Actually 16-bit is already more than enough for almost any kind of picture you'd like to create - more colors can be created in 16-bit than what you're eye can tell from one-another.
Goind to 32 bit has only 1 major thing as Sandwich said - ituses 8 bits for transparency data.
24bit on the other hand uses 8-8 for R,G,B so it does not support transparency.
As for the whole resolution issue - higher resolution is always better since all the lines are straighter. On the other hand it always takes more power.
The later option -mappings and ect. - however would take more power IMHO.
Higher maps are good at close distances - however this one's most important for cap-ships, since the problem with them is that the textures are streched onto huge polys, and end up wiht low DPI (dot/inch) - even for 640*480 they would need better maps.
Using more textures was already mentioned.
As for multiple resolutions:
How about using vector graphics - I mean if all the HUD was done in vector graphics then it could run in any resolution, even the size of all the gagues could be fine-tuned for everyone.
However that would be a load of work and mostly new hardcoded support of 2D vector graphics.
On the other hand an outside application could be used to create the apropiate .pcx-s from vector-g. each time the resolution is changed in the launcher.
Finally high-resolution HUDs can always be scaled down in a similar method - or even in-game as it was already suggested.
As for scaling up/down I can't believe the lack of attention to certains tricks:
Any video player program does it, they simply show a pixel more times to fill up the screen. Those are the grindy - ugly looking scaleups - however there are a whole buch of blur tricks to refine that image.
DivX for instance uses such things, so does ordinary MPEG-1.
Even a 3*3 Average matrix can have good results.
I think we need a codec or video format guru who has knowledge of actually how those filters work.
Than it's only putting these filters into the code and we can even scale up .ani-s.
Actually a whole range of those filters are supported by DirectX so we simply have to acces them.
If I sound pompous, than I'm sorry.
I'm no DX programmer, I don't know the exact hierarchy DirectX functions and layers, I was merely commenting on things used every day in video manipulation.