Line Doubling vs Framebuffer Scaling in Retro Game Upscalers

The moment you start shopping for a retro upscaler, you hit the fork in the road: line doubling vs framebuffer scaling. These two architectures decide everything about your lag, your image quality, and which consoles you can even plug in. It is not a brand preference or a price point. It is a physical law. The OSSC is a line multiplier: it takes each incoming scanline and draws it again, immediately, no storage, no waiting. The RetroTINK 5X-Pro is a framebuffer scaler: it captures the whole frame into memory, processes it, and then outputs a new frame. One adds microseconds. The other adds at least one frame period. That is the entire subject. Every feature, every failure mode, and every buying decision hangs off it.

The OSSC is the reference line multiplier. It does exactly one thing and does it with no memory. Each input line from a 240p source is repeated to produce a 480p, 720p, or 1080p output. Line2x repeats each line once. Line3x repeats each line twice. Line4x and Line5x for the 240p sources follow the same pattern up to the limits of the output DAC. Since the scaler never stores a frame, the latency is measured not in milliseconds but in scanlines: the OSSC's own spec is under 2 scanlines of added lag. A scanline at 60Hz is roughly 32 microseconds. Two scanlines is 64 microseconds. The time it takes a 60Hz frame to appear is 16.67 milliseconds. The OSSC's entire processing chain is a tiny fraction of one frame. That is why the Shmups Forum testing methodology reports OSSC latency as effectively zero, and why FirebrandX's CRT profiles for the device exist at all: the device is just a wire with repetition.

The cost of that purity is architectural. A line multiplier cannot do anything except repeat lines. It cannot scale to a non-integer multiple of the input. A 240p source at 4x gives 960p, and that is the output. If your display is 1080p, the scaler inside your TV has to do the last 120 lines of scaling. That TV scaler adds the very lag you bought the OSSC to avoid. The OSSC cannot deinterlace 480i properly because deinterlacing requires a frame buffer to merge fields. It cannot rotate the image. It cannot zoom into a region of the frame. It can apply a scanline filter that fakes a CRT, since a scanline filter draws dark lines between every output line, but it cannot do it with the variable intensity of a real CRT beam. What it can do, and what it does better than anything on earth, is produce a pixel-perfect, zero-lag image for 240p consoles over RGB SCART.

The RetroTINK 5X-Pro is the reference framebuffer scaler for the opposite reason: it does everything. It captures the entire source frame into memory, applies its processing stack, and then outputs a new frame at your chosen resolution. The input cannot be a single scanline anymore. The processing includes motion-adaptive deinterlacing, which requires both fields of a 480i frame to be present before it can merge them. It includes a scanline filter that operates across the full frame. It includes zoom, rotation, and aspect ratio correction. And it accepts composite, S-Video, component, and RGB SCART. It handles the consoles the OSSC cannot even see.

Where the lag comes from

The price of that flexibility is measurable and physical. Any processing step that buffers a full frame adds at least one frame time of lag. At 60Hz, one frame is 16.67 milliseconds. The RetroTINK 5X-Pro in framebuffer mode measures between 0.25 and 0.5 frames on RetroRGB's Time Sleuth. That is better than one frame since the scaler starts outputting the processed frame before the entire next input frame arrives, a technique called low-lag frame buffering. But it is still a world away from the OSSC's 64 microseconds. The RetroTINK 4K, which adds HDR processing and black frame insertion, sits at 0.25 to 1 frame depending on output resolution. The HDR tone mapping and the 4K scanline masks take time to compute. The physics does not care about brand. It cares about memory access.

You need a real number to hold both architectures against, so here is the one that matters: a frame at 60Hz is 16.67 milliseconds. That is not a marketing claim. It is the reciprocal of 60. It is the physical constant that separates the two devices. The OSSC's entire pipeline costs under 0.1 milliseconds. The RetroTINK 5X-Pro costs between 4 and 8 milliseconds. The RetroTINK 4K costs up to 16.7 milliseconds when HDR is on. The difference between the OSSC and the RetroTINK 5X is roughly one hundred times the difference between the RetroTINK 5X and a TV in Game Mode, which adds 10 to 40 milliseconds. The gap between a line multiplier and a framebuffer scaler is not a matter of tuning. It is structural. You cannot tune your way out of a frame buffer. The only way to avoid the lag is to not have the buffer.

