[NiLuJe]

The only thing that works with a plain factor is the fixed cell font rendering.

For images, scaling indeed works as "get me a wxh image out of whatever I'm feeding you". (Or a best fit for w; a best fit for h; a best fit for screen; there's a few magic values/combinations to achieve that, but the default is to keep the AR).

The assumption being that you know the screen size, you know the specific area you want to fill, so you don't really care about the source's size.

----

That said, it's entirely possible to request the image's dimensions without fully decoding it (for... most image formats, anyway), but that's not something that's currently exposed, even to API users.

I might revisit that later when I try my hands at a NanoSVG frontend, because knowing the optimal raster size for those *does* make sense, but it's highly unlikely to end up being exposed to the CLI tool, only to API users.