Meme Text Generator

How auto-fit font sizing works: for each caption, the starting font size is the canvas width divided by 10 (rounded down). The tool then measures the caption's rendered width at that size using the canvas's built-in text measurement, and if it's wider than the canvas minus a 40px margin, it shrinks the font by one pixel and re-measures, repeating until the text fits or the font size reaches a hard floor of 14px, whichever happens first — so a very long caption will always stop shrinking at a still-readable minimum size rather than becoming illegibly tiny or causing the loop to run forever. How the outlined look is drawn: text is rendered using the font `bold {size}px Impact, "Arial Black", sans-serif` — Impact where it's installed, falling back to Arial Black or the browser's generic bold sans-serif otherwise. The outline width scales with font size (font size ÷ 12), and each caption is drawn twice at the identical centered position: once with `strokeText` in solid black for the outline, then immediately with `fillText` in solid white for the fill, which is what produces the familiar crisp white-on-black-outline look rather than a soft or uneven edge. Positioning: the top caption's baseline sits a fixed 10px below its own font size from the top of the canvas (so larger top text sits a little further down, keeping its top edge roughly aligned across different caption lengths), and the bottom caption's baseline sits a fixed 20px above the bottom edge of the canvas. Both are horizontally centered. Image handling: an uploaded image is scaled down to a maximum display width of 900px if it's larger (using the same `Math.min(1, 900 / width)` scaling approach used by this site's other canvas-based image tools), preserving its aspect ratio, and drawn once as the base layer before either caption is added. The blank-canvas option instead fills an 800×600 canvas with a flat light gray background. Either way, every redraw — triggered by typing in either caption box — repaints the entire canvas from scratch (background or image first, then top caption if non-empty, then bottom caption if non-empty), so the preview is always a single accurate flattened image with no leftover text from a previous edit. Edge cases: leaving either caption blank simply skips drawing that line entirely, so a meme can have only a top caption, only a bottom caption, both, or (before you've typed anything) neither. Downloading is disabled until an image has been chosen or Use Blank Canvas has been clicked, since there's nothing to export before then. The download itself is a PNG produced with the canvas's native `toBlob` export, the same approach used by this site's other image-generating tools.