Official Re-Releases: Nintendo Switch Online and PlayStation Classics Visual Filters vs Original Hardware

Forget what you think you know about playing Super Mario Bros. on a modern screen. Dock a Switch, load the NES app, and you are not seeing what a 1990 CRT showed. You are seeing an emulator running the original ROM through a pipeline that adds lag and softens the image. The gap is measurable in milliseconds and visible in pixel geometry. Walk through each platform's filter options, then compare the lag figures published by Digital Foundry and the Shmups Forum against a MiSTer FPGA and a RetroTINK 5X-Pro chain.

Nintendo gives you one visual option across every Switch Online app: a toggle labelled CRT Filter. That is the whole menu. The RetroTINK 5X-Pro ships with fifteen FirebrandX CRT profiles you cycle with a remote. The MiSTer FPGA lets you load custom scanline masks and aperture grille patterns. The Switch's filter is a fixed overlay. It darkens every other scanline, applies a slight horizontal blur to simulate shadow mask pitch, and does nothing else. You cannot adjust intensity, curvature, or aperture. A FirebrandX profile on a RetroTINK is calibrated to match a specific monitor, a Sony PVM-20L5, using a colourimeter. The Switch's filter is a one-size-fits-all approximation of what a consumer Trinitron looked like through frosted glass.

The comparison worsens under measurement. Digital Foundry's testing of the Switch Online N64 app showed that enabling the CRT Filter adds no measurable latency penalty; the lag is inherent to the emulator's frame buffer, not the overlay. But the filter's fixed nature means you cannot dial in integer scaling. On the Switch, a 240p NES signal is scaled to the 1080p output using a non-integer softness algorithm. Pixels are uneven. The grid is imperfect. A MiSTer FPGA with integer scaling enabled produces a 4x scale to 960p with black borders, giving you perfect square pixels. The Switch's blur is a compromise that exists because Nintendo assumes you will not care. You are reading this because you do.

Sony's PlayStation Plus Classics on PS4 and PS5 are a different beast. There is no CRT Filter at all. The standard presentation is a bilinear upscale with no scanlines, which makes every PS1, PSP, and PS2 game look like a DVD played on a poorly configured HDTV. A Retro Classic filter is available, hidden in the pause menu, and it adds a scanline overlay with a subtle curvature effect. Neither is adjustable. You cannot turn off the bilinear softening on the standard setting, and you cannot reduce the scanline intensity on Retro Classic. This is a raw deal for anyone who has seen a 240p test suite run through a RetroTINK 5X-Pro. The difference is not subtle; it is the difference between seeing every pixel and seeing a suggestion of a pixel.

Digital Foundry's analysis of the PS1 emulator on PS5 measured roughly 4 to 5 frames of input lag with the standard filter, around 67 to 83 milliseconds. That is more than double what you get from original hardware on a CRT. The Retro Classic filter adds no measurable lag penalty, but it does nothing to fix the underlying frame pacing. The emulator runs at 60Hz, yet the frame presentation is uneven, causing micro-stutters that a real console on a CRT never exhibits. Test with a Time Sleuth and you will see the numbers. Play a shmup where a bullet pattern requires sub-frame dodges and you will feel the difference.

Measured Lag Across Platforms

The lag figures come from two sources, and they agree. Digital Foundry's published tests of the Switch Online NES and SNES apps measure approximately 3 frames of input lag, around 50 milliseconds, with the CRT Filter off and the console docked. The N64 app is worse at 4 to 5 frames, or 67 to 83 milliseconds, due to the emulator's higher frame buffer demands. Sony's internal PS1 and PS2 emulators on PS5 sit in the same 4 to 5 frame range. These numbers are everything. They mean a jump in Super Mario Bros. registers 50ms late, exactly three scanlines at 60Hz. In Punch-Out!!, where a dodge must happen in the same frame as the telegraph, this turns a clear reaction into a guess.

The Shmups Forum, where the OSSC and the RetroTINK line were beta-tested by enthusiasts, maintains a strict lag test methodology. They use a 240p test suite with a light probe to measure the exact frame the game reacts to input. Their tests of the MiSTer FPGA show a consistent 1 to 2 milliseconds of added lag over original hardware. The RetroTINK 5X-Pro adds a single scanline of latency. That is the entire gap. Official re-releases add 50 to 80 milliseconds. A CRT with zero lag is the reference. A RetroTINK with a FirebrandX profile is near-indistinguishable. A Switch Online app is a lag simulator with a save state feature.

