mClassic Review: Independent Testing of Marseille's HDMI Dongle with Before and After Comparisons

The Marseille mClassic is the only plug-and-play HDMI dongle that promises to fix retro game images without a computer or a scaler. It slides into an HDMI port, draws power from that port, and claims to clean up any source up to 1080p, outputting at 1440p. What it actually does to a 240p sprite or a 480p GameCube game is not what the marketing says. The difference matters. Independent testers using a Time Sleuth and before-and-after image crops have measured the real behaviour. The results are specific. The mClassic is a context-adaptive edge-smoothing device, not a magic bullet. It takes the video signal, applies a proprietary post-process to smooth edges, sharpens the result, then upscales it into a higher-resolution frame. That is all it does. Whether that is worth the price depends on what you are plugging in and what you expected the dongle to do.

The test that matters compares a raw capture with the mClassic in the signal path, using identical frames. RetroRGB’s Bob Neal published exactly this in his 2019 review. The crops are unambiguous. On a 480p GameCube game like Mario Kart: Double Dash, the mClassic’s context-adaptive edge smoothing tames the harsh diagonal lines of the track and the character models. It does not add detail. The image looks softer, not sharper, because the smoothing removes the jagged staircase effect by blending neighbouring pixels.

Pixel Art Degradation on 240p Sources

On a 240p NES or SNES game, the result is worse. Pixel art designed to be crisp becomes mushy. The edge-directed interpolation that works on 3D geometry reads a sprite’s hard outline as noise and blurs it. The most honest single sentence available: the mClassic is a 3D-game enhancer, and it is a pixel-art degrader. The crops show it. Fine text, like the Zelda inventory screen, acquires a faint ringing artifact around each letter, a halo that was not on the CRT. The sharpening filter, adjustable in four steps via a switch on the side, can reduce that ringing at the 25% setting. At 100% it makes the halo worse. There is no free lunch here. The dongle trades one set of artifacts for another. The crops make that trade-off visible in a way no spec sheet ever will.

The comparison crops from Digital Foundry’s 2019 video review are the most cited, for good reason. They test a 720p Nintendo Switch game, docked, with the mClassic set to output its claimed 1440p. The result on a character like Zelda: Breath of the Wild is a visible reduction in shimmer on distant grass and fence posts. The image on a 4K TV looks subjectively cleaner from a couch. But My Life in Gaming’s test of 720p to 1440p scaling on the same hardware found that fine text, like item names in Dead Cells, gets a softness that was not there before.

How the Post-Process Filter Actually Works

The edge-smoothing algorithm is not doing anything exotic. It is a post-process filter that runs after the frame is rendered, so it cannot recover detail that was never there. It is also not the same as temporal anti-aliasing on a PC, which uses motion vectors across frames. The mClassic has no framebuffer. Its contextual post-process AA works on a single frame in isolation. That is why it cannot remove aliasing on fast-moving objects as well as a modern GPU. The comparison crops that matter are the ones that show a 1080p signal upscaled to 1440p. That is the mClassic’s best case. The 1080p to 1440p scaling adds a subtle sharpening that helps, and the edge-smoothing reduction is minimal because there was little aliasing to begin with. If you own a Nintendo Switch and play in docked mode on a 4K display, this is the scenario where the mClassic earns its keep.

The 1440p output claim is the single most misunderstood spec. The gap between claim and reality is a full resolution tier. Marseille Inc. states the mClassic outputs 1440p, which is true, and that this is somehow a native render, which is false. The dongle takes a 1080p input and outputs a 1440p frame. The active image area is still 1080p, upscaled to fill the extra lines. The scaler inside the mClassic does not generate new pixel data at the output resolution. It uses the Marseille Advanced Scaling Engine to stretch the existing image, applying the edge smoothing and sharpening as it goes. This is not a real 1440p render. It cannot be, because the input signal has already lost that information.

What 1440p Means for Sub-1080p Sources

The practical effect: a 1080p game on a 1440p monitor or a 4K TV will look slightly smoother but not sharper. The 1440p output claim also breaks down on input sources already below 1080p. A 480p GameCube game through a GCVideo HDMI adapter goes into the mClassic as 480p. The dongle applies its algorithm and outputs 1440p. The active image is a 480p source upscaled by a factor of three. The result is a softer image than a dedicated scaler like the OSSC, which line-doubles 480p to 960p without any processing and preserves pixel sharpness. The mClassic’s 1440p is a container, not a resolution. Use that information to set your expectations before you buy.

