johnn wrote:In order for us game developers to be able to leave 2.25 fully, we need the save runtime feature back.
I can't understand why you choose not to re-insert it with 2.33.
that seemed to be more of a mistake than a decision
johnn wrote:My feeling is that the blender developers priorities lack vision of the grander picture.
are there more than 2 people working on the game engine part of blender?
that was a rather large, and from my point of view, pointless rant. there has been a standalone runtime posted on elysiun.com for 2.33
and it isn't obscenely difficult (for a programmer with experience with whatever tool is at hand) to build it yourself.
johnn wrote:And while I'm at it: A time plan for improvements for the game engine and the progress of those improvments should be a sticky in this forum.
As things are now I don't know when or if the following things, for example, will be implemented:
* Other pysical objects than spheres
johnn wrote:* Built in LOD
so, uhh, you want blender to spend cpu cycles to calculate your lower detail mesh at runtime, making everything slower. Or do you want a way for the artist to specify lower detail meshes?
the latter is very much possible already
johnn wrote:* Built in real time shadowing
which kind? how? you want blobs projected on the geometry instead of parallel to it? you want dynamic rendered textures projected on the geometry? you want blurred dynamic rendered textures projected on the geometry? you want stencil shadows [as seen in doom III] with that too? how about dynamic rendered indexed shadows?
(the last two methods mentioned allow self shadowing. Stencil shadows need an interface to prohibit them from passing through walls, indexed shadows require convex regions of meshes be seperated. All methods require that some setting be set somewhere to determine which objects cast shadows, and which objects recive them)
this stuff could slow blender down significantly, and aren't really a priority. The priority of the moment appears to be getting a few culling methods to work.
someone posted on blender3d.org (I can't seem to find it right now) some plans for future changes in gameblender. Perhaps the second most notable graphics related one was a "fix for the alpha bug," or actually drawing faces with the alpha flag from farthest to nearest, so they do not appear to overlap incorrectly
johnn wrote:* Ragdoll-physics (or an object with physics connected to other object with physics stay connected after hitting other objects)
I have a few things to say about this
first, the current physics engine as far as I know has no capability for multiple bodies to be connected and simulated as linked in some way. ODE (which was put in blender until SOLID became avalible again) had this capability, but until recently there was no interface to it.
furthermore, there is no beginings of an interface to define the kinds of constraints required to make a reasonable looking ragdoll in blender.
finally, I believe it is possible to do this already, using python. It would be a bit of a nuisance as you can't interact with the physics and collision detection as would be possible if you were editing the code directly.