Identifying Fake and Clone HDMI Anti-Aliasing Adapters Sold as Retro Gaming Upscalers

What You Are Actually Being Sold

In a cramped booth at Golden Computer Centre, a seller holds up a small dark-gray HDMI dongle and calls it an mClassic at a price that undercuts every authorised retailer. The box has the right colour, the right font, even a passable Marseille logo. The moment you pick it up, your hand tells you something is wrong. It feels hollow. Too light. The HDMI connector wiggles. A board-mounted jack on a genuine unit has no business being loose. Before you hand over cash, know how to spot a fake. The counterfeit market in Sham Shui Po is built on selling you a DVD-player chip disguised as a Marseille mClassic. The difference between the two is the difference between a clean 240p image and a disaster of combing artifacts and lag that ruins every retro game you own.

The Real ASIC and What It Processes

The authentic device, made by Marseille Inc., is a plug-and-play HDMI dongle that applies contextual, post-process anti-aliasing to any 1080p input. A custom ASIC analyses each frame's edges and smooths them in real time. It accepts a maximum input resolution of 1920×1080 and outputs a 2560×1440 frame at 60 Hz. It draws 1.5 W from the HDMI port itself. No power brick. No USB cable. A physical switch toggles three modes: off, upscale, and retro. Independent tests using a Time Sleuth latency meter measure its processing lag at 1-2 ms, effectively imperceptible. Its context-adaptive algorithm smooths a PS1 sprite without smearing text or adding halos.

Why the Clone Chip Fails

The clone is built on a DVD chipset designed for 480i video, not 240p games. It lacks the custom ASIC entirely. Scaling is done by a generic chip that treats every signal as interlaced, destroying the progressive-scan information that makes retro games look right on a modern display.

Plug any clone into a SNES, Genesis, or NES and you will see the failure within seconds. The clone sees a 240p signal and misdetects it as 480i. It applies a destructive deinterlace that discards half the vertical resolution and adds combing artifacts, jagged horizontal lines that flicker across moving sprites. Super Mario World looks like it is being played through a shredded screen. This is not a minor quality gap. It is a completely different signal treatment, caused by a scaler chip designed for a DVD player's 480i output, not a console's progressive-scan 240p. The Marseille device processes 240p as 240p. It preserves the scanline effect and the clean edges that post-process AA is designed to smooth. That is why an authentic unit costs several times as much. The ASIC solves a problem the clone's chip does not even know exists.

• mClassic manufacturer: Marseille Inc.
• mClassic input resolution maximum: 1920×1080 (1080p)
• mClassic output resolution maximum at 60 Hz: 2560×1440 (1440p)
• mClassic processing chip: Marseille mClassic ASIC (custom)
• mClassic anti-aliasing method: Contextual, post-process anti-aliasing with scaling
• mClassic power consumption: 1.5 W typical

Weight and the First Three Checks

You do not need a microscope. You need seven minutes and a working knowledge of what an authentic unit does not do. First, weigh it. The real thing is 28 g. Anything significantly lighter that floats in your palm is a red flag. Second, look at the HDMI connector's colour. The official unit is dark gray, not black or silver. The LED is a specific blue in active mode and amber in standby. Clones use cheap red or green LEDs visible through the case. Third, check the packaging for the "mClassic" trademark and the Marseille logo. A clone's box carries a similar but not identical mark, often missing the registration symbol.

Firmware, PCB and Output

Fourth, try the firmware update. The genuine device has a USB-C port on the side, and Marseille Inc. publishes update tools on its website. A clone will not connect to that software. The absence of a firmware update path is a definitive fail. Fifth, inspect the PCB if you can open it. The genuine board has the Marseille ASIC clearly marked. A clone has a chip with the marking sanded off or replaced with a generic part number. Sixth, test the 1440p output claim. The authentic device outputs a 1440p frame with a 1080p active image area upscaled, not a native 1440p render, and it switches modes cleanly on any display. A clone drops the signal or outputs a blank screen when you switch resolutions.

The 240p Test Suite

Finally, run a 240p test suite from a flash cart. The genuine unit shows a clean test pattern. A clone displays the characteristic combing artifacts of 240p-as-480i misdetection.