This is why the classic advice holds. If you play 240p games on a CRT or a 1ms OLED and you want the absolute fastest response, buy the OSSC. If you play 480i PlayStation 2 games, or you need composite for a Nintendo 64, or you want scanlines that look like a PVM-20L5, buy the RetroTINK 5X-Pro. And if you are using a GBS-Control, you are getting a motion-adaptive deinterlacer that handles 480i passably and adds about 4 milliseconds, but it exhibits chroma artifacts on fast-moving red objects. There is no free lunch. There is only the lunch you choose to eat.

Put the two side by side and you can trace exactly why each one wins the games it wins. The OSSC's only output is an integer multiple of the input. Line2x, Line3x, Line4x, Line5x for 240p. The RetroTINK 5X-Pro outputs arbitrary resolutions up to 1440p. The OSSC preserves the original motion cadence exactly because it never drops or duplicates a frame. The RetroTINK 5X-Pro in framebuffer mode can introduce judder if the output refresh rate does not match the input, since it has to duplicate or drop frames to stay in sync. The OSSC cannot do 480i deinterlacing at all; it passes the interlaced signal through, which means your TV has to deinterlace, and your TV is the worst deinterlacer in the chain. The RetroTINK 5X-Pro deinterlaces 480i into progressive video with no flicker. That is the only way to play a PS2 at 480i on a modern panel without splitting headaches.

The sync tolerance tells the same story. The OSSC has a sync tolerance tuned for arcade and console RGB signals, but it will drop sync on a SNES white flash if the console's sync voltage sags too far. The RetroTINK 5X-Pro has a broader tolerance because it buffers the frame and can hold the last good frame during a sync glitch. The frame lock speed is the flip side. When a PS1 game switches from a 240p menu to 480i gameplay, the OSSC re-locks in about a second. The RetroTINK 5X-Pro takes two to three seconds. It has to flush the buffer and re-negotiate the output. Neither is broken. They just fail in exactly the place the other one works.

Integer scaling mathematics

The math of integer scaling is the line multiplier's entire reason to exist. A 240p source scaled 4x gives 960p. That means 120 lines of black on a 1080p display because 1080 minus 960 is 120, and the scaler cannot push pixels into those lines without breaking the grid. The RetroTINK in framebuffer mode can scale to exactly 1080p, filling every line, because it resamples the image across the full frame. The visual difference is a razor-sharp pixel grid versus a slightly softened grid that fills the whole screen. On a 65-inch OLED viewed from six feet, you will not see the difference. On a 27-inch monitor at two feet, you will.

What actually costs you money

The OSSC sits in the budget enthusiast bracket and takes RGB SCART input. The RetroTINK 5X-Pro occupies the mid-range and takes everything except VGA. The RetroTINK 4K sits at the flagship tier and outputs real 4K with HDR. The OSSC Pro adds a framebuffer, adaptive line multiplication, and HDMI input, which means it can process a modern console's 1080p signal. The GBS-Control is the entry-level option if you buy the board and the control module, and it will handle 480i deinterlacing competently if you can live with the chroma artifacts and the 4 to 8 milliseconds of lag. Check the manufacturer's store for current pricing on each device. If you only play 240p consoles, buy the OSSC. If you only have one slot in your life for this hobby, buy the RetroTINK 5X-Pro. If you already own a 4K OLED and you want the absolute best image for everything including modern consoles, the RetroTINK 4K is the only device on the market that does HDR CRT emulation. It takes the scanline masks and the phosphor simulation and wraps them in tone-mapped HDR. It is the only one that looks like a PVM-20L5 playing on an OLED. Nothing else comes close.

