>Really, the point at which this billinear filter is being inserted is
>the wrong one (bilinearly filtering the tiles introduces issues that
>bilinearly filtering an entire game-window-sized image would not.
Yes. For example the energy bar drawing routine is doing pixel
replacements on a scaled image based on a preprogrammed color
range. This is causing an artifact. Are there other sections of
Wesnoth code that are similarly problematic?
Pro (for game-window approach):
- avoids whole classes of potential artifacts/issues
- easier to migrate to graphics hardware someday
- avoids Wesnoth-specific interpolation algorithm
- skeleton bowstring will look a little better :)
Con:
- not well suited to Wesnoth's current rendering architecture
- can't precompute scaling like Wesnoth does now
- more scaling operations overall in a given game
Except for the energy/xp bar artifacts, the tiling approach is
producing pretty good results now, and is not a particularly invasive
code change. If the known artifacts are surmountable AND there aren't
other problems lurking, I think the tiling approach is right for the
immediate term.
However, I am only going to bother tackling the energy bar artifact if
the resulting code is eventually accepted by Wesnoth mainline. Am I
wasting my time?
Cheers,
Jeff
