line multiplication. It has no framebuffer. In Line2x through Line5x modes, the OSSC adds less than 1ms of delay, measured in microseconds per scanline. This is the theoretical minimum for a digital scaler. The catch is input compatibility: the OSSC accepts RGB SCART, component, and VGA, but not composite or S-Video. If your system outputs only composite, the OSSC cannot use it without an external transcoder, and that transcoder adds its own delay.

RetroTINK 5X-Pro: The Single-Device Solution

The RetroTINK 5X-Pro uses a framebuffer but keeps response under 1ms in Frame Lock and Gen Lock modes. A Time Sleuth test by RetroRGB in 2021 measured 0.25ms at 1080p output. The 5X accepts composite, S-Video, component, and RGB SCART, making it the single-device answer for the widest range of hardware. Triple Buffer mode, designed for tricky sync signals, adds one to two frames of lag (16 to 33ms at 60Hz). Never use Triple Buffer for competitive play.

RetroTINK 4K: The Endgame

The RetroTINK 4K is the ceiling. A 2024 Time Sleuth test measured 0.2ms at 4K output in Frame Lock mode. It adds black frame insertion at 120Hz, which eliminates sample-and-hold blur without adding delay. This unit is the only scaler that can output a 4K signal with sub-millisecond response. For a Hong Kong apartment gamer with one screen for modern and retro hardware, the 4K is the endgame.

Device	Mode	Latency	Test Year
Sony PVM-20L5 CRT	Any	0ms (analogue baseline)	N/A
OSSC	Line2x–Line5x	<1ms (microseconds per scanline)	2022
RetroTINK 5X-Pro	Frame Lock, 1080p	0.25ms	2021
RetroTINK 4K	Frame Lock, 4K	0.2ms	2024
RetroTINK 5X-Pro	Triple Buffer	16–33ms (1–2 frames at 60Hz)	2021
Framemeister XRGB-Mini	Game Mode	25ms (1.5 frames at 60Hz)	2013
Generic HDMI converter (Sham Shui Po)	240p treated as 480i	30–66ms (2–4 frames)	Community test

The choice between OSSC and RetroTINK 5X-Pro for fighting games comes down to which hardware you play. The OSSC line multiplication is purer on the response side, but it cannot process 480i without flicker. If you play Super Turbo on a Super Famicom in 240p, the OSSC is ideal. If you play a PS2 fighting game that runs in 480i, the OSSC's bob deinterlace flickers and the RetroTINK 5X's motion-adaptive deinterlacing is the better option. The 5X-Pro also handles composite and S-Video, so a N64 or a Famicom without an RGB mod is usable without an external transcoder.

What the Tournament Organisers Use

The Hong Kong Retro Game Association runs tournament setups using both scalers. Their verified reference chain is a system to RGB SCART to RetroTINK 5X-Pro in Frame Lock mode to a modern TV in Game Mode. They test every screen with a Leo Bodnar Lag Tester before a tournament. If the TV adds more than 16.67ms (one frame) at 1080p, they replace it. The 2024 RTINGS.com measured values for modern TV Game Mode lag range from 5ms to 20ms at 60Hz. A TV at the high end of that range plus a scaler at 0.25ms is still under one frame of total added lag. That is playable for competitive fighting games.

When the TV Is the Bottleneck

The failure case is a TV without a fast Game Mode. A Samsung or LG in Hong Kong with Game Mode on still adds 10 to 15ms at 1080p and 30 to 40ms at 4K. If the TV's 4K Game Mode is slow, run the scaler at 1080p output and let the TV do the final upscale. The TV's scaler adds its own lag, but it is usually lower than the penalty of the 4K processing path. Test it. If the game feels loose, drop the output resolution.

Rhythm games are less forgiving of variable response than fighting games. A fighting-game punish window is two to three frames. A rhythm game like Dance Dance Revolution judges the step timing window at 1.67 frames (28ms) for a Perfect on the original arcade hardware. Beatmania timing is tighter. Any fluctuation in delay from frame to frame is worse than a consistent, slightly higher lag. The scaler must deliver a fixed, repeatable delay.

The OSSC and RetroTINK 5X-Pro in Frame Lock mode both provide consistent sub-millisecond response. The RetroTINK 4K with 120Hz black frame insertion improves motion clarity for rhythm games on an OLED. An OLED's sample-and-hold technique causes motion blur at 60Hz that can obscure fast-moving notes or arrows. Black frame insertion at 120Hz clears that blur by inserting a black frame between each game frame, resetting the pixel response. The 4K's BFI mode adds no delay because it adjusts frame timing rather than buffering. For a rhythm game player, the 4K on a 120Hz OLED is the current best option.

The timing-window problem appears when the screen itself adds variable lag. A cheap converter in Sham Shui Po that treats 240p as 480i adds two to four frames of lag that fluctuates with the deinterlacer's motion estimation. A rhythm game that judges timing in milliseconds becomes unplayable. The fix: bypass the converter entirely and use a scaler that identifies 240p correctly. The 240p test suite is the standard tool for checking whether a scaler misdetects the signal. Run it before you buy.

Every device between the system and the screen adds measurable delay. The chain is the hardware, cable, scaler, panel. Remove everything else. A splitter for a capture card adds its own handshake and buffer time. A capture card adds at least one frame. The Hong Kong Retro Game Association tournament chain has no capture card in the play path. The capture card gets a separate split from the hardware, not from the scaler output, and the play screen runs direct.

Sync Drops and How to Stop Them

HDMI handshake re-sync delay is a real problem. A scaler that loses sync on a SNES white flash takes two to three seconds to re-establish the connection. During that time, the screen is black. The player has lost control. The RetroTINK 5X-Pro and OSSC Pro handle sync drops better than earlier designs. If a game consistently causes sync drops, enable the scaler's sync regeneration option or add a sync stripper cable.

