Right now the RGSS Bitmap class is mostly implemented using System.Drawing.Bitmap, for simplicity. Once changed, the related texture is disposed and recreated. Notably, some special classes are more optimized (i.e. the tileset).
While this was easier to set up, it's not the best long-term solution as it is considerably less performant than a true OpenGL solution would be (or at least only re-uploading subtextures rather than the entire image). There is a noticeable FPS snag when using heavy Bitmap features. The vanilla engine was non-performant in this category as well; but the hit is more visible in this current implementation (and we have an opportunity to exceed the performance capabilities of the original engine here).
I have a few vague ideas for how the different functions could be implemented, but this will take a significant amount of work.
I won't mark this issue for the 1.0 milestone as it's not critical to the functionality, but it's an important performance feature.
Right now the RGSS Bitmap class is mostly implemented using System.Drawing.Bitmap, for simplicity. Once changed, the related texture is disposed and recreated. Notably, some special classes are more optimized (i.e. the tileset).
While this was easier to set up, it's not the best long-term solution as it is considerably less performant than a true OpenGL solution would be (or at least only re-uploading subtextures rather than the entire image). There is a noticeable FPS snag when using heavy Bitmap features. The vanilla engine was non-performant in this category as well; but the hit is more visible in this current implementation (and we have an opportunity to exceed the performance capabilities of the original engine here).
I have a few vague ideas for how the different functions could be implemented, but this will take a significant amount of work.
I won't mark this issue for the 1.0 milestone as it's not critical to the functionality, but it's an important performance feature.