Separate PNG output resolution from internal coordinate system

[adrhill]

First of all, thanks for the awesome package, it's a joy to work with! The documentation is stellar as well.

I've created a few SVG illustrations that I'm very happy with. However, after modifying the output format to PNG (by changing @svg to @png), I've noticed that the resolution of the PNG corresponds to Luxor's internal coordinate system.

I find it rather bothersome to work around this, since I would have to scale all "internal" dimensions of my shape primitives just to obtain the output resolution I desire.
As a workaround, I currently output SVGs and convert them to PNG using Inkscape (e.g. inkscape -w 2000 input.png -o output.png). Is there a way to avoid this manual step and set DPI / scale PNGs internally?

[cormullion]

A good question! I'm not too knowledgeable about this area - for me, PNGs are done in pixels/points, whereas SVGs are resolution-independent, to be scaled later if necessary. So I've not encountered your scenario.

These (https://www.cairographics.org/manual/cairo-Transformations.html#cairo-user-to-device) are the Cairo scaling functions; if you can work out how/when to call these to meet your requirements, I wonder if it could be integrated somehow... ?

[cormullion]

Thinking about a workflow: does this encapsulate the problem:

function master()
    Drawing(500, 500, :svg)
    origin()
    background("blue")
    ngon(O, 100, 4, 0, :stroke)
    finish()
    return svgstring()
end

function main(;PPI=72)
    svgdrawing = readsvg(master())
    w = svgdrawing.width
    h = svgdrawing.height 
    Drawing(w, h, "/tmp/output.svg")
    origin()
    placeimage(svgdrawing, centered=true)
    finish()

    w = svgdrawing.width * PPI / 72
    h = svgdrawing.height * PPI / 72
    Drawing(w, h, "/tmp/output.png")
    origin()
    @layer begin
        scale(PPI/72)
        placeimage(svgdrawing, centered=true)
    end 
    finish()
    preview()
end

main(PPI=144)

[adrhill]

These (https://www.cairographics.org/manual/cairo-Transformations.html#cairo-user-to-device) are the Cairo scaling functions; if you can work out how/when to call these to meet your requirements, I wonder if it could be integrated somehow... ?

The current transformation matrix, ctm, is a two-dimensional affine transformation that maps all coordinates and other drawing instruments from the user space into the surface's canonical coordinate system, also known as the device space.

I guess if ctm is the identity matrix by default, calling cairo_scale would do what I need?
A scaling between "user space" and "device space" might indeed be what I'm asking for here.

Thinking about a workflow: does this encapsulate the problem?

That would indeed fulfill my needs!