Type that survives a phone and a compressor

Three things working against you

Text on video faces problems that text on a page does not.

The background changes every frame. White text that is perfectly legible over a dark jacket disappears the moment the subject moves in front of a window.

The compressor hates thin strokes. A hairline serif at 1080p is a few pixels wide. The encoder allocates bits by how much changes between frames, and fine detail over moving video is exactly what it discards first. Thin type comes back broken and fringed.

It is being read on a phone. A 1080p frame on a phone held at arm's length subtends roughly the same visual angle as a paperback page, and the reader has a few seconds rather than as long as they like.

Sizing, in numbers

Work in percentages of frame height, because that is what transfers between resolutions.

• Absolute minimum for body-sized text: about 2.5% of frame height. At 1080p that is 27 px; at 2160p, 54 px.
• Comfortable for captions and lower thirds: 3.5 to 4.5% of frame height. 38–49 px at 1080p.
• Vertical short-form captions: larger again, 5 to 7% of frame height, because they are the primary channel.
• A title meant to be read at a glance: 6 to 10%.

If you are designing at 4K and delivering at 1080p, check the numbers at the delivery size. Type that looks generous on a 4K canvas is half the size you thought on the file people watch.

Weight, not size, is usually the fix

A common mistake is making text bigger when the problem is that it is too light. Never use a hairline, thin or light weight on video. Medium, semibold or bold, always.

The reason is the compressor. A bold stroke is 6 or 8 pixels wide and survives quantisation; a 2-pixel stroke does not. This is why type that looks elegant in your editor looks broken after upload.

Similarly: avoid high-contrast display serifs, where the thin strokes are very thin. A slab serif or a sturdy sans survives.

Contrast over a moving background

Four mechanisms, in rough order of how often they are the right answer:

1. A semi-opaque box behind the text. Black at 60–80% opacity. Ugly in a design sense, completely reliable, and used by every broadcaster on earth for exactly that reason.
2. A drop shadow, tight and dark: offset 2–4 px, blur 4–8 px, 60–80% opacity. Nearly invisible as an effect, and it keeps white text readable over most backgrounds.
3. An outline or stroke, 2–4 px in a contrasting colour. Standard for short-form captions. It survives any background and it looks like short-form captions, which may or may not be what you want.
4. A gradient scrim — a soft dark gradient across the bottom third, under the text. Less heavy than a box and works when the text position is fixed.

What does not work: plain white text with nothing behind it, which will vanish over a bright sky exactly once in every video and you will not notice because you know what it says.

Position and the safe area

From the earlier lesson: keep text within 90% of the frame — 5% in from every edge.

For vertical short-form specifically:

• The bottom 15–20% is covered by the caption, username and interface buttons.
• The top 10% carries the status bar and platform chrome.
• Put anything that must be read in the middle third.

Every editor draws these guides. Turn them on, and check a real screenshot from the actual platform if you can, because the safe zones differ between them and change.

Timing

The reliable test: you should be able to read the text aloud twice, at a normal pace, while it is on screen.

You will always underestimate this, because you wrote it and you know what it says. A three-word name super needs about two and a half seconds; a sentence needs four or five.

For captions specifically, the numbers from the earlier lesson hold: 15–17 characters per second is comfortable, above 20 people stop reading, and a minimum of about one second on screen even for a single word.

Consistency

Pick two type sizes and one family for a whole piece, and stick to them. The most common amateur tell is not bad typography but inconsistent typography — a different size, position and style for every piece of text because each was made separately.

Make one title template and duplicate it. Every editor supports saving a title as a preset or a template; in Resolve you save it to the Media Pool or the Edit page's Titles, in Kdenlive as a title template file.

Today

Export thirty seconds with a line of text in a thin weight and the same line in bold, upload it somewhere that re-encodes, and look at both on a phone. That comparison settles the weight question permanently.