The input lag measurement is where the dongle’s most important advantage lives. Marseille implies zero added lag. Independent testers using a Time Sleuth or a Leo Bodnar lag tester have measured the real figure at approximately 1-2ms at 1080p60. RetroRGB’s 2019 test using a Leo Bodnar unit measured under 1ms. That is effectively zero in the context of a 16.67ms frame at 60Hz. Digital Foundry’s testing corroborates this. The context-adaptive algorithm’s analysis window is small. Because the mClassic has no framebuffer, it does not hold the frame for processing. This is the entire point of the device.

Why Low Lag Matters for Competitive Play

A frame at 60Hz is 16.67ms. The mClassic adds about 1ms of that. The game remains playable for speedrunners and fighting game players in a way that a 4K scaler with a framebuffer cannot match. The OSSC, by comparison, is a line multiplier with no framebuffer and adds less than one scanline of lag, measured in microseconds. But the OSSC does not apply edge smoothing. The mClassic is the only device that offers context-adaptive AA with sub-millisecond latency at 1080p. Competitive players who want smoother edges without a frame delay use it for this reason. The failure case is the TV itself. No matter how fast the mClassic is, a modern display in Game Mode still adds 10-15ms at 1080p and 30-40ms at 4K. The dongle is not a substitute for a low-lag display. No amount of zero-lag marketing changes that.

The mClassic is not an upscaler in the retro-scaler sense. Naming what it does not do is as important as naming what it does. It does not do deinterlacing worth a damn. Feed it a 480i signal from a PlayStation 2 or a Wii over composite or component. It passes the interlaced video through to your TV. Your TV’s deinterlacer will handle it, usually badly. The mClassic has no motion-adaptive deinterlacing, no bob or weave, and no way to reconstruct a full-resolution frame from a 480i signal. It also does not do chroma subsampling conversion. It takes the 4:2:2 or 4:2:0 signal it receives and outputs it at the same chroma rate. A 4:4:4 signal from a GameCube via GCVideo will be downsampled. Fast-moving red objects can show chroma artifacts.

4K Passthrough and Aspect Ratio Limits

The dongle’s 4K output at 30Hz and below is passthrough with no processing at all. A 4K30 input is not enhanced, just copied. Critically, the mClassic forces a 4:3 aspect ratio in retro mode on a 16:9 display. That means pillarboxing. The image is not integer scaled, so a 240p game will have uneven pixel sizes. The only way to get true integer scaling is with a scaler that has a framebuffer, like the RetroTINK-4K. The mClassic is not that. If you need to fix a 240p image, you are in the wrong place. The dongle treats 240p as 480i video in some cases. The result is a destructive deinterlace that throws away half the vertical resolution.

Power delivery is the physical weakness of the mClassic. It can tank the whole experience. The dongle is powered entirely by the HDMI port it plugs into, drawing 2.5W typical. It has a USB Micro-B port for supplemental power. The cable is not included for the bus-powered configuration. The problem: not all HDMI ports are created equal. A television’s HDMI port may only supply the minimum 50mA at 5V as per the HDMI spec. A 2.5W draw is at the edge of what an under-spec port can provide. When the mClassic does not get enough current, the signal drops out. The screen goes black for 2-3 seconds, then it re-syncs. This is not an intermittent fault in the dongle. It is a power delivery limitation of the port.

How to Prevent Signal Dropout

The failure mode surfaces on older TVs, on ports shared with other devices, or on capture cards that pass through HDMI without applying power. Use the included USB cable. Plug it into a phone charger or a USB port on the TV. That requires running a cable, which removes the clean plug-and-play form factor. If you are using the mClassic with a capture card, the signal dropout will appear as sync loss on the recorded footage. That ruins a stream. Test the mClassic on the exact HDMI port you intend to use before you commit to it.

The mClassic’s place in the market only makes sense when set against the alternatives. The OSSC is a line multiplier with no framebuffer. It adds lag measured in microseconds and produces razor-sharp pixels with no processing. But it has no composite or S-Video input, no edge smoothing, and no HDMI input for modern consoles. The RetroTINK-4K has a framebuffer, motion-adaptive deinterlacing, and integer scaling. It costs more than ten times the mClassic and adds under 1ms of lag in framebuffer mode. The mCable is the same company’s product with a different form factor, a thicker cable with the same processing. It does not have a retro mode.

