For instance, let's say I have a 1024x1024 image, and a 32x32 tile size.
I've heard Lummox say in the past that BYOND sort of buffers the view sending distance to the largest possible offset that needs to be sent for every client in the world once a large icon is used. Is this still the case? If so, would this not make a single obscenely large object render the entire project's view sending algorithm completely untenable?
Or does it manage it by something of a quadtree by stuffing the minimum visual bounds of an object into a subdivison?
Also, a second question: Are there plans to fix the problem that objects scaled and rotated by matrices still use the AABB of the untransformed image, rather than the AABB of the transformed image for view sending?
What I mean is:
This is the unmodified AABB for the image that determines its physical footprint in game.
x1 = x*TILE_WIDTH + bound_x + step_x + pixel_x
y1 = y*TILE_HEIGHT + bound_y + step_y + pixel_y
x2 = x1+icon.width
y2 = y1+icon.height
Whereas, when you scale an image up, if you move past the original AABB's coordinates, the scaled image disappears as though it's icon ends there.
Is my assumption correct that having a single really large icon would increase the artificial view-sending bounds for every player in the world, thus increasing the network bandwidth on a per-player basis, and secondly, is there a plan in place to fix the objects' footprint not being based on transformed coordinates?