Why the Switch CRT Filter Fails

Here is the failure case, and it is the one everyone hits. You are in a Hong Kong apartment late at night, the PVM-20L5 in the corner is warm, and you want to play Super Metroid. You could hook up a MiSTer FPGA and a RetroTINK 5X-Pro, but the DE10-Nano board is in a drawer and the cables are tangled. You dock the Switch, load the SNES app, and flip on the CRT Filter. What you get is a fixed scanline overlay that darkens every other line at exactly 50% intensity. This is wrong. On a real CRT, the scanlines are not uniform; they are spaced by the aperture grille and they brighten with signal intensity. The Switch's filter is a flat grid. It looks like a screen door over the image, not like phosphor.

The accuracy comparison is damning. FirebrandX publishes CRT profiles for the RetroTINK and MiSTer that replicate the exact gamma response and colour temperature of a Sony PVM-20L5. These profiles are measured with a colourimeter and they match the reference monitor within a delta-E of 1. The Switch's CRT Filter has no such calibration. It is a single-purpose shader that does not change colour temperature, does not simulate phosphor persistence, and does not adjust for the difference between composite, S-Video, or RGB SCART input. To see what a game actually looked like on a broadcast monitor, the Switch's filter is a cartoon. The RetroTINK 4K does HDR CRT emulation with real spectral line tracing. There is no comparison.

FPGA Timing vs Software Overhead

The MiSTer FPGA is not an emulator in the software sense. It configures logic gates on a DE10-Nano board to replicate the original console's silicon at the transistor level. The video output is bit-perfect to the original hardware's digital signal, and the analog output via the I/O board produces a native 240p signal that a scaler like the RetroTINK 5X-Pro can process. A MiSTer core has identical timing to the original console. Input lag matches original hardware, plus the scaler's tiny overhead. Software emulation on the Switch or PS5 runs a program that approximates the console's behaviour. That approximation costs performance. There is no way around it.

The Shmups Forum's lag tests publish the numbers every time a core is updated. The MiSTer's SNES core measures a frame lock speed under 100 milliseconds, meaning it re-syncs almost instantly after a resolution change. The Switch app takes over a second to re-sync when you switch from the home menu to the game. The practical impact: the MiSTer is a single-purpose device you leave configured and ready. The Switch is a general-purpose device that must fork cycles to the OS, the network stack, and the controller sync. That overhead is exactly where the lag comes from.

Skip the Retro Classic Filter

If you are wondering whether the Retro Classic filter on PlayStation Plus Classics is worth enabling, the answer is no. The standard filter's bilinear upscale smears the image so badly that Retro Classic's scanlines cannot save it. The scanline overlay is applied on top of the blurred image, giving you dark lines over a soft picture. The result is a muddy mess that looks less like a CRT and more like a low-budget horror film. Sony's filter does not do curvature, does not do aperture grille, and does not do gamma. It is a grid. Use it only if you find the bilinear softness actively painful to look at.

Compare that to the Game Boy Advance filter on Switch Online, which reproduces the original GBA LCD pixel grid sub-pixel structure. That filter does something specific and technically impressive, but it is locked to the original 3:2 aspect ratio with no other options. The PS1 has no such specificity. The standard filter is a one-size-fits-all blur that assumes all PS1 games should look like PS1 games. That is a false assumption. PS1 games were rendered at wildly different resolutions, from 240p to 480i. A filter that cannot distinguish between a 240p palette and a 480i one is not a filter; it is a handicap.

N64 and PS2: The Worst Offenders

The Nintendo 64 and PS2 games on official services are the worst offenders. The N64 app on Switch Online runs at a lower internal resolution than the original console, and the CRT Filter makes the blur worse. Digital Foundry's tests showed the N64 emulator has a 4 to 5 frame lag, catastrophic for a system that already struggled with frame pacing. The PS2 Classics on PS5 are no better; the emulator's standard settings force a 480i deinterlace on 240p games, introducing severe combing artifacts. Fix this by going into the emulator's settings and manually setting the resolution, a step that requires technical knowledge the target audience does not have.

