Add Text to an Image

Legibility beats style on a photograph

Text over an image is only readable if it contrasts with whatever is behind it, and no font choice fixes a poor contrast. Bright text on a bright area disappears even in a heavy weight, and the instinct to reach for a heavier font makes it worse rather than better. The reliable approaches are a text colour that contrasts with the background, a translucent panel behind the text, or placing the text over the darkest or lightest part of the frame and colouring it to suit. Anything that works reliably is more useful than a subtle treatment that nobody can read.

NoteA heavier font does not fix low contrast. Use a colour or a panel that separates the text.

Font size has to be judged at final size

A caption that looks reasonable while you are editing a large preview can become unreadable at the size the image is actually viewed. This is the most common problem with added text, and it happens because the preview is magnified. The test is to look at the result at the size it will be displayed, which for a social post is thumbnail sized on a phone and for a print is a few centimetres tall. If it is not readable at that size, no amount of adjustment at full size will make it work, and the text needs to be shorter or larger rather than restyled.

• Judge legibility at the size the image will be viewed.
• A preview is magnified. A full size preview is not a fair test.
• On a phone, a social post is viewed small. Assume that.
• Shorten the text before you shrink the font further.

NoteIf it is unreadable at final size, make it shorter or larger. Restyling will not save it.

Safe margins and where not to place text

A full bleed image is usually cropped differently depending on where it is displayed, and text near an edge is the first thing to be cut. A social platform cropping to a square will remove the top and bottom of a portrait image, and a story format will crop the sides of a landscape one. Placing text a reasonable margin in from every edge protects it across those variations. The corners are the least safe area of all, because they are cropped by more formats than the middle. If the image is the priority and the text is secondary, keeping the text out of the safe area is more important than making it large.

• Full bleed images are cropped differently by different platforms.
• Text near an edge is the first thing lost in a crop.
• Corners are cropped by more formats than the middle. Avoid them.
• Keep a margin from every edge that could be cropped.
• On a social post, assume the bottom is the first area to go.

Keeping the text editable and the file safe

Rasterised text becomes part of the image pixels and cannot be changed afterwards, which is the main limitation of doing this in a browser rather than in a layout tool. That is fine for a finished post, and inconvenient for something you will revise. Two habits help: keep the unedited original so the text can be added again with different wording, and if the text needs to be localised or reworded regularly, compose it in something that keeps it as text. The added text also means the output is re-encoded, so the quality depends on the format and setting used to save, and an original JPEG gains a second lossy pass.

• Rasterised text cannot be edited after saving.
• Keep the original so the text can be added again with different wording.
• For text that changes often, compose it in something that keeps it as text.
• Saving re-encodes the image. A JPG gains a second lossy pass.
• Export as PNG for text that must stay sharp, at the cost of file size.