Convert PNG to BMP

This is the difference between converting a PNG and converting a photograph, and it is the one decision the conversion actually asks of you. A PNG carries an alpha channel with 256 levels; BMP has no usable transparency at all. So every see-through pixel has to become an opaque one, and something has to say which colour.

That choice is permanent. There is no way to recover the transparency from the BMP afterwards — the information is not hidden, it is replaced. Which makes it worth thirty seconds of thought rather than accepting a default and finding out later.

White for anything going into a document, a printed page or a light interface. Black for a dark display or a dark application. And for an embedded screen or a piece of equipment, the exact colour of the panel behind the artwork, which is usually specified somewhere in the documentation you were given.

Getting it wrong is visible rather than subtle: a logo filled with white and placed on a dark panel appears inside a white rectangle. That is not a fault in the conversion — it is the transparency that used to hide the rectangle no longer existing.

Anti-aliasing works by making the pixels along a curve or a diagonal partly transparent, which is how a rounded corner looks smooth instead of jagged. Those pixels are not fully transparent, so they are blended with the background colour rather than replaced by it.

Blended against the intended colour, the edge stays smooth and nobody notices. Blended against a different one, the artwork carries a faint halo of the wrong colour all the way around — the classic sign that a logo was flattened onto white and then placed on something else. It is another reason the fill colour has to match the destination rather than merely being neutral.

PNG is lossless and BMP is uncompressed, so every opaque pixel is carried across exactly. No detail is given up, no edges are softened, and the image in the BMP is the image that was in the PNG.

That makes this a rare conversion where the only loss is the one described above. It also means there is nothing to gain: the picture cannot be improved by the change, and any expectation in that direction is worth setting aside before starting.

PNG compresses losslessly, and it is very good at exactly the material people convert here — flat colour, hard edges, large areas of the same value. BMP stores every pixel as plain bytes with no compression at all.

So the gap is widest precisely for logos and interface graphics. A flat 1000 by 1000 icon might be 40 kilobytes as a PNG and around 3 megabytes as a BMP, and the difference has nothing to do with quality. The size depends only on the dimensions, which is why reducing the maximum width is the only lever that changes it.

Software and hardware that read pixels directly rather than implementing a decoder: embedded displays, industrial and laboratory applications, point-of-sale equipment, older Windows programs, and occasionally a course assignment that specifies the format for exactly that reason.

For every other purpose, keep the PNG. It is smaller, it keeps its transparency, it opens everywhere, and it is the file you would want back if the destination ever changes. Convert at the boundary and treat the BMP as something you hand to a machine rather than something you store.

Drop a whole folder: each file converts in turn on your own machine, keeps its name with the extension swapped, and the batch comes back as one ZIP. There is no upload, no queue and no daily allowance.

The constraint worth planning for is size. Uncompressed outputs add up quickly, and a folder of interface graphics can produce hundreds of megabytes that all have to fit in the browser’s memory while the archive is assembled. For a large set, converting in smaller groups is the practical approach.