OKAY so if I copy-paste out of my image editor, like this, I get a PNG.
(better than a BMP, at least, and it makes sense for screenshots)
but if I copy a jpeg file in explorer, it pastes as a jpeg.
so here's my idea: I write a little script that yanks the PNG out of the clipboard, saves it as a JPEG to a temp folder, then copies that jpeg file. So instead of directly posting out of my image editor, I do copy, click the script icon, then paste.
that'd be only a minor change to my workflow, it'd fix the PNG problem, and it wouldn't be too hard to write, I don't think.
basically I'm trying to avoid the problem of posting giant PNGs when they'd be better served as a small JPEG. Twitter doesn't care what you post, it will JPEG them if it feels like it, and I have been unintentionally depending on that functionality
@foone you might be able to create a userscript for the fedi frontend that you use which grabs the PNG, dumps it into an html canvas, and then exports it as a jpeg, then replaces the file in the file input element with the jpeg. Like a greasemonkey script.
I did this pattern in my own app once, looking back at the code https://git.sequentialread.com/forest/graffiti-app/src/branch/main/static/app.js#1454
I think I ended up nixing the client side conversion to jpeg because it didn't work properly in every browser. Image rotation issues on iOS probably related to apples proprietary photo format 
@foone https://developer.mozilla.org/en-US/docs/Web/API/HTMLCanvasElement/toDataURL. I think this might make a DOM blob instead of a huge string? Or maybe there's a different method that does that, I don't remember and I can't find the old code from my phone rn