When you buy a clone HDMI upscaler for retro gaming in Sham Shui Po, you are not getting a lesser version of the same thing. You are getting a different thing entirely. The Marseille custom ASIC is a post-process AA engine that analyses each frame's edges and applies context-adaptive smoothing. A clone's generic scaler chip is a DVD player converter with no concept of game signals. This difference shows up in every frame. The clone adds 2-4 frames of lag on top of the display's inherent latency, because the DVD chip's deinterlacer buffers a full frame before it can process it. It discards the odd or even scanlines of a 240p signal, halving the vertical resolution before it reaches your TV. The authentic unit adds only 1-2 ms of processing latency and preserves the full 240p image. A retro gamer who values responsiveness will not accept a clone as a substitute.

Learning to spot a fake Marseille dongle is a skill you develop once and use forever. Counterfeiters improve their game over time. The most reliable indicator is build quality. The genuine unit has a solid, injection-moulded case with a smooth matte finish. The HDMI connector is board-mounted with a reinforced joint that does not wiggle when you move the cable. A clone has a hollow feel, a glossy or rough finish, and a board-mounted jack that flexes under pressure. That design flaw causes signal dropouts within a year of cable changes. Check the USB-C port. The official device has a recessed port that accepts a standard cable. Clones have a shallower cutout or a non-standard pinout that will not charge or update. Finally, look at the physical switch. The real thing has a three-position switch with a positive, tactile click. Clones have a cheap sliding switch that feels mushy and breaks after a few dozen uses.

Golden Computer Centre and Ap Liu Street

If you are shopping for a deal on a scaler in Hong Kong, you will eventually find yourself in Sham Shui Po, specifically at Golden Computer Centre on Fuk Wa Street or the electronics stalls of Ap Liu Street. A fake scaler here is not a rare find. It is the default inventory of at least a dozen stalls. The sellers are professionals who know exactly what they are selling. The most common fake is the mClassic clone, priced at a fraction of the authentic unit's cost. You will also find counterfeit OSSC and RetroTINK units with missing silkscreens and sanded-off FPGAs.

How to Buy Safely in the Maze

Golden Computer Centre is a four-floor maze of tiny booths. Sellers are skilled at switching between a genuine unit and a fake when they think you are not looking. Your best defence is to buy from a shop with a physical address you can verify, not a stall that pulls boxes from under the counter. If you are buying late at night when the markets are closed, your only option is online, but the same rules apply. Check the seller's feedback for complaints about fakes. Demand photos of the actual device, not stock images.

Marseille's marketing claims the device outputs 1440p. That is true, but the claim hides a gap. The scaler takes a 1080p input and outputs a 1440p frame with a 1080p active image area upscaled, not a native 1440p render. The device is not generating new pixel data at the output resolution. It applies post-process AA to the 1080p image and then scales the result to fill a 1440p frame. The active game image is still 1080p with added smoothing. For a retro gamer playing on a 1440p monitor, this is fine. The job of the unit is to smooth jaggies, not to add native resolution. But it explains why the 1440p output looks softer than a native 1440p render from a PC. Independent tests using a Time Sleuth confirm the 1-2 ms latency. No test has shown the device adds real resolution beyond 1080p. Treat the 1440p claim as marketing, not a specification.

The Time Sleuth is the standard tool for measuring lag in a scaler. The numbers it produces are the only ones that matter when you are choosing between an authentic unit and a clone. A genuine Marseille device measures 1-2 ms of processing latency, within the margin of error for most test setups and effectively imperceptible in gameplay. A clone measures 30-50 ms of added lag. That is 2-3 frames at 60 Hz. Enough to make a platformer feel sluggish and a fighting game unplayable. The difference is not subjective. It is a number you can measure, caused by the clone's deinterlacer buffering a full frame before processing while the ASIC analyses each line as it arrives. If a seller will not let you test with a 240p test suite or a Time Sleuth, assume the worst and walk away.

If you are in Golden Computer Centre and a seller offers you an mClassic at a price that seems too low, you are not getting a deal. You are getting a clone, and the seller knows it. The failure mode is simple. You buy the clone. You plug it into your SNES. Within minutes you see the combing artifacts and lag that ruin your game. You return it. The seller has a strict no-refund policy for "opened" electronics. You are out the money. The alternative is to buy an authentic unit from a reputable dealer, pay the full retail price, and know you are getting a device that has been tested. If the shop is closed or you are online late at night, buy from a seller with a physical address and a return policy. Use a credit card that offers purchase protection. Your bank will reverse the charge if the item is fake, but only if you have documentation of the fraud.

If you remember one thing, it is that the physical weight and board-mounted jack of a device are your first line of defence against a counterfeit. No fake mClassic has ever passed a 28 g weight test or a wiggle test with the HDMI connector.