Why do the RGB and composite versions look so different?

They are two different physical outputs. The RGB monitor shows the CGA palette colors as sharp pixels. The composite TV receives a single analog NTSC signal, where the horizontal dot patterns blend into artifact colors that don't exist in the RGBI palette. Same data, different display hardware.

Is this composite simulation accurate?

Yes. It builds the real composite signal from the pixel dot-patterns and demodulates it like an NTSC TV, using a faithful port of reenigne's chroma-multiplexer algorithm (the same model DOSBox uses). It does not sample the RGBI colors directly, which is the common error in other tools.

CGA composite simulator

The exact same CGA image, shown two ways: on an RGB monitor (sharp, the 16 RGBI colors) and over NTSC composite into a TV (where dot patterns blend into artifact colors that aren't in the palette at all). This is the difference that defined how CGA games actually looked — decoded correctly, not faked.

Click to choose an image or drag & drop · samples:

CGA mode dithering

CGA palette

background

composite monitor knobs

hue 0 saturation 100 sharpness 0

RGB monitor (the 16 RGBI colors)

NTSC composite (decoded artifact color)

Same pixels, different hardware. Nothing changes between the two panes except the display path. The left is what an RGB monitor's digital input shows; the right is what a TV makes of the analog composite signal — where four pixels per color-carrier cycle become hue, saturation, and brightness. The artifact colors on the right genuinely do not exist in the 16-color RGBI palette.

Why this is the interesting part of CGA

For years, players saw CGA games over whatever display they had — and on a composite TV (very common in the early 1980s US), the picture was completely different from the crisp RGB version developers might have authored on. Some games were designed for composite specifically, exploiting the artifact colors to get a richer-looking palette than four flat colors. The cyan/magenta/white "palette 1" look people remember is the RGB story; the composite story is browns, oranges, and blended hues the hardware was never "supposed" to show.

The trick to simulating it correctly is to model the signal, not the palette. CGA's pixel clock (14.31818 MHz) is exactly four times the NTSC color subcarrier (3.579545 MHz), so a horizontal dot pattern is a chroma signal of some phase and amplitude — which a TV decodes as color. Sampling the RGBI values directly (the common shortcut) can never reproduce those colors. This simulator ports reenigne's chroma-multiplexer decoder — the same model DOSBox uses — so the right-hand pane matches real hardware. The full method and sources are on the accuracy page, and the theory is in how CGA color works.

To convert and download images through these modes, use the image converter; to explore the RGB palettes themselves, see the CGA mode simulator.

Frequently asked

Does the dithering affect the composite result?: Yes, strongly — dithering changes the dot patterns, and on composite the dot patterns are the color. Try switching dithering methods and watch the right pane shift while the left only changes texture.
What do the knobs do?: They mirror a composite monitor's controls: hue rotates the artifact color wheel, saturation scales color intensity, sharpness trades blur for crispness. Old vs New CGA picks the card revision, which shifts the calibration.
Is anything uploaded?: No. Everything runs in your browser on a canvas; the image never leaves your device.