At this point you have the whole architecture in your head, so here is the route by priority. If your only goal is zero lag for 240p over RGB, buy the OSSC, plug in a SCART cable, set it to Line4x, and do not touch anything else. That is the entire setup. The output will be 960p and your TV will scale it the rest of the way, which adds the TV's lag, not the OSSC's. If you get a white flash during a SNES game and the screen drops to black for two seconds, that is the sync tolerance. Fix it by adjusting the sync threshold in the OSSC menu or by enabling the sync combiner on your SCART cable. If you want to play 480i PlayStation 2 games without flicker, or you need composite for a Nintendo 64, or you own anything that does not output RGB, buy the RetroTINK 5X-Pro. Set the output to 1080p, enable the motion-adaptive deinterlacer, and pick the FirebrandX profile that matches your display. That profile is free, it is downloadable, and it is the difference between a scanline filter that looks like a screen door and one that looks like a CRT.

The failure case is the same for both. It is 1am, you just bought a console you do not have a cable for, and the only scaler in the house is the one built into your TV. You are going to plug the console into the TV's composite input. It is going to look like garbage, and it is going to lag. The TV's scaler does not know what a 240p signal is and it treats it as 480i, which adds combing artifacts and doubles the lag. The fix is to not do that. The fix is to have bought a RetroTINK 2X-Pro. It is the entry-level version of the 5X that does line doubling only, has zero added lag, and accepts composite, S-Video, and component. It is the single best entry-level purchase you can make in this entire hobby. It will make that 1am console look exactly as good as it did on a CRT in 1995, which is to say, playable. The RetroTINK 2X-Pro's sub-1ms latency was measured by RetroRGB using a Time Sleuth in 2019, and it remains the reference budget buy.

There are exactly five questions everyone asks when the architecture finally clicks, and the answers are not what the marketing pages say.

1. Does the OSSC add any lag at all? In the sub-millisecond range, yes: less than 2 scanlines, which is under 64 microseconds. This is measured, not claimed. The Shmups Forum testing methodology treats this as effectively zero, and it is the number every other scaler is compared against.
2. How much lag does the RetroTINK 5X-Pro add in framebuffer mode? Between 0.25 and 0.5 frames according to RetroRGB's Time Sleuth testing. That is not the one-frame minimum since the scaler starts outputting before the input frame completes, but it is still 50 to 100 times more lag than the OSSC.
3. Can a line multiplier do non-integer scaling? No. It is physically incapable. A line multiplier has no frame memory, so it cannot resample. It repeats lines. The output resolution is always an integer multiple of the input line count. If your display is not an exact integer multiple, the display's own scaler does the last step.
4. Which one is better for 480i PlayStation 2 games? The RetroTINK 5X-Pro, unambiguously. The OSSC cannot deinterlace 480i; it passes it through, so your TV has to deinterlace, and your TV will add combing and lag. The 5X's motion-adaptive deinterlacer merges the fields correctly, producing a clean progressive image with no flicker and no combing.
5. Is the RetroTINK 4K worth the premium over the 5X-Pro? Only if you own a 4K display with HDR and you want the HDR CRT emulation. That feature takes the scanline masks from the FirebrandX profiles and tone-maps them into HDR, which produces a luminance range a standard CRT never had. It is a different image, not a better one. If you do not want that, the 5X-Pro is the better buy.

Where the measurements come from

Every latency number in this guide comes from one of two sources: the Shmups Forum's OSSC latency measurement thread, which has been measuring scaler lag since 2016, or RetroRGB's Time Sleuth tests, which have been the reference for HDMI scaler latency since 2019. Both publish their methodology. The Shmups thread uses a photodiode and a 240p test pattern. RetroRGB uses a Time Sleuth tester. If you want to verify the RetroTINK numbers yourself, you can buy a Time Sleuth, but the 2021 figure of 0.25-0.5 frames for the 5X-Pro is still the one quoted in the community and it has not been superseded. The RetroTINK 4K figure of 0.25-1 frame is from the 2024 launch testing and varies with output resolution and HDR processing load.

The GBS-Control is the only device mentioned here whose motion-adaptive deinterlacing exhibits chroma artifacts on fast-moving red objects. That is the exact reason it costs a fraction of the RetroTINK 5X-Pro.