well i'm sure that it must be possible to develop a "3d camera" behavior plugin that could let you choose a side of a 3Dbox that would then draw to the screen what is in front of that face, but the thing is to do it... i would like to do it, but do not have enough knowledge to do so, i'm still studying the sdk's, but i'm far from being able to do something, i only found until now where the window is builded in the VRuntime.h, that itself uses RenderCommon.h, now i'm looking for the 3dbox, to see if i could find a way to refer to one of it's face, and then i'll have to figure out how i need ro build what i want
edit: just found out the 3Dbox i think it's Quad.h
edit2: i think that i wont be able to do it on my on, anyway i think i just found a program specially made to make 3d games, 3D rad, if somebody is intested here's the link: http://www.3drad.com/
there is an sdk tutorial. you should definitely start with that. it'd be extremely difficult to figure out the sdk without some guidance. here's the basic tutorial.
if you have trouble figuring it out, or getting it to compile, or whatever, the appropriate forum to ask questions about that end of things is construct engineering. I actually think it would be very simple to make a 3d camera, within the actual runtime of construct, because construct is already using a 3d camera, it's just stuck angled at the background, so it would only need the extra commands for yaw pitch roll (not sure which one we can already use). The problem is whether it would be practical or not.
I think it's probably possible to do it without altering the runtime in a plugin. I haven't worked directly with directx calls that much yet, but it seems like it should be pretty straightforward. I think it would work better as an object than a behavior though, as of now, all behaviors can be used on any object that is placed on the layout, like box, 3d box, sprite, line, etc, but not on mousekeyboard, array. so creating a behavior that only works with 3dbox would require rewriting 3d box to provide feedback to the behavior about what it's doing. that's not hard, but it's also not necessary
So back to the practical part. if it's simple, why not just add it to the runtime? because it may just cause more problems than it solves. your game will look like a bunch of flat 2d cutouts when you switch from looking at them straight on, all lined up like a diorama. Then you would need options for sprites to make them always face toward the camera, or be able to rotate them along all axis's. This is all possible through sprite distortion events, but it would be about as fun as trying to make the world rotate around you instead of using the 3d camera. Then there's the issue of having no 3d layout editor. All these features and extra options everywhere are going to confuse new users who will try them out and basically not understand why there seems to be a halfimplemented 3d aspect to construct, at which point the board will be constantly flooded by questions on how to make it work well. Part of the beauty of construct is that by sticking completely to 2d, you can have a simplified system of condition/action(cause/effect - stimulus/response) relationships that's just not possible in 3d where all the spatial information related to any event adds just one too many layers of complexity to make this system work as neatly.
but by all means, try out the sdk tutorial. a 3rd party plugin that does this would be harmless to the rest of the system, and to newcomers as well. also, if you know some c++, and don't mind fiddling around. creating such a plugin, if it's possible, should be pretty straightforward