The mClassic: Borderline

The mClassic dongle from Marseille is a borderline case. Marseille claims sub-millisecond response, but independent testing with a Time Sleuth shows approximately 1 to 2ms of processing delay from the context-adaptive anti-aliasing algorithm. That is still under one frame. It is not zero. The mClassic also requires HDMI input, which means the source must already output digital video. For a Wii with a Wii Dual HDMI mod or a PlayStation with PS1Digital, the mClassic can reduce aliasing without adding a full frame. For a stock Super Famicom, the mClassic is not usable without an external HDMI converter, which defeats the purpose.

The screen is the largest variable in the chain. A modern TV's Game Mode is not a single setting. On a 2024 LG or Samsung, Game Mode reduces processing but still leaves the TV's internal scaler active. The delta between Game Mode and non-Game Mode on a 2024 model is 25 to 100ms. Game Mode is mandatory. It is not sufficient.

PC Mode, if the TV offers it, is faster than Game Mode on some models. PC Mode disables additional post-processing that Game Mode leaves on. The trade-off: PC Mode often restricts colour space or disables HDR. For a retro scaler outputting 1080p or 1440p, PC Mode is worth testing. The Leo Bodnar Lag Tester measures display lag at the centre of the screen. The Shmups Forum competitive lag testing thread documents specific TV models and their real Game Mode and PC Mode lag values. Check it before buying a screen.

The 2024 RTINGS.com measured values for modern TV Game Mode lag range from 5ms at 1080p to 40ms at 4K. The 5ms figure is from a high-end OLED running 1080p input. The 40ms figure is from a mid-range LCD running 4K input. If you are buying a display for competitive retro gaming, test it at the resolution you will actually use. A scaler outputting 1440p to a TV that handles 1440p faster than 4K is a better choice than running the scaler at 4K and relying on the TV's slow 4K path.

The Hong Kong Retro Game Association publishes their tournament equipment list as a community-verified reference point. The chain is a Super Famicom or PS1 with an RGB SCART cable, a RetroTINK 5X-Pro in Frame Lock mode at 1080p output, and a 2024 LG OLED in Game Mode at 1080p input. The scaler and TV are tested with a Time Sleuth and a Leo Bodnar Lag Tester before every event. Total added lag is under one frame. The association also runs a MiSTer FPGA setup for games where original hardware is unavailable. The MiSTer's direct video output in 240p to the 5X matches the response of original hardware.

Rhythm Game Standards Are Stricter

The association's rule for rhythm games is stricter: only the RetroTINK 4K at 120Hz with BFI enabled is approved for official tournaments. The 4K's 0.2ms response and BFI motion clarity eliminate the OLED sample-and-hold blur that rhythm game players find disorienting. The association tests each rhythm game's timing window with a high-speed camera against the original CRT setup. The measured difference is under 1ms. That is the community-verified standard.

Setting Up Your Own Chain

If you are setting up your own Hong Kong apartment, buy a RetroTINK 5X-Pro directly. The scaler's firmware updates are free and frequent. The 5X-Pro's profile system lets you save settings per system, so you do not re-calibrate every time you switch. The failure case is a scaler with outdated firmware that drops sync on a game you play. Update the firmware before you build the chain.

The HDMI handshake between the scaler and the screen adds a fixed minimum time every time the connection is established. This is not a variable you control. The handshake includes EDID exchange and HDCP negotiation. A scaler that loses sync and re-handshakes during gameplay causes a two to three second black screen. That is a round lost in a fighting game. A failed song in a rhythm game.

The most common cause of sync drops is a scaler with tight sync tolerance that cannot handle the voltage dip during a full-white frame on a SNES. The RetroTINK 5X-Pro's sync regeneration option and the OSSC Pro's adaptive sync handling reduce this. If sync drops persist, a sync stripper cable from Insurrection Industries or HD Retrovision stabilises the signal before it reaches the scaler. The cable adds no measurable delay.

The second cause of handshake loops is an EDID mismatch when a capture card is in the chain. The scaler and the capture card negotiate with the screen simultaneously, and the screen's EDID changes depending on which input is active. The fix: power the scaler and the screen first, then connect the capture card after the handshake is stable. The Hong Kong Retro Game Association rule is no capture card in the play path. If you stream, split the hardware signal before the scaler. Send one path to the capture card and one to the scaler. The capture card gets a delayed signal. The play screen gets the fast one.

The single thing that most often goes wrong is assuming the TV's Game Mode is fast enough without measuring it. A 2024 TV in Game Mode can still add 20ms at 1080p and 40ms at 4K. That is one to two and a half frames of lag. A fighting game player who cannot land a punish window that should be two frames wide is not playing worse. The screen is stealing those frames. Measure the TV with a Leo Bodnar Lag Tester or the 240p test suite manual lag camera method before committing to a setup. The Hong Kong Retro Game Association tests every display before a tournament. Test yours before a session.

The lowest-lag retro setup for fighting and rhythm games suits the Hong Kong apartment gamer who needs one screen for modern and retro hardware and is willing to spend on a RetroTINK 5X-Pro or RetroTINK 4K. It suits the competitive player who perceives milliseconds of lag and needs a repeatable, measurable chain. It does not suit someone buying a pre-modded system as a plug-and-play gift with no interest in settings menus. It does not suit a PC gamer trying to upscale old 3D titles from a desktop. It does not suit a modern gamer wanting better anti-aliasing on a PlayStation 5. Those are different problems with different solutions. This one is about the chain, the numbers, and the feel.