Short answer: choose a unit and count, press Generate text. The seed is printed in the summary — keep it and the same text comes back exactly, which is what makes a before-and-after screenshot comparable.
Why a seed instead of randomness
Design review is a comparison: this spacing against that spacing, with the same words. A generator that reshuffles on every reload changes the variable you were trying to hold still. Seeding makes the filler part of the test fixture instead of part of the noise.
What it will not do
- No real sentences. The word list is filler, so nothing reads as a claim that might accidentally ship.
- No HTML. The output is plain text; wrap it in your own tags so the markup stays under your control.
- No Unicode trickery. Every word is ASCII, so a font test measures the font you chose, not a fallback.