Caption Colour, Contrast and Legibility: The Complete Guide

Caption legibility is usually treated as a styling preference. It is closer to an engineering constraint. Text over video has to stay readable while the thing behind it changes sixty times a second, on a screen held at arm's length, frequently outdoors, often at reduced brightness to save battery. Most caption styling decisions are made in exactly the opposite conditions: a large bright monitor, indoors, on a single paused frame.

Why contrast ratios are necessary but not sufficient

The WCAG AA requirement of 4.5:1 for normal text is the right starting point and a widely accepted floor. But it was written for text on a known, static background. Video violates both assumptions: the background is unknown at design time and it moves.

The practical consequence is that a caption which measures a comfortable 6:1 against your average frame can drop below 2:1 for the twenty frames where someone walks past a window. Those twenty frames are enough to break reading, because the reader loses their place and has to recover.

The three reliable techniques

Stroke (outline)

A dark stroke around light text is the workhorse of short-form captions. It is cheap, it never covers footage, and it works because the stroke guarantees a local contrast boundary regardless of what is behind it. Two to four pixels at 1080p is the usable range; below two it disappears, above four it starts to swallow letterforms — especially in Devanagari, where the shirorekha merges.

Plate (background box)

A semi-opaque rounded rectangle behind the text. This is the most reliable option and the one broadcast has used for decades. The cost is that it covers footage and reads as more formal. Use 60 to 80 percent opacity — fully opaque looks bolted on, and below 50 percent stops guaranteeing anything.

Shadow

A tight, offset drop shadow. Softer visually than a stroke, and adequate on footage that is consistently mid-toned. It is the weakest of the three on bright backgrounds because a blurred shadow does not create a hard boundary. Keep the blur small; a large soft shadow adds mud, not contrast.

Colour choices that survive real footage

Brand colour captions are tempting and usually a mistake. Saturated colours have lower luminance than white, so they lose contrast faster, and mid-tone brand colours are the worst offenders. Keep brand colour for the highlight word, not the body.

• White body text with a dark stroke: the most robust default, and it suits the bright, warm footage typical of Indian creator content.
• Pale yellow body text: slightly better on very bright backgrounds, and a long-standing broadcast convention.
• Brand colour on the active or highlighted word only, with the rest of the line in white.
• Never mid-tone grey, mid-blue or mid-green for body text — they fail against both dark and bright footage.
• Avoid pure black text unless you are always on a light plate.

A ten-second test that catches nearly everything

1. 1Scrub to the single brightest frame in your video — usually a window, a sky, a white wall or a reflective surface.
2. 2Pause there and look at the caption. If it is uncomfortable, it is broken; that frame will appear.
3. 3Now scrub to your darkest frame and check the stroke is not swallowing the letters.
4. 4Drop your phone brightness to about 40 percent and look again. This is how a large share of your audience will actually see it.
5. 5Hold the phone at normal viewing distance, roughly 40 to 50 centimetres, rather than close to your face.
6. 6Finally, watch the video once at full speed without reading the captions deliberately. If your eye trips anywhere, fix that moment.

Script-specific adjustments

Latin type is not the hardest case. Devanagari, Tamil, Telugu and Bengali all carry marks above or below the baseline, and their strokes are typically thinner at the same optical size. Same stroke settings, different result.

Accessibility is not a separate goal

Everything above is usually framed as accessibility work, and it is. It is also straightforwardly commercial: roughly the same adjustments that make captions usable for a viewer with low vision make them usable for a viewer on a bus in daylight with the sound off. That is not a minority case in India — it is the default viewing condition.