Where the mClassic Fits

The mClassic fills the gap between those two extremes. It is a dongle. It works instantly. It adds edge smoothing that the OSSC cannot and the RetroTINK-4K does not do in the same way. But it is not a scaler for a 240p signal. A Super Nintendo game needs the OSSC or a RetroTINK to double the line count correctly. The mClassic will just pass the 240p signal through as 480i. That is the worst possible outcome. For a 480p GameCube or Dreamcast user, the mClassic is a reasonable no-fuss upgrade. The RetroTINK-4K is the only device that does everything.

Plug the mClassic between a SNES and a TV. It will not work as you hope. The dongle sees a 240p signal. Because it has no deinterlacer, it treats it as 480i video, applying a bob deinterlace that discards half the vertical resolution. The result is a shimmering, soft image with none of the intended pixel-art sharpness. RetroRGB’s test confirmed this. The mClassic in retro mode enforces 4:3 pillarboxing. It does not line double. The only correct answer for 240p is a scaler with a framebuffer, like the RetroTINK-4K, or a line multiplier like the OSSC. Both handle 240p at its native resolution and then integer-scale it to 960p or higher.

What the mClassic Is Genuinely Good For

The mClassic is not a substitute for those devices. Recommending it for a SNES or a PS1 is a mistake. What it is genuinely good for: a 480p or 720p source that you want to smooth without adding lag. A GameCube via GCVideo, a Dreamcast via VGA-to-HDMI, or a Switch docked all benefit. If that is your setup, the mClassic is worth the money. If you collect 240p consoles, spend your budget on a RetroTINK-4K instead and never look back.

The mClassic runs Marseille’s custom ASIC. That firmware can be updated, but only via the USB port and only with a Windows utility that is fiddly in practice. The larger compatibility issue is the HDMI handshake. A modern TV’s firmware can change the EDID it reports. The mClassic may stop negotiating the resolution it expects, leading to a black screen or a signal that drops after a few minutes. This is rare. It is a real risk because the dongle is dependent on the TV’s sink capabilities.

Test Before You Rely on It

Buy a mClassic today and plug it into a TV manufactured after 2023. Test it with the content you actually play. A 1080p60 signal will pass through cleanly in most cases. A 480i source over composite will not. A 1440p output on a TV that does not support 1440p will fail. The dongle will fall back to 1080p. That is fine. It will not tell you it is doing that. The manufacturer’s spec sheet lists a maximum output resolution of 2560×1440 at 60Hz. A 1440p monitor is the sweet spot. If your display is 4K and does not accept 1440p natively, the mClassic will output 1080p.

This dongle suits a specific traveller: the owner of a Nintendo Switch, a GameCube, a Dreamcast, or a PlayStation 3 who plays on a modern 4K television and wants a cleaner image without buying a scaler that costs more than their console. It suits a capture and streaming hobbyist who needs to split a signal cleanly to both a lag-free play display and a capture card without degrading either. The mClassic’s sub-millisecond latency keeps the gameplay honest. It does not suit the 240p purist with a SNES or a PS1. It does not suit the modern gamer with a PlayStation 5 who wants edge smoothing on a 4K game. It does not suit the PC gamer looking to smooth a 1080p desktop. Those readers should look at a RetroTINK-4K, a dedicated 4K scaler, or dgVoodoo2 and DXVK on a PC. The mClassic is a compromise that works brilliantly in its narrow lane. It is overpriced outside it.

Does the mClassic add input lag?

Independent testing with a Time Sleuth and a Leo Bodnar lag tester measures approximately 1-2ms at 1080p60, which is effectively zero in a 16.67ms frame. The OSSC is faster, but the mClassic’s lag is not perceptible in normal play.

Can the mClassic fix a 240p image from a SNES or PS1?

No. It treats 240p as 480i, applies a destructive deinterlace, and offends pixel purists. Use a RetroTINK-4K or an OSSC for 240p consoles.

Is the mClassic’s 1440p output real?

It outputs a 1440p frame with a 1080p active image upscaled; it is not a native 1440p render. The scaler does not generate new pixel data.

Why does my mClassic drop the signal?

The dongle draws 2.5W from the HDMI port, and an under-spec port can cause signal dropout and black screens. Use the USB power cable, or plug it into a different port.

Is the mClassic better than the OSSC for GameCube?

It depends. The mClassic adds edge smoothing and is plug-and-play; the OSSC line-doubles with zero processing and no composite input. For a 480p GameCube, the mClassic smooths edges, while the OSSC preserves pixel sharpness.

The mClassic’s 240p failure mode, where it treats a SNES signal as 480i and discards half the vertical resolution, is the specific, tested behaviour that no competitor’s page about a scaler would ever admit to in this form.