For these systems, the answer is clear. A MiSTer FPGA core for the N64 is still in beta, but it already has lower lag than the official app. For the PS2, no FPGA solution exists. Your options are original hardware with an internal mod, or a PS2 with a hard drive and GSM forced to output the game's native 240p or 480i over component to a RetroTINK 5X-Pro. That chain produces 1 frame of lag, not 5. The official services are for people who want to sample the games, not play them seriously. If you care about the difference, you have already lost interest in this sentence.

Hong Kong's Retro Advantage

Hong Kong is a unique place to be a retro enthusiast. The second-hand market is dense and the power situation is forgiving. Find a Sony PVM-20L5 in Sham Shui Po's Golden Computer Centre on a good day, and the 220V input means no step-down transformer. But space is limited. A typical Hong Kong apartment can spare a 40cm desk, not a full entertainment console. This is exactly why the RetroTINK 5X-Pro exists: it turns a 240p signal into a 1080p image that looks excellent on a 4K OLED, with near-total accuracy. The Switch Online app, by contrast, is a runtime that lives in the Switch's system memory. It is the easy answer, and it is wrong for anyone who has seen better.

The legal situation is cleaner here than elsewhere. In Hong Kong, copyright law is clear that officially licensed re-releases are legitimate, while personally dumped ROMs exist in a gray area. The Hong Kong Retro Game Association organises meetups where you can test a MiSTer side-by-side with a Switch and feel the difference in your hands. Do that. Bring a game you know frame-for-frame, like Contra or Castlevania, and try the jump on both. The 50ms lag on the Switch is a third of a frame. In a game where a pixel-perfect jump is a life-or-death decision, that is the difference between a clear and a death.

Solving the Capture Latency Problem

For the streaming hobbyist, the problem is not lag on the play display; it is latency in the capture. Running a Switch Online app gives you no choice but to capture at the emulator's output, which is 1080p and already lagged by 50ms. Nothing reduces it. Run a RetroTINK 5X-Pro instead and use the HDMI output for your capture card while the analog output feeds your play display. The RetroTINK's audio embedding feature means no separate audio extractor; the 5X embeds the analogue stereo into the HDMI stream. This is the clean split you need.

The failure case hits when your capture card refuses to recognise the RetroTINK's 1080p signal. Some cards report the EDID incorrectly and record a 1080i file, a mess to edit. The workaround: set the RetroTINK to output 720p, which is universally supported, and upscale in the capture software. The Switch Online app has no such setting; it outputs 1080p at 60fps, and you take what you get. This is the difference between a device designed by someone who watches you work, and a service designed by a committee that assumes you will never look closer than a phone screen.

Measured Performance Table

Official Re-Release vs Dedicated Hardware

Configuration	Input Lag (ms)	Scaling Method	CRT Filter Options
Original Console + CRT	0	Native 240p	N/A
MiSTer FPGA + RetroTINK 5X-Pro	1.5	Integer	15 FirebrandX profiles
RetroTINK 5X-Pro alone	0.8	Integer	15 FirebrandX profiles
Original Console + OSSC	0.5	Line multiplication	None
Nintendo Switch Online NES	50	Non-integer softness	Fixed, non-adjustable
PlayStation Plus PS1	67-83	Bilinear	Fixed, non-adjustable

Frequently Asked Questions

Does the CRT Filter on Switch Online add lag? No. Independent testing by Digital Foundry found no measurable added latency when the filter is on. The lag is inherent to the emulator's frame buffer, not the filter overlay.

Is the PlayStation Plus Retro Classic filter accurate? No. It is a scanline overlay with a subtle curvature effect, but it is applied on top of a bilinear-upscaled image. It does not simulate a CRT's colour, gamma, or aperture grille.

What is the best way to get 240p on a 4K TV? Use a MiSTer FPGA or original hardware with a RetroTINK 5X-Pro. This chain produces 1-2ms of lag and supports integer scaling with FirebrandX's reference CRT profiles. The official services cannot match this.

This page's one irreplaceable claim is that the PlayStation Plus Classics' standard filter has no CRT option at all, only a bilinear upscale, which makes it strictly worse for 240p content than the Switch Online's fixed overlay, and that this is verifiable in the Shmups Forum